Physics potential of the International Axion Observatory (IAXO)
E. Armengaud, D. Attie, S. Basso, P. Brun, N. Bykovskiy, J. M. Carmona, J. F. Castel, S. Cebrián, M. Cicoli, M. Civitani, C. Cogollos, J. P. Conlon, D. Costa, T. Dafni, R. Daido, A.V. Derbin, M. A. Descalle, K. Desch, I.S. Dratchnev, B. Döbrich, A. Dudarev, E. Ferrer-Ribas, I. Fleck, J. Galán, G. Galanti, L. Garrido, D. Gascon, L. Gastaldo, C. Germani, G. Ghisellini, M. Giannotti, I. Giomataris, S. Gninenko, N. Golubev, R. Graciani, I. G. Irastorza, K. Jakovcic, J. Kaminski, M. Krcmar, C. Krieger, B. Lakić, T. Lasserre, P. Laurent, O. Limousin, A. Lindner, I. Lomskaya, B. Lubsandorzhiev, G. Luzón, M. C. D. Marsh, C. Mergalejo, F. Mescia, M. Meyer, J. Miralda-Escudé, H. Mirallas, V. N. Muratova, X. F. Navick, C. Nones, A. Notari, A. Nozik, A. Ortiz de Solórzano, V. Pantuev, T. Papaevangelou, G. Pareschi, K. Perez, E. Picatoste, M. J. Pivovaroff, J. Redondo, A. Ringwald, M. Roncadelli, E. Ruiz-Chóliz, J. Ruz, K. Saikawa, J. Salvadó, M. P. Samperiz, T. Schiffer, S. Schmidt, U. Schneekloth, M. Schott, H. Silva, G. Tagliaferri, F. Takahashi, F. Tavecchio, H. ten Kate, I. Tkachev, S. Troitsky, E. Unzhakov, P. Vedrine, J. K. Vogel, C. Weinsheimer, A. Weltman, W. Yin
PPrepared for submission to JCAP
Physics potential of the InternationalAxion Observatory (IAXO)
E. Armengaud, a D. Attie, a S. Basso, b P. Brun, a N. Bykovskiy, c J. M. Carmona, d J. F. Castel, d S. Cebri´an, d M. Cicoli, e,f
M. Civitani, b C. Cogollos, g J. P. Conlon, h D. Costa, g T. Dafni, d R. Daido, i A.V. Derbin, j M. A. Descalle, k K. Desch, l I.S. Dratchnev, j B. D¨obrich, c A. Dudarev, c E. Ferrer-Ribas, a I. Fleck, m J. Gal´an, d G. Galanti, b L. Garrido, g D. Gascon, g L. Gastaldo, n C. Germani, g G. Ghisellini, b M. Giannotti, o I. Giomataris, a S. Gninenko, p N. Golubev, p R. Graciani, g I. G. Irastorza, d, K. Jakovˇci´c, q J. Kaminski, l M. Krˇcmar, q C. Krieger, l B. Laki´c, q T. Lasserre, a P. Laurent, a O. Limousin, a A. Lindner, r I. Lomskaya, j B. Lubsandorzhiev, p G. Luz´on, d M. C. D. Marsh, s C. Mergalejo, d F. Mescia, g M. Meyer, t J. Miralda-Escud´e, g,u
H. Mirallas, d V. N. Muratova, j X. F. Navick, a C. Nones, a A. Notari, g A. Nozik, p,v
A. Ortiz de Sol´orzano, d V. Pantuev, p T. Papaevangelou, a G. Pareschi, b K. Perez, w E. Picatoste, g M. J. Pivovaroff, k J. Redondo, d,x
A. Ringwald, r M. Roncadelli, y E. Ruiz-Ch´oliz, d J. Ruz, k K. Saikawa, x J. Salvad´o, g M. P. Samperiz, d T. Schiffer, l S. Schmidt, l U. Schneekloth, r M. Schott, z H. Silva, c G. Tagliaferri, b F. Takahashi, i,aa
F. Tavecchio, b H. ten Kate, c I. Tkachev, p S. Troitsky, p E. Unzhakov, j P. Vedrine, a J. K. Vogel, k C. Weinsheimer, z A. Weltman, ab W. Yin. ac,ad a IRFU, CEA, Universit´e Paris-Saclay, F-91191 Gif-sur-Yvette, France b INAF - Osservatorio astronomico di Brera, Via E. Bianchi 46, Merate (LC), I-23807, Italy c European Laboratory for Particle Physics (CERN), Geneva, Switzerland d Grupo de F´ısica Nuclear y Astropart´ıculas, Departamento de F´ısica Te´orica,Universidad de Zaragoza C/ P. Cerbuna 12 50009, Zaragoza, Spain e Dipartimento di Fisica e Astronomia, Universit`a di Bologna,via Irnerio 46, 40126 Bologna, Italy f INFN, Sezione di Bologna, viale Berti Pichat 6/2, 40127 Bologna, Italy Corresponding author a r X i v : . [ h e p - ph ] J un Institut de Ci`encies del Cosmos, Universitat de Barcelona, Mart´ı i Franqu`es 1, 08028Barcelona, Spain h Rudolf Peierls Centre for Theoretical Physics, Clarendon Laboratory, Parks Road, OxfordOX1 3PU, UK i Department of Physics, Tohoku University, Sendai, Japan j NRC KI Petersburg Nuclear Physics Institute, 188300, Gatchina, Russia k Lawrence Livermore National Laboratory, Livermore, CA, U.S.A. l Physikalisches Institut, Universit¨at Bonn, Nussallee 12, 53115 Bonn, Germany m Department Physik, Universit¨at Siegen, Siegen, Germany n Kirchhoff Institute for Physics, Heidelberg University, INF 227 69120 Heidelberg Germany o Physical Sciences, Barry University, 11300 NE 2nd Ave., Miami Shores, FL 33161, USA. p Institute for Nuclear Research (INR), Russian Academy of Sciences, Moscow, Russia q Rudjer Boˇskovi´c Institute, Zagreb, Croatia r Deutsches Elektronen-Synchrotron (DESY),Notkestr. 85, 22607 Hamburg, Germany s Oskar Klein Centre for Cosmoparticle Physics, Stockholm University, Albanova, StockholmSE-10691, Sweden t W. W. Hansen Experimental Physics Laboratory, Kavli Institute for Particle Astrophysicsand Cosmology, Department of Physics and SLAC National Accelerator Laboratory, Stan-ford University, Stanford, California 94305, USA u Instituci´o Catalana de Recerca i Estudis Avan¸cats, Passeig Llu´ıs Companys 23, 08010-Barcelona, Spain v Moscow Institute of Physics and Technology, 9 Institutskiy per., Dolgoprudny, MoscowRegion, 141700, Russian Federation w Department of Physics, Massachusetts Institute of Technology, Cambridge, MA 02139, USA x Max-Planck-Institut f¨ur Physik (Werner-Heisenberg-Institut), F¨ohringer Ring 6, D-80805M¨unchen, Germany y INFN, Sezione di Pavia, via A. Bassi 6, I-27100 Pavia, Italy z Institute of Physics, University of Mainz, Mainz, Germany aa Kavli IPMU (WPI), UTIAS, The University of Tokyo, Kashiwa, Japan ab The Cosmology and Gravity Group, Department of Mathematics and Applied Mathemat-ics,University of Cape Town, Private Bag, Rondebosch, 7700, South Africa ac IHEP, Chinese Academy of Sciences, Beijing, China ad Department of Physics, KAIST, Daejeon, KoreaE-mail: [email protected]
Abstract.
We review the physics potential of a next generation search for solar axions: theInternational Axion Observatory (IAXO). Endowed with a sensitivity to discover axion-likeparticles (ALPs) with a coupling to photons as small as g aγ ∼ − GeV − , or to electrons g ae ∼ − , IAXO has the potential to find the QCD axion in the 1 meV ∼ ontents N DW = 1 223.3.2 Models with N DW > – 1 – Introduction
Axions and axion-like particles (ALPs), as well as other more generic categories of particles(weakly interacting sub-eV particles, WISPs) at the low-mass frontier [1–3] are acquiring astrong interest as a portal for new physics, candidates to the dark Universe, or as solutionsof poorly understood astrophysical phenomena. The detection of these particles in terrestrialexperiments is currently pursued by a number of experimental techniques, see [4, 5] for recentreviews. In this experimental landscape, the International Axion Observatory (IAXO) [6, 7]stands out as one of the most ambitious projects under consideration. It is our purpose hereto review the theoretical, cosmological and astrophysical motivation to carry out the primarygoal of IAXO, i.e. the search for solar axions with sensitivity much beyond previous similarsearches and well into unexplored parameter space. We will focus on more recent develop-ments affecting regions of parameter space at reach of IAXO, as well as those highlightingits novelty and complementarity within the larger set of axion experimental efforts.The QCD axion is a hypothetical 0 − particle predicted in the Peccei-Quinn mechanismto solve the strong CP problem [8] of the standard model (SM) of particle physics. In the pureSM, the parity (P) and time-reversal (T) violation effects observed so far can be attributedto the phase of the CKM matrix, which is relatively large δ ∼ o and has its origin inthe Yukawa couplings of the Higgs to fermionic fields. The problem is that theory predictsthe existence of another P,T violating phase, ¯ θ , which appears in the Lagrangian densitymultiplying the topological-charge density of QCD, L ¯ θ = − α s π G aµν (cid:101) G aµν ¯ θ . (1.1)This term produces CP violating effects like electric-dipole moments (EDMs) for hadrons,which have never been observed. The strongest upper limit on strong CP violation comes fromthe neutron EDM, | d n | < . × − e fm [9]. Given the calculation d n = (2 . ± . θ × − e fm [10], one finds an extremely strong upper bound, | ¯ θ | < . × − . (1.2)Indeed, ¯ θ arises in the SM as the sum of two contributions: the θ -angle defining a gauge-invariant QCD vacuum and a common phase of the quark-mass matrix. The latter has anorigin similar to the CKM phase and the former has no clear a-priori relation with them soit is extremely suspicious that these two will cancel so precisely as (1.2).The solution proposed by Peccei and Quinn [11, 12] to alleviate the “strong CP issue” isbased on the observation that the QCD vacuum energy, V QCD (¯ θ ), has an absolute minimum at ¯ θ = 0. They proposed the existence of an extra global U(1) symmetry, spontaneouslybroken and colour anomalous. Due to the colour anomaly, the concomitant Goldstone-bosonfield, dubbed “axion” [13] by Wilczek and denoted here by A , develops in the effective low-energy theory an anomalous coupling to gluons, L Ag = − α s π G aµν (cid:101) G aµν Af A , (1.3)where f A is the so-called axion decay constant, a new energy scale related to the scale ofPQ symmetry breaking. Effectively, ¯ θ becomes replaced by ¯ θ + (cid:104) A (cid:105) /f A , where (cid:104) A (cid:105) is the In a strict sense this is true if ¯ θ is the only source of CP violation in the SM. One expects a tiny shift ofthe minimum due to the non-zero CKM and from any other source of CP violation that can be communicatedradiatively to the gluonic sector. – 2 –acuum-expectation-value (VEV) of the axion. Since the axion has no other potential energyterms in the Lagrangian (as long as PQ is a symmetry at the classical level) it will adjustitself to minimise V QCD (¯ θ + A/f A ) by taking the VEV (cid:104) A (cid:105) = − θf A , which cancels all the CPviolating effects of ¯ θ !. Note that the axion field effectively becomes a dynamical version (itis a space-time dependent field) of the ¯ θ angle, which is a mere constant.The QCD potential gives the axion field a mass, m A = √ χf A (cid:39) . GeV f A , (1.4)(where χ = ∂ θ V QCD (cid:12)(cid:12) ¯ θ =0 is the topological susceptibility of QCD) and induces mixing of theQCD axion field with the η (cid:48) and the rest of the 0 − mesons. By virtue of this mixing, theQCD axion develops model-independently couplings to hadrons and, most importantly, acoupling to two photons, L Aγ = − αC Aγ πf A F µν (cid:101) F µν A = − g Aγ F µν (cid:101) F µν A = g Aγ (cid:126)E · (cid:126)BA. (1.5)Here F µν is the electromagnetic field-strength, (cid:101) F µν its dual and (cid:126)E, (cid:126)B are the electric andmagnetic fields, respectively. The coupling constant g Aγ has units of an inverse energy scale,which is 1 /f A except for an electromagnetic loop factor required for photon emission, α/ π ,and the mixing coefficient C Aγ = − .
92 [14]. The parameter C Aγ can receive additionalmodel-dependent additive contributions. These are usually of the order of 1 in simple mod-els [15–18] but can be much larger in some engineered cases [19, 20]. Note that both themass (1.4) and couplings, like (1.5), are inversely proportional to f A . Ratios of coupling/massare independent of f A and only depend on QCD physics and a few hopefully O(1) model-dependent parameters so the properties of the QCD axion are reasonably constrained.The PQ solution to the strong CP problem is a minimal one, and is appealing for tworeasons. The first is that the predicted axion particle can be experimentally confirmed orrejected (as opposed to some alternatives of the strong CP problem, e.g. [10]). So far QCDaxions with f a (cid:38) GeV have already been robustly excluded by solar axion searches, stellarevolution, cosmology and laboratory experiments. The second is that the QCD axion with alarge decay constant f a (cid:29) GeV is an excellent cold dark matter (CDM) candidate. Thecombination of the strong CP problem and the resounding evidence of CDM in the Universemotivate very strongly the search for a low mass QCD axion.Indeed, the popularity of the QCD axion has made generations of theorists to employthe word “axion” for other, equally-hypothetical, particles that are not related to QCD, thestrong CP problem or the PQ mechanism simply because they have certain similarities tothe QCD axion. The similarities often exploited are: being a low-mass 0 − particle, a pseudo-Goldstone boson of an “axial” symmetry, featuring anomalous couplings to topological-chargedensities like (1.3) for general non-abelian gauge-bosons, featuring anomalous couplings totwo-photons, being low-mass and couple to two photons, to name a few. We will give moredetails below, but here we just note that the abuse is so overwhelming that we must bow toit and introduce here a disclaimer about nomenclature. Henceforth we will denote the trueQCD “axion” as “QCD axion” or “the axion”, particles with similar properties as axion-likeparticles (ALPs) and the whole family as simply “axions”. The QCD axion will be A and ageneric ALP will be denoted as a . Couplings like (1.5) will be generic for ALPs but the massrelation is exclusive of the QCD axion. It is extraordinary that axions appear so profusely– 3 –n extensions of the SM. They are often encountered as pseudo-Nambu-Goldstone bosons ofnew global symmetries (spontaneously broken at a high energy scales) and in holomorphictheories of dynamical couplings, string theory being a prime example [21–24]. Because ofthe similar properties of the QCD axion and ALPs, axion experiments have the potentialto discover more than the solution of the strong CP problem and the dark matter of theUniverse. An example is the search for solar axions and the proposed International AxionObservatory (IAXO), the focus of this monograph.The two-photon coupling allows the production of axions in the collision of photonsand charged particles via the Primakoff-effect γ + q → a + q . The corresponding solaraxion flux can be calculated to be 1 . × C aγ (10 GeV /f a ) / (m year) with a mean axionenergy (cid:104) E a (cid:105) = 4 . ∼ × C ae (10 GeV /f a ) / (m year) with (cid:104) E a (cid:105) = 2 . C ae is of order unity in some important models like DFSZ [26, 27] orgeneral Grand Unification Theory axions (GUTs) [28], it is absent at tree-level in othermodels, and the loop-induced contribution is very small. Fortunately, one can always resortto the former Primakoff flux. The solar axion flux is indeed copious even for very smallvalues of f a because the Sun is huge, but the chances of detecting solar axions on Earth-based experiments are severely hampered by a small detection probability. Naive estimatesof the natural value of detection cross sections of order σ ∼ g aγ ∼ × − C aγ (10 GeV /f a ) were initially discouraging. However, P. Sikivie presented in 1982 an experimental conceptthat uses the low mass of axions to boost the detection probability by making it coherentover macroscopic magnetic fields, the axion “helioscope” [29, 30].Axions travelling through a transversely polarised B -field can convert into photons atany point along the magnetic region. The photon polarisation is aligned to the external B -field due to the (cid:126)E · (cid:126)B coupling. The conversion probability can be understood as axion-photon oscillations [31] with oscillation length λ a = 8 πE/ (cid:112) m a + (2 g aγ BE ) ( E is the axionenergy), but also as the square of the coherent sum of the conversion amplitudes at eachpoint along the line of sight [32]. After a homogeneous magnetic length L , the probability ofconversion is P ( a → γ ) = (2 g aγ BE ) m a + (2 g aγ BE ) sin (cid:112) m a + (2 g aγ BE ) L E , (1.6)which is coherent ( ∝ L ) if the magnetic length is smaller than the oscillation length L (cid:46) λ a .For small m a this can be very large and so can be the enhancement! For instance, for theparameters proposed for IAXO, this conversion probability is P ( a → γ ) (cid:39) − C aγ (cid:18) GeV f a (cid:19) (cid:18) B (cid:19) (cid:18) L
20 m (cid:19) , (1.7)and stays in the coherent regime for m a (cid:46)
16 meV, making the IAXO search for solar axionsrealistic. Insisting on an incoherent detection scheme, one would need hundreds of kilometresof low-background instrumented detectors to obtain a similar conversion probability with σ ∼ g aγ . This is far too long compared to the most ambitious direct DM experiments lookingfor WIMP recoils, which are also parasitically used to search for solar axions. Their virtue isthat their detection does not rely on macroscopic coherence and therefore they are relativelyinsensitive to the axion mass. Unfortunately, they will not be in the foreseeable future assensitive as indirect stellar evolution constraints, or as current helioscopes, see e.g. [33, 34].– 4 –or helioscopes, the smallness of the axion mass is a virtue, as the coherent lengthcan be macroscopic and the use of long and intense magnets can greatly enhance the axiondetection probability with respect to incoherent detection. This so-called coherent inversePrimakoff-process is at the core of the power of the Helioscope technique to find solar axionswith IAXO but it can lead to other very interesting phenomena.Indeed, the most mature experimental efforts to detect axions in lab experiments [4, 5]rely on their conversion to photons in macroscopic magnetic fields [29, 31]. This approachincludes the search for galactic axion dark matter [29] and the photon regeneration experi-ments (“shining light through a wall”) [30, 32, 35]. There is a large complementarity betweenthe different strategies. While relic axion searches enjoy high sensitivity in terms of g aγ , theytypically accomplish this only for a narrow m a range and rely on the assumption that theDM is mostly composed of axions. Laboratory experiments are free from cosmological or as-trophysical assumptions on the production of ALPs, but they have a comparatively reducedsensitivity. Helioscopes enjoy an attractive compromise between axion source dependency(the solar flux is a robust prediction from any axion model) and competitive sensitivity, aswill be shown. Indeed, a next generation helioscope is the only demonstrated technique thatcan discover QCD axions in the meV mass range [36]. The CERN Axion Solar Telescope(CAST) [37–40] , in operation at CERN for more than a decade, represents the current state-of-the-art of the axion helioscope technique. The International Axion Observatory (IAXO)has been recently proposed as a follow-up of CAST, scaling the helioscope concept to thelargest size realistically allowed. IAXO will be largely based on experience and conceptsdeveloped by CAST.In this paper we provide an updated description of the physics case of IAXO. We organiseour paper by reviewing the contexts in which the discovery of the QCD axion or an ALPin the parameter space accessible by IAXO could have strong implications in other contextsof theoretical particle physics, cosmology and astrophysics. We set the stage in section 2with a detailed theoretical background for axions in extensions of the SM, reviewing recentadvances in axion model building and axion couplings, and placing particular emphasis onstringy axions.Section 3 is devoted to axions as dark matter. Indeed, a great deal of the appeal of axionsis that they can be produced in the early Universe via non-thermal processes (realignmentmechanism and decays of cosmic strings and domain walls [41, 42]), which makes them wellmotivated cold dark matter candidates. The relic density depends on f a , m a but also on theinitial conditions and the cosmological history until today. For the pure QCD axion, one candistinguish between two broad classes: • If the PQ symmetry spontaneously breaks before inflation and is not restored after-wards, the axion field becomes homogeneous during inflation at some a-priori-unknownVEV. A broad range of axion masses can produce the correct relic density just adjust-ing to the adequate initial conditions, including the m A ∼ meV values that could bediscovered with IAXO. • If the symmetry is restored after inflation, initial conditions are reset and the QCDaxion becomes coherent only at very small scales compared to today’s horizon. Anaverage over disconnected patches removes the uncertainty of initial conditions but thequantity of axions radiated from cosmic strings is still uncertain and those radiatedfrom domain wall (DW) decay is model dependent. Models with short-lived DWs( N DW = 1) can give the totality of the DM for m A (cid:38) µ eV, the uncertainty in the– 5 –alculation encompassing the m A ∼ meV values accessible by IAXO. In models withlong-lived DWs, the DM yield can be greatly increased, which points to m A ∼ meV asthe most favoured mass range to account for all the DM in the Universe.We discuss the status of the meV QCD axion as a DM candidate in detail in section 3.General ALP models have ample freedom to become DM candidates in a broad range ofvalues of f a and m a [43].Axions can be also produced from thermal processes or decays of heavy particles, con-tributing to the radiation energy density as effective relativistic degrees of freedom in theearly Universe and/or today. Current cosmological observations present some tension thatcould suggest the presence of this Dark Radiation, in addition to the known neutrino species.For some values of the relevant parameters, this axionic Dark Radiation could give observablesignals in helioscopes like IAXO. This scenario is described in section 4.The fact that ALPs have masses protected from large radiative corrections is verysuggestive to use them as candidates for the inflaton. A recent scenario showed that, foran adequate potential and coupling to photons (for reheating purposes), an axion accessibleto IAXO might indeed be responsible for cosmic inflation. This interplay between ALPs andinflation is reviewed in section 5.The existence of axions can have very important consequences in astrophysics. Indeed,the properties of well-known stellar systems have been providing the strongest constraintson axion properties for a long time [44]. More intriguing axion effects may account forunexplained astrophysical observations. Two of these cases deserve special consideration: theanomalous cooling rate observed in a number of stellar systems and the excessive transparencyof the intergalactic medium to very high energy (VHE) photons. In both contexts, theexistence of very light axions with properties at reach of IAXO has been repeatedly invoked asan explanation, although claims will be taken with caution, as more conservative explanationscannot be excluded. The observational situation as well as the potential interpretation interms or axions are reviewed for both cases in sections 6 and 7, respectively.To conclude the physics case update, we briefly review the conceptual design of IAXO [7]in section 8, present updated sensitivity projections of the experiment and conclude in sec-tion 9. General Relativity (GR) is a perturbatively unitary theory only up to the Planck scale.Whether or not GR is non-perturbatively unitary is still an open question but the require-ment of a Wilsonian (perturbative) unitarization of GR has a very promising venue in stringtheory where point particles are replaced by extended objects: strings. For theoretical con-sistency, string theory requires additional spatial dimensions beyond the known three. Ex-periments and observations constrain these additional dimensions to be compact and verysmall. The compactification schemes determine many of the properties of the low-energyfour-dimensional effective description.The field content in the higher dimensional theory includes many anti-symmetric tensorfields ( p -forms) which may be integrated over any non-trivial cycle in the compactificationgeometry. In four-dimensions, the values of such integrals appear as dynamical scalar fields,and the gauge symmetry of the p -forms translates into a global shift symmetry making– 6 –hem candidate axions [21–24]. These are commonly known as ‘closed string’ axions. Inaddition, D-branes (places where open strings can end) that fill four-dimensional spacetimecan support additional ‘open string’ axions, which can realise the PQ-mechanism [45] orappear as independent ALPs.In some string models, the mass spectrum and couplings of these scalars to ordinarymatter can be explicitly computed [46, 47]. This serves as the starting point for severalinteresting applications to cosmology, astrophysics and particle physics. In particular, onelinear combination of these scalars will play the rˆole of the standard QCD axion if coupledto the QCD sector realised on D-branes, i.e. if leading to the low-energy coupling (1.3). Allother scalars are instead ALPs, which might behave very similarly to the QCD axion butwhose mass and decay constant are not related by QCD scales but by some other detailof the model. Effectively, mass and couplings become independent parameters. Hence thecorresponding parameter space is much wider, featuring for instance regions where ALPscan successfully: ( i ) drive inflation with the production of observable primordial gravitywaves [48], ( ii ) account for the observed dark matter content of the Universe [43, 49], ( iii )contribute to dark radiation by behaving as extra neutrino-like degrees of freedom [50, 51],or ( iv ) become the longitudinal component of extra U (1) gauge bosons with mass well belowthe string scale which could belong to either the visible or a hidden sector [52–54].It is worth mentioning that some of these potential axions might acquire large masses [47].For instance, they could be simply excluded in the compactifications leading to the SM gaugegroup [55]. Moreover, they could be eaten up by anomalous U (1)’s with masses of order thestring scale in the Green-Schwarz mechanism of anomaly cancellation [56]). Finally, theycould acquire a very large mass if they are stabilised in a supersymmetric way since theywould become as massive as the corresponding supersymmetric partners, the so-called sax-ions , which have to be heavier than ∼
50 TeV to decay before primordial nucleosynthesisand avoid altering the successful Big Bang predictions [57–59].For each of the above discussed ‘closed string’ axions, there is generically also a ‘modulus’scalar field that parametrises the size or shape of the compactification manifold. In contrastto the ‘closed string’ axions, moduli fields do not enjoy a shift symmetry, are not in generalprotected from large quantum corrections to their mass and therefore might be easily tooheavy to be considered axions. However, the imaginary part a of a moduli field Φ = φ + i a is exactly shift symmetric and thus is a natural candidate for an axion.The real part, the saxion φ , parametrises either the size or the shape of the extradimensions and comes from the dimensional reduction of the ten-dimensional metric. Hencethe saxion is a gauge singlet with only gravitational couplings to ordinary matter. Moreover,as already mentioned, φ does not enjoy a shift symmetry, and so can be lifted and becomemassive by any kind of perturbative effect. Therefore it is crucial to study moduli stabilisationin order to work out the axion mass spectrum for phenomenological applications. There aretwo benchmark scenarios [22, 47]: • The moduli are fixed by non-perturbative effects. Since the same mechanism givesmass to both φ and a , they will naturally tend to acquire masses of the same order ofmagnitude, which is constrained by the saxion phenomenology to m a ∼ m φ (cid:38)
50 TeV,too massive for our purposes. • The saxion masses are fixed by perturbative effects. Axions will remain massless be-cause they are protected by their perturbative shift symmetry, except for the non-perturbative effects which are now unconstrained by saxion phenomenology. In this– 7 –ast case, the actual value of the axion mass is model-dependent. However, since thenon-perturbative effects depend on exponentials of volume factors, the axions a can beexponentially lighter than saxions and thus naturally very light.Let us stress that fixing all the moduli via non-perturbative effects is highly non-genericbecause of various technical issues that can forbid non-perturbative effects [60]. Therefore weconclude that the low-energy four-dimensional limit of string compactifications genericallygives rise to the presence of multiple light axions which can include the QCD axion plus oneor more axions whose mass spectrum and couplings depend on the microscopic details ofthe physics and the geometry of the extra dimensions [47]. The axions which acquire smallmasses are of course of phenomenological interest.The decay constants of stringy axions depend on their microscopic origin as closedor open string modes. In particular, the size of f a is determined by the geometry of thecompactified dimensions. The closed string axion coupling f a can be either of order thecompactification scale (the Kaluza-Klein scale M KK ) for strings associated with internalcycles in the extra dimensions bulk, or it can be of order the string scale M s , for stringsassociated with resolutions of local singularities [22, 47].Given that M s ∼ g / s M Pl / √V and M KK ∼ M s / V / , where M Pl = 2 . · GeVdenotes the (reduced) Planck mass, g s the string coupling, and V the internal volume instring units, compactifications at large volume and weak string coupling —both preferableproperties of string models under perturbative control— predict f a (cid:28) M Pl .In models where all the moduli are safely heavier than 50 TeV, M s and M KK tend to bearound the GUT scale which is therefore the natural value to expect for the decay constantof closed string axions: f closed a ∼ GeV. Notice also that in the closed string axions theeffective PQ symmetry is always broken in the 4D effective field theory below the Kaluza-Klein scale. However, for sufficiently large V , the decay constant of closed string axions canfall within the intermediate scale window [22] and thus the reach of IAXO. However, in thesemodels with relatively low string and Kaluza-Klein scale, some moduli tend to get lighterthan 50 TeV and cause cosmological problems.Let us turn now to open string axions. They arise as the phases of complex fieldscharged under anomalous U (1)s that could live on a stack of D-branes [45]. In the processof anomaly cancellation, the U (1) acquires a mass of order M s by eating up a closed stringaxion [61]. Therefore, the effective field theory below the string scale contains an effectiveglobal U (1) PQ-like symmetry. The axion decay constant is then set by the magnitude ofthe radial part of the charged open string mode. In turn, in supersymmetric theories thisradial part is set by a model-dependent Fayet-Iliopolous term. For models with D-branes atsingularities where the SM sector is sequestered from the sources of supersymmetry breakingin the bulk [62, 63], the decay constant of open string axions can be suppressed with respectto M s due to either sequestering effects or a mixing of sequestering and U(1) kinetic mixing,leading to an intermediate scale f open a . In some of these models the resulting decay constantcan be as small as f open a ∼ GeV [47, 64], in an interesting region testable by IAXO.In sum, axions are ubiquitous in string theory, and while most axions may be renderedheavy, some may remain light in the low-energy theory and contribute to a variety of physicalprocesses that can be accessible to axion experiments like IAXO. A significantly lower f closed a could be achieved if the cycle supporting the axion is located in a highlywarped region of the compactification, where the effective action is however under less control. – 8 – .2 The low-energy Lagrangian and field-theoretic models The low-energy description of the properties and interactions of an axion is strongly deter-mined by the requirement of possessing a global shift symmetry, a → a + (cid:15)f a . (2.1)If the symmetry is exact, the axion will be massless, as a mass term L a ∝ m a a does notrespect the shift symmetry. Low mass axions can exist if the symmetry is very weaklybroken. There are two conceptually different ways of achieving this breaking. On the onehand, the shift symmetry can simply be explicitly broken by some term in the Lagrangianthat is extremely small. In these cases there is a set of couplings or parameters { p } which,when taken to zero, leave the Lagrangian density invariant under the shift symmetry. Inthis case, the axion mass can receive radiative corrections from the high-energy sector of thetheory, hinting to a possible hierarchy problem. However, they will always be proportionalto these coefficients. If those are very small, the theory can have low mass axions and stillbe technically natural. On the other hand, the shift symmetry might be violated only byanomalous couplings like (1.3). These soft-breaking terms can and do generate non-trivialpotentials for axions but do not generate UV-sensitive radiative corrections to their mass.The resulting low mass axion is therefore naturally light. In practice it is often assumed thatglobal symmetries are not respected by quantum gravity effects because sufficiently classicalblack-holes have no hair and thus can swallow arbitrary global charges. Therefore, unless theglobal symmetry is a remnant from a gauge symmetry or has some other means of protection,one assumes that quantum gravity will generate some kind of explicit symmetry breakingand thus a small potential for axions (and thus a mass). This kind of reasoning will play arole in Sec. 3.In either case, the Lagrangian of a low-mass axion can be divided into a shift-symmetricand a shift-breaking part. The second must be subdominant to consider our axion naturallylight. We assume that our axion is a periodic angular field defined in a ∈ ( − π, π ) v a where v a is a scale related to the spontaneous symmetry breaking of the shift symmetry. Below v a and the electroweak scales, but above the QCD confining scale the Lagrangian of an axioncan be written as L a = 12 ( ∂ µ a )( ∂ µ a ) + (cid:88) ij c aij v a ( ¯ ψ i γ µ γ ψ j ) ∂ µ a − E α πv a F µν (cid:101) F µν a (2.2) − N α πv a G µν (cid:101) G µν a + ... (2.3)where c (cid:48) aij s are dimensionless coupling coefficients, ψ i are the SM quarks and leptons, E and N are the electromagnetic and colour anomalies of the shift symmetry and the ellipsisstands for additional explicit symmetry breaking terms. If N (cid:54) = 0 and additional explicitbreaking terms are sufficiently small, the axion becomes the QCD axion. If we compare with(1.3), we find that we should define f A = v A / N . In this case, it is important to note that V QCD (¯ θ ) is 2 π periodic in ¯ θ and so it will in A/f A . Therefore, there are N physically differentvalues of the QCD axion VEV that minimise the QCD potential and cancel CP violation,¯ θ − (cid:104) A (cid:105) /f A = 0 , , , ...N −
1. Once again, this will be crucial for our section on axion CDM,Sec. 3.Below the QCD confining scale, Λ
QCD , quarks and gluons are no longer adequate degreesof freedom and any axion coupling to G (cid:101) G will mix with the neutral mesons and acquire a– 9 –on-trivial potential. We define the low-energy Lagrangian for a generic axion as L a = 12 ( ∂ µ a )( ∂ µ a ) + (cid:88) ij C aij f a ( ¯ ψ i γ µ γ ψ j ) ∂ µ a − C aγ α πf a F µν (cid:101) F µν a (2.4) − V QCD (cid:18) ¯ θ − N av a (cid:19) + ... where the ellipsis stands now for shift-symmetry couplings to mesons and explicit symmetrybreaking terms. We also define the phenomenological coupling constants g aff (cid:48) = C aff (cid:48) m f + m f (cid:48) f a , g aγ = α πf a C aγ , (2.5)where m f are SM fermion masses.The distinction between f a and v a is only relevant for the QCD axion in our context sowe will use f a = v a for ALPs (we remove V QCD but we allow for other sources of ALP massin the ellipsis). The coupling constant C aγ has a model-dependent contribution ∝ E and amodel-independent contribution from meson-mixing just for the QCD axion [14], C Aγ = − . EN (QCD axion) , (2.6) C aγ = E (ALP) . The couplings to fermions are very similar. They are inherited from the high-energy coef-ficients c aij except for a model-independent part which arises for the QCD axion couplingto hadrons. The relevant couplings for the low-energy phenomenology discussed here are toprotons, neutrons and electrons. The case of an ALP is quite unconstrained so we quote herethose of the QCD axion [14], C Ap = − . C App , C An = − . C Ann (2.7) C Ae = α C Aγ π log (cid:18) Λ QCD m e (cid:19) + c aee (2.8)where ∆ C App , ∆ C Ann are the model-dependent parts that arise from the model-dependentQCD axion couplings to SM quarks. The model-independent axion-electron coupling is zeroat tree-level so we have included the loop correction arising from the axion-photon cou-pling [65], although it is typically very small. Similar loop effects appear for the proton andneutron couplings but are not very relevant.If SM fermions do not transform under the PQ symmetry, QCD axions do not couplewith them at tree level ( c (cid:48) s are all zero). These are called “hadronic axions”, of which theKim-Shifman-Vainshtein-Zakharov (KSVZ) [66, 67] model is an often quoted example. Arecent work [18, 68] has classified a wealth of these models constrained by their cosmologicalviability. Simple models involve one extra charged singlet under the SM gauge group anda new heavy coloured fermion and populate the range 0 . < | C Aγ | < .
75. The rangebroadens when multiple coloured new fermions are allowed [18] but only very special chargeassignments and field contents lead to significantly different values of C Aγ . Therefore, it turnsout that this study provides a very nice bracketing of the photon coupling in axion models,encompassing predictions beyond the KSVZ-type constructions. Therefore, henceforth wewill use the above-mentioned range when we plot the predictions for axion models for theaxion-photon coupling. – 10 –et us now consider the cases where SM fermions transform under the PQ symme-try. The simplest flavour preserving models involve two-Higgs-doublet models: the Peccei-Quinn model itself where f a ∼ v EW —ruled out long ago— and the Dine-Fischler-Srednicki-Zhitnitsky (DFSZ) [26, 27], where an additional SM singlet is responsible for the very largevalue of f a . Two existing DSFZ variants have E / N = 8 / C Ae = cos β/ , sin β/ β = v /v . These potentially large couplings to electrons makethese models especially interesting for explaining the stellar cooling anomalies and to bedetected by IAXO.Introducing new SM Higgs doublets/singlets, new PQ fermions and/or non-flavour di-agonal couplings gives rise to axions featuring a broad range of couplings to SM particles,including flavour violating ones [69], which are strongly constrained by rare decays. Refer-ence [70] demonstrated that, by adjusting the couplings to quarks and leptons, it is possibleto arrange the model-dependent couplings to cancel the model-independent contributions tothe proton, neutron and electron couplings to the QCD axion, the so-called astrophobic axion.This model and those accidentally similar are not severely constrained by stellar evolution(indeed supernova, neutron star, white dwarf and red-giant constraints become essentiallyirrelevant) and might be most advantageously discovered by IAXO itself. It is worth men-tioning that other recently proposed models/constructions like the minimal flavour violatingaxion [71], the axi-flavon [72] and flaxion [73] do not seem to be particularly astrophobic ingeneral, although a recent study focused on minimality found two interesting exceptions [74].Another interesting QCD axion model is SMASH [75, 76], together with other axi-Majoron models [77]. SMASH was born as a minimal self-consistent model solving the mostpressing issues of particle physics and cosmology: neutrino masses, inflation, dark matter,baryogenesis, the strong CP problem and the stability of the Higgs potential. SMASH is aKSVZ-type QCD axion model where the PQ symmetry mixes with lepton number and itsspontaneous breaking gives a majorana mass to right-handed neutrinos. Note that the QCDaxion here plays the role of the majoron as well. The SMASH axion as well as other majoronmodels does not have tree-level coupling to electrons, but they feature a potentially largeradiative correction with right-handed neutrinos in the loop. Another interesting connectionbetween the QCD axion and neutrinos is the Ma-axion [78].At the end of the day, concerning solar axion detection, the main important modeldependency is whether axions have couplings to electrons with C Ae ∼ O (1) like DFSZ, orthe couplings to electrons are loop-suppressed like in KSVZ. In the event of a discovery,IAXO could measure the QCD axion mass as well as the photon and electron couplingsindependently [79, 80]. Therefore it can potentially distinguish between generic possibilitiesand pinpoint very conspicuous models in optimistic conditions. We review this possibility inSec. 8.1. The QCD axion is regarded as one of the best motivated candidates to be the DM of theUniverse. Its couplings to SM particles are suppressed due to a large decay constant f A ,which ensures the stability and collisionless properties of DM. Furthermore, the axions areproduced non-thermally in the early Universe, and hence they are “cold” in the sense thattheir velocity dispersion is small enough to fit the observed large scale structure.– 11 –he cosmological production mechanism of axion DM, called the vacuum realignmentmechanism, was first discussed in [81–83]. In the early Universe, the axion field may havean expectation value which is different from the vacuum expectation value (VEV) at theminimum of the effective potential at the present Universe. The effective potential of theaxion field arises due to instantons: non-perturbative and topologically non-trivial config-urations of the gluon fields in QCD, which become increasingly relevant as QCD becomesnon-perturbative, i.e. as the temperature of the Universe approaches the confining phase-transition around O (0 .
16) GeV. When the QCD potential becomes sizeable, the axion fieldstarts to oscillate around the minimum of the effective potential, and such a coherent oscil-lation of the classical axion field contributes to the matter energy density of the Universe.Taking into account this production mechanism, it was concluded in [81–83] that the QCDaxion DM would overclose the Universe (Ω A >
1) if the decay constant takes values above f A ≈ O (10 ) GeV corresponding to the axion mass of m A ≈ O (10 − ) eV.Although the early discussion described above strongly motivated us to consider theaxion as a DM candidate, the estimate of the relevant mass range was somewhat simplistic andshould be refined in light of recent developments of theoretical and observational cosmology.First of all, the precise estimation of cosmological parameters revealed that DM constitutesonly a fraction of the total energy density of the Universe. According to the recent resultsof the Planck collaboration, the matter density parameter is determined as Ω c h = 0 . ± . h represents the value of the Hubble parameter today, H = 100 h km sec − Mpc − . An order of magnitude improvement in the accuarcy of Ω c leadsto a tighter constraint on the decay constant f A (cid:46) O (10 ) GeV for axion DM producedby the realignment mechanism [85]. Furthermore, the axion DM abundance estimate is notso straightforward if we follow the evolution of the axion field in the context of inflationarycosmology. If the Peccei-Quinn (PQ) symmetry is restored after inflation, topological defectssuch as strings and domain walls are formed, and they could also produce significant amountsof cold axions [86, 87]. On the other hand, if the PQ symmetry is never restored after inflation,such defect contributions can be neglected since their density is significantly diluted due tothe exponential expansion during the inflationary epoch.The difference between the two possible scenarios is sketched in Fig. 1. Here we showhow the spatial distribution of an angular field θ ( x ) evolves over time, where we introducethe dimensionless axion field θ ( x ) = A ( x ) f A , (3.1)which ranges from − π to π . Let us assume that the PQ symmetry has been broken at avery early stage, and that the θ field takes a certain initial value θ i within the Hubble radius ∼ H − I at that time. If inflation occurs at a later time, the physical scale at which θ takesa uniform value θ i exponentially grows, while the Hubble radius remains almost constant.After inflation, the Hubble radius grows with time, H − ∼ t . As shown in the upper panelof Fig. 1, if the PQ symmetry is never restored after inflation, the size of the patch of theUniverse in which θ takes the value θ i can be much larger than the Hubble radius even atthe present time. In this case, we can assume that the axion field has the same initial value θ i throughout the observable Universe at the onset of its coherent oscillation. We call thisscenario the pre-inflationary PQ symmetry breaking scenario.The situation is drastically different if the PQ symmetry is restored after inflation, whichis shown in the lower panel of Fig. 1. If the reheating temperature after inflation is largeenough , the θ field has large fluctuations after inflation, taking random values from − π to π – 12 – igure 1 . A schematic view of the time evolution of the spatial distribution of θ field in the pre-inflationary PQ symmetry breaking scenario (top) and the post-inflationary PQ symmetry breakingscenario (bottom). In the pre-inflationary scenario, inflation makes the initial value of θ = θ i uniformin all the observable Universe. In the post-inflationary scenario, the Universe ends up containing manydifferent patches that had different values of θ at the time of QCD phase transition. We also expectthat topological defects such as strings and domain walls form around the borders of the patches. – 13 –n microscopic scales and restoring the PQ symmetry. After the Universe expands and coolsenough, the PQ symmetry is spontaneously broken and θ takes a certain value θ within theHubble radius at that time. Note that, however, θ takes different values ( θ , θ , . . . ) outsidethe Hubble radius, since such regions are causally disconnected. The Hubble radius continuesto expand as time goes on, and the field relaxes to the uniform value within the new Hubbleradius in order to minimize the gradient energy. Such behaviour continues until the time ofthe QCD phase transition, at which the mass energy of the axion field cannot be neglectedand it starts to oscillate around the minimum of the potential. The typical scale of spatialvariation of the axion field at the onset of the oscillation is given by the Hubble radius at thetime of the QCD phase transition ∼ H − . Such a scale (typically a few comoving mpc if theUniverse is radiation dominated at the time) becomes shorter than the Hubble radius at latertimes, and we expect that our present observable Universe contains many different patcheswhich had different values of θ at the time of the QCD phase transition. Furthermore, wealso expect that topological defects such as strings and domain walls are formed around theboundaries of different patches. Therefore, the relic axion density should be estimated bysumming over all possible field configurations, which include various initial values for θ at theonset of the coherent oscillation and those produced by the collapse of topological defects.We call this scenario the post-inflationary PQ symmetry breaking scenario.Since there is a conceptual difference in the spatial distribution of the axion field betweenthe pre-inflationary and post-inflationary PQ symmetry breaking scenarios, estimates of theaxion DM abundance will be different accordingly. In the following subsections, we willdiscuss these two scenarios separately.
First, let us consider the pre-inflationary PQ symmetry breaking scenario. In this case, wecan ignore the contribution from topological defects, and the relic axion DM abundance canbe estimated by simply following the evolution of the homogeneous axion field θ , which can bedescribed by the standard realignment mechanism. As mentioned earlier, the axion field hasa certain initial value θ i throughout the observable Universe, and the relic axion abundancedepends on this initial value as well as on f A .The axion potential arises from QCD non-perturbative effects and it is thus extremelysuppressed at high temperatures due to the asymptotic freedom of the QCD running coupling.The axion is therefore effectively massless at high temperatures. As the temperature ofthe Universe approaches the QCD confinement scale, the axion mass becomes significantand starts influencing the axion field evolution. This phenomenon can be modelled by thefollowing T − dependent potential, V QCD ( A, T ) = χ ( T ) (cid:20) − cos (cid:18) Af A (cid:19)(cid:21) , (3.2)where we emphasise that χ = χ ( T ) depends on T . Note that we have redefined A/f A → A/f A − ¯ θ for simplicity. Recently, lattice calculations of the topological susceptibility in fullQCD became available [88], whose behavior at high temperatures is close to the power law χ ( T ) ∝ T − n with n = 8 .
16 predicted by the dilute instanton gas approximation [89–91].Below the QCD phase transition, χ becomes a constant χ = (75 . [92]. In whatfollows, we estimate the axion DM abundance by using the latest lattice QCD results [88]. The state of the art calculations of the topological susceptibility χ ( T ) based on lattice simulations of full – 14 –he zero momentum mode of the axion field obeys the following field equation,¨ θ + 3 H ˙ θ + m A ( T ) sin θ = 0 . (3.3)The realignment production of axions occurs when the Hubble friction becomes smaller thanthe potential force, i.e. H (cid:46) m A . At that time the axion field starts to oscillate aroundthe minimum of the potential. The solution of Eq. (3.3) guarantees the conservation ofthe axion number in the comoving volume, from which we can obtain the present density ofaxion DM. Assuming radiation domination during the onset of oscillations, and the harmonicapproximation sin θ ∼ θ one finds [75],Ω A, real h ≈ . (cid:18) θ i . (cid:19) × (cid:16) f A × GeV (cid:17) . for f A (cid:46) × GeV , (cid:16) f A × GeV (cid:17) . for f A (cid:38) × GeV . (3.4)The case f A (cid:38) × GeV corresponds the onset of axion oscillations happening when theaxion mass has already reached its T = 0 value ( χ = χ ). In the case of interest to Helioscopedetection, f a (cid:28) × GeV, the onset of axion oscillations happens before the confiningphase transition, when χ depends very strongly on the temperature. Unfortunately, Eq. (3.4)is based on an approximate solution, which is valid as long as | θ i | (cid:28) π (corresponding to f a (cid:38) O (10 ) GeV). If | θ i | takes a larger value, the factor θ i in Eq. (3.4) is replaced by a θ i -dependent correction term, which can be quantitatively estimated by solving the non-linearfield equation [Eq. (3.3)] numerically. The anharmonic correction factor becomes very large inthe limit | θ i | → π , and in such a case Ω A, real h can account for the total cold DM abundancefor the lower values of the decay constant f A accessible to IAXO. We have computed therequired value of θ i to obtain 100% , , , .
3% and 1% of the observed DM abundanceand plot them in Fig. 2. One can see that very high fine-tunings of θ i are required to have100% for f A ∼ GeV, already noticed by [85], but the values are quite reasonable for afew %. It is precisely in these meV-mass scenarios, in which the direct axion DM detectionis more difficult and one predicts typically a small DM abundance, where IAXO can be theonly way of discovering the QCD axion.In the pre-inflationary scenario, quantum fluctuations of an axion field during inflationlead to isocurvature fluctuations in its realignment dark matter density that are imprintedin the temperature fluctuations of the cosmic microwave background, whose amplitude isstrongly constrained by observations [99, 100]. This constraint applies to any ALP DMcandidate, not only to QCD axions. The upper limits on isocurvature fluctuations result in aconstraint that relates the Hubble scale during inflation H I (which gives the typical size of thefluctuations in the a field), the fraction of axion cold DM to the total, r aDM = Ω a, tot h / . f a (during inflation) that might depend on anharmonic effects if θ i is large. In the simplest cosmological scenario, the constraint is [101] r aDM d ln Ω a dθ i H I πf a (cid:46) − . (3.5)The case of the QCD axion is shown in Fig. 2 (right) as an upper bound on H I for differentvalues of r ADM . Note that the above graph uses our numerical calculations of θ i based QCD result in some discrepancy among different numerical treatments [88, 93, 94]. In particular, the resultof [93], which shows a much milder temperature dependence, leads to an order of magnitude smaller value forthe axion mass explaining the total DM abundance compared with that obtained in [88]. Such a result mightbe interpreted as huge lattice artifacts caused by strong cutoff effects [88, 94]. – 15 – − − m A [eV] − t a n ( θ i / ) Ω A / Ω c . . f A [GeV] − − m A [eV] − H I [ G e V ] Ω A / Ω c . . f A [GeV] Figure 2 . Left: Value of the misalignment angle after inflation (initial conditions of the axion DMfield) required to obtain today different fractions of the observed DM abundance. Right: Upper boundon the Hubble constant during inflation from Planck’s absence of isocurvature fluctuations for axionDM having such DM fractions. These bounds can be evaded in a number of scenarios [95–98]. Bothcalculations are performed with the topological susceptibility and equation of state of the SM from[88]. on [88], which is specific to the QCD axion. One can see that values of f A ∼ GeVand a significant contribution of axion DM r ADM ∼ O (1), require quite low H I . Notethat a bound on H I constrains the maximum possible reheating temperature to be T RH ∼√ H I M Pl and this is constrained to be T RH (cid:38) H I period ofinflation may also be responsible for generating a large enough misalignment angle to giverise to QCD axion or ALP DM [108, 109].Future CMB polarisation probes like CMB-S4 and LiteBIRD can measure the CMBpolarisation effect of gravitational waves from which we can obtain the value of H I . However,due to the moderate sensitivity improvement with respect to the current limits the measuredvalues will not be far from H I ∼ GeV. Thus, if a future CMB polarisation experimentmeasures H I , we will be forced to consider high scale inflation, which limits even tiny fractionsof axion DM in this pre-inflation scenario. It is worth noting here that the isocurvaturespectrum can be reduced in a number of non-minimal models [95–98], so pre-inflationaryphenomenology is not completely eliminated in this scenario. However, we might say thatthe simplest axion models of high-scale inflation favour the scenario where the PQ symmetryis broken after inflation, which we discuss in the following subsection. The axion appears as a pseudo Nambu-Goldstone boson in the low energy effective theory,and such a description is only valid at energies below the scale of symmetry breaking v A .Once the symmetry is restored, we cannot use the effective field theory description in terms of– 16 –he axion field A ( x ). Instead, we can consider the evolution of a gauge singlet complex scalarfield Φ( x ) (the PQ field for QCD axions), which transforms as Φ → Φ e iα with α being a realconstant parameter under the global U(1) PQ symmetry. The PQ symmetry is spontaneouslybroken when the PQ field acquires a VEV |(cid:104) Φ (cid:105)| = v PQ / √
2, and after that the axion field A ( x ) is identified as an angular direction of the PQ field, i.e. Φ = ( v PQ / √
2) exp( iA ( x ) /v PQ ).The VEV of Φ, v PQ here plays the role of v A . The Lagrangian for the PQ field is given by( √− g ) − L = ∂ µ Φ ∂ µ Φ ∗ − V eff (Φ , T ) , (3.6)where V eff (Φ , T ) represents the finite-temperature effective potential for the PQ field, V eff (Φ , T ) = λ (cid:32) | Φ | − v (cid:33) + λ T | Φ | . (3.7)If we assume that the reheating temperature T R after inflation is sufficiently high, T R (cid:29) v PQ ,the PQ symmetry is restored ( (cid:104) Φ (cid:105) = 0) at early times. Subsequently, it is broken when thetemperature drops below T (cid:46) v PQ . When the U(1) PQ symmetry is spontaneously broken, vortex-like objects called stringsare formed due to the Kibble mechanism [115]. After their formation, the string networkevolves according to an approximate scaling solution [116–118], where the typical length scaleof long strings is given by the cosmic time. In order to maintain the scaling property, theenergy stored in strings is dissipated as radiation of massless axions, which leads to a furthercontribution to the present axion DM abundance [86].In addition to the dynamics of strings, we must take into account that of domain walls,which appear at the epoch of the QCD phase transition. The appearance of domain walls isunderstood through the effective potential of the axion field [Eq. (3.2)] written in terms of v PQ , V QCD ( A, T ) = χ ( T ) (cid:20) − cos (cid:18) N DW Av PQ (cid:19)(cid:21) , (3.8)where N DW = N is the colour-anomaly of the PQ symmetry, which turns out to be a positiveinteger and it is called the “domain wall number” in this context. The above potentialexplicitly breaks the U(1) PQ symmetry into its discrete subgroup Z N DW , in which the QCDaxion field transforms as A → A + 2 πv PQ k/N DW ( k = 0 , , . . . , N DW − χ ( T ) → N DW degenerate vacua (cid:104) A (cid:105) k = 2 πv PQ k/N DW . Since the field value (cid:104) A ( x ) (cid:105) must be uncorrelated over distances larger than the Hubble radius at the QCD phasetransition ∼ H − , the QCD axion field relaxes into different vacua from one Hubble volumeto another. Continuity of the VEV demands that there must be a sheet-like boundary betweenthese regions in which the energy density is as high as V QCD ≈ χ ( T ). Such a non-trivialfield configuration is called a domain wall, and its formation is an inevitable consequence ofthe QCD axion model [119]. The PQ symmetry may also be restored non-thermally due to non-perturbative field dynamics afterinflation [75, 110–114] . The evolution of topological defects in this case is similar to that in the thermallyrestored case once they enter the scaling regime, and there is no significant difference in the prediction ofDM abundance between the two cases. However, in the non-thermally restored case, relativistic axions can beabundantly produced, which tends to violate dark radiation constraints if such axions are not thermalized [75]. – 17 – igure 3 . Effective potential of the PQ field (top) and that of the axion field (bottom). The left (right)panel shows the potential at the temperature much higher (lower) than the critical temperature of theQCD phase transition. The axion field corresponds to the direction along the bottom of the potential V (Φ) (pink lines), and such a direction is exactly flat at high temperatures. At low temperatures,the periodic axion potential appears due to the presence of the topological susceptibility χ ( T ), whichgives rise to N DW different minima. In these figures, we choose N DW = 4. Figure 4 illustrates a two-dimensional section of topological defects in the axion modelwith N DW = 3. We note that in every case strings are attached by N DW domain walls, sincethe value of the phase of the PQ field (cid:104) A ( x ) (cid:105) /v PQ must continuously change from − π to π around the string core. Therefore, we expect that hybrid networks of strings and domainwalls, called string-wall systems, are formed at around the epoch of the QCD phase transition.The domain wall number N DW determines the number of degenerate vacua in the ef-fective potential of the A field. We have already mentioned that in models where only oneglobal U(1) PQ symmetry exists, the value of N DW is determined by the colour anomaly co-efficient, N . In simple models this coincides with the number of new quark flavours thattransform under the global U(1) PQ symmetry [119]. For instance, in the KSVZ model thereis one hypothetical heavy quark which transforms under U(1) PQ and N DW = N = 1 whilein DFSZ-I all of the standard model quarks transform under U(1) PQ and we have N DW = 6.The subsequent evolution of the string-wall systems differs according to whether N DW =1 or N DW >
1. If N DW = 1, strings are pulled by one domain wall, which causes thedisintegration into smaller pieces of a wall bounded by a string [120]. Therefore, thesestring-wall systems are unstable, and they collapse soon after their formation. On the otherhand, if N DW >
1, the tension force of domain walls acts on strings from N DW different– 18 – tring �� �� =- � π � �� �� =- � π ��� �� = � π ��� �� = � π ��� �� = � Figure 4 . A top view of string-wall systems with N DW = 3. Gray circles represent the cores ofstrings, which are attached by three domain walls. Coloured lines correspond to the position of thecentre of domain walls. Domain walls exist around boundary of three disconnected vacua, in whichthe axion field has a value A/v PQ = 0, 2 π/
3, and − π/
3. Around the centre of domain walls, it hasa value
A/v PQ = π/ π (magenta), and − π/ directions, which makes the systems stable. Once such stable string-wall systems are formed,their evolution also obeys the scaling solution, in which the typical distance between twoneighbouring walls is given by the Hubble radius. Since the energy density of such scalingdomain wall networks decays slower than that of radiation or matter, it eventually overclosesthe Universe, which conflicts with many observational results [119, 121]. Therefore, QCDaxion models with N DW > ∼ v PQ , theexplicit symmetry breaking terms appearing from dynamics at much higher energy scales Λcould be represented by higher dimensional PQ violating operators like∆ V = c N Λ (cid:18) ΦΛ (cid:19) N + h.c., (3.9)– 19 –here c N is a dimensionless constant. With this parametrisation, even if Λ is as high as thePlanck scale, one needs to suppress the modulus of the coefficients c N with N < N DW . The string-wall systems areshort-lived in the case with N DW = 1, while they are long-lived in the case with N DW >
1. Inboth cases the estimation of the relic QCD axion DM abundance is quantitatively differentfrom the conventional case, since we must take account of the contribution from the collapseof string-wall systems. In the next subsection, we briefly review the recent results on theaxion DM abundance and relevant mass ranges in these two scenarios.
In the post-inflationary PQ symmetry breaking scenario, the present total QCD axion DMabundance can be somewhat artificially split into a sum of three contributions, 1) the con-tribution from the realignment mechanism, 2) that from global strings, and 3) that from thedecay of string-wall systems,Ω A, tot h = Ω A, real h + Ω A, string h + Ω A, dec h . (3.10)We note now that ALPs can also develop string-wall networks and be produced in similarways than QCD axions if they are endowed with periodic potentials with one or severalminima. The QCD axion is peculiar in that we know its potential and its temperaturedependence.We summarize the timeline of the history of the early Universe in Fig. 5. Here, it isassumed that inflation has happened at sufficiently high energy scale, and that the subsequentreheating is so efficient that the PQ symmetry is restored after inflation. The PQ symmetryis spontaneously broken when the temperature of the Universe becomes T ∼ v PQ . At thattime strings are formed, and they continue to produce massless axions until the epoch of theQCD phase transition.The simple picture of the strings decaying into massless axions does not hold once theaxion mass becomes non-negligible, which corresponds to the temperature of the Universeof T ≈ θ , since the valueof θ is different for each Hubble volume. After performing such averaging procedure, thecontribution from the realignment mechanism becomes [75]Ω A, real h ≈ (3 . ± . × − × (cid:18) f A GeV (cid:19) . , (3.11)where the uncertainty originates from the estimation of the topological susceptibility. Thisresult assumes infinite distance from the strings and is thus probably an overestimate [75].For temperatures below T (cid:46) A, dec . Since the contributionfrom the string-wall systems is different according to whether N DW = 1 or N DW >
1, in thefollowing subsections we will discuss these two cases separately.– 20 – nflation ( p ) reheatingrestoration of PQ symmetryPQ symmetry breaking, formation of stringsaxion production from strings: Ω A ,string formation of domain walls, realignment production of axions: Ω A ,real collapse of string - wall systems: Ω A ,dec ( N DW = ) collapse of string - wall systems: Ω A ,dec ( N DW > ) inflaton dominationradiation dominationPQ phase transition QCD phase transition radiation dominationmatter dominationnowT [ GeV ] - - - Figure 5 . A schematic diagram of the thermal history of the Universe in the post-inflationary PQsymmetry breaking scenario. Two possibilities for the evolution of string-wall systems, the case with N DW = 1 and that with N DW >
1, are parallelly shown in the region below T ≈ There has been a lot of controversy on the estimation of Ω A, string h [126–133] andΩ A, dec h [134, 135], which arises from the difficulty in understanding the energy loss processof topological defects and analyzing the spectrum of the axion produced from them in aquantitative way. A straightforward approach to this problem is to perform first principlefield theory simulations of topological defects in the expanding Universe [132]. Along thisline, the computational methods to estimate the energy spectrum of radiated axions havebeen developed in [118, 136–144]. In the following, we review these results and discuss theiruncertainties. – 21 – .3.1 Models with N DW = 1Let us first consider the case with N DW = 1. In this case, the collapse of the string-wallsystems occurs immediately after the formation, and the parameter dependence of the totalDM abundance is similar to that of the realignment contribution (3.11).The number density of axions produced from strings at time t s before the formation ofdomain walls can be estimated as [139] n A ( t s ) (cid:39) ξv (cid:15)t s (cid:34) ln (cid:32) √ λv PQ t s √ ξ (cid:33) − (cid:35) , (3.12)where the parameters ξ and (cid:15) are defined in terms of the energy density of strings ρ string andthe mean energy of axions produced from strings (cid:104) E a (cid:105) , respectively, ξ = ρ string ( t s )( µ string /t s ) , (cid:15) = (cid:104) E a (cid:105) (2 π/t s ) , (3.13)with µ string = πv ln( √ λv PQ t s / √ ξ ) being the tension of strings. The results of field theorysimulations indicate that the network of strings evolves toward the scaling solution, in which ξ takes an almost constant value of O (1) [132, 136]. The results of the simulations also showthat the mean energy of axions produced from strings is comparable to the Hubble scale atthe production time, and (cid:15) takes a value of O (1) [136, 139]. This fact implies that most ofthe axions produced from strings become non-relativistic during the radiation-dominated era,and they contribute to the cold DM abundance. We note that the estimate (3.12) implies thatthe axion abundance scales linearly with ξ . This dependence originates from the assumptionsused to derive Eq. (3.12), that the energy density of strings obeys that of the scaling solutionand that the total comoving energy of the system consisting of strings and axion radiationsis conserved. Adopting the values ξ = 1 . ± . (cid:15) = 4 . ± .
70 suggested in [139] andmultiplying Eq. (3.12) by an appropriate dilution factor, we estimate the contribution fromstrings as Ω A, string h ≈ . +6 . − . × − × (cid:18) f A GeV (cid:19) . . (3.14)In addition to the above contribution of axions produced from strings, there is a contri-bution from those produced from the collapse of string-wall systems. The results of numericalsimulations in [139] indicate that the corresponding abundance is slightly smaller than (3.14),Ω A, dec h ≈ . +2 . − . × − × (cid:18) f A GeV (cid:19) . ( N DW = 1) . (3.15)If we adopt the estimates shown in Eqs. (3.11), (3.14) and (3.15), the total axion abun-dance (3.10) reads Ω A, tot h ≈ . +1 . − . × − × (cid:18) f A GeV (cid:19) . . (3.16) Errors shown in Eqs. (3.14), (3.15), and (3.16) are not the standard uncertainty estimate ( i.e. those fromthe propagation of uncertainty law), but maximum and minimum values obtained by using the results ofnumerical simulations. – 22 –equiring that it explains the cold DM abundance observed today Ω A, tot h = Ω c h (cid:39) . f A ≈ (3 . . × GeV, or the axion mass, m A ≈ (0 . . × − eV.Our estimate so far is based on the result of numerical simulations performed in [139],but there are some debates on the interpretation of the simulation results. In particular, theconventional simulation method [139] cannot realise a large logarithmic enhancement factorin the string tension µ string , and it was pointed out that the effects of such high string tensionmay further modify the estimate of the DM abundance [140, 141].Recent studies have reported small (seemingly logarithmic) corrections to the scalingsolution [118, 140, 144], which indicate a slow increase of the string length parameter ξ .The authors of [118] have studied the uncertainty due to the extrapolation of the simulationresults to realistic values of the string tension and emphasize a much broader uncertainty.The extrapolation includes the uncertainty on the mean energy of radiated axions as well asthe increase of the string length parameter. The main source of the uncertainty on the axionabundance is the ambiguity in the analysis of the spectrum of axions produced from strings.It was pointed out in [118] that a different interpretation on the shape of the spectrumwould alter the prediction for the axion DM abundance by a few orders of magnitude whenextrapolated to realistic values of the string tension. In particular, if the mean energy isindeed comparable to the Hubble scale as claimed in [86, 127, 131, 132, 136, 139], the axionDM abundance can be further enhanced compared to the estimate shown above due to theincrease of the string length parameter. Although such infrared (IR) dominated spectrumis incompatible with recent simulation results [118, 140, 144] obtained based on a stringtension whose value is smaller than realistic ones, we cannot exclude the possibility that theenergy of radiated axion is dominated by IR modes in realistic cases. Intriguingly, simulationsperformed in [118] shows some indication that the spectrum slightly changes towards an IRdominated shape with increasing the string tension, and more careful studies on the axionspectrum in simulations with larger dynamical ranges are warranted to confirm such a trend.We emphasize that even a small change of the shape of the spectrum could drastically enhancethe axion DM abundance when extrapolated to realistic parameter values, and typically theaxion DM mass becomes as large as O (meV) in such scenarios, which is accessible to IAXO.In order to quantify the potentially large uncertainty on the string contribution, let usrecast it to the following form,Ω A, string h ≈ . × − × K (cid:18) f A GeV (cid:19) . , (3.17)where K represents the axion production efficiency from strings, which gives the number ofaxions produced until the radiation from strings is terminated at a time t e around the epochof the QCD phase transition, i.e. n A ( t e ) = KH ( t e ) v , and we have used the condition m A = 3 H to estimate t e . The value of K depends on the logarithmic correction to thestring tension ln( √ λv PQ t e / √ ξ ), which requires extrapolation over an enormous separationrange between √ λv PQ and t − e . The estimate shown in Eq. (3.14) corresponds to K ≈ K as large as (cid:46) × [118] which corresponds tothe axion DM mass of m A (cid:46) . m A ≈ (0 . ± . × − eV [143], that correspondsto the production efficiency K ≈
13 in Eq. (3.17). The results of these simulations also showthat the string networks become denser for larger values of the string tension, which leads to alarger value of ξ ∼ ξ ∼ (cid:15) ∼ (cid:15) = 4 . ± .
70 obtained in the conventional field theory simulations [139]. This implies thatphysics at smaller scales can be relevant to the determination of the axion DM abundance. Wenote that the new simulation method introduced in [142] is based on an effective theory thatbreaks down at some smaller distance scales, and hence it is still not straightforward to figureout how the physics of small scale strings affects the axion production efficiency. Furtherstudies on the dynamics of string-wall networks are required to include precise modeling ofphysics at smaller distance scales.Regarding the fact that there remains the large uncertainty on the estimation of theaxion abundance produced from strings, here we treat K ≈
13 corresponding to the resultof [143] as a lower limit and K (cid:46) × obtained in [118] as an upper limit on the axionproduction efficiency. This corresponds to the following range of the axion DM mass in themodels with N DW = 1,2 . × − eV (cid:46) m A (cid:46) . × − eV ( N DW = 1) , (3.18)or the axion decay constant,1 . × GeV (cid:46) f A (cid:46) . × GeV ( N DW = 1) . (3.19)We again emphasize that the latest simulation results [118] show a trend that the IR modesare getting more important for larger values of the string tension, and that the extrapolationwith such a feature shows a preference for a higher axion DM mass close to the upper limit on m A shown above. Note that this mass range is derived based on the assumption that axionsproduced from strings account for the total cold DM abundance. In other words, IAXO willbe able to probe the parameter space where axions can constitute 100% of the observed coldDM abundance within the range of uncertainty. N DW > N DW >
1, the string-wall systems live longer than those in the case with N DW = 1. Theyeventually collapse due to the effect of the explicit symmetry breaking term. The presentabundance of axions produced from these string-wall systems can be estimated asΩ A, dec h ≈ . × C / d A / ˜ (cid:15) a N (cid:18) Ξ10 − (cid:19) − / (cid:18) f A GeV (cid:19) − / ( N DW > , (3.20)where Ξ ≡ ∆ V v (3.21)– 24 –arameterises the magnitude of the explicit symmetry breaking term ∆ V that induces theenergy bias between different domains. The dimensionless parameters C d , A , and ˜ (cid:15) a ap-pearing in Eq. (3.20) represent the decay time of the string-wall systems, the area of domainwalls, and the mean energy of axions radiated from them, whose values can be estimatedfrom numerical simulations [139]. They generically take values of O (1) but slightly dependon N DW .If the lifetime of the string-wall systems is not sufficiently long, one must take account ofthe realignment and string contributions in addition to Ω A, dec h . For the string contribution,we extend Eq. (3.17) to the case with N DW > A, string h ≈ . × − × KN (cid:18) f A GeV (cid:19) . , (3.22)and assume that K has a similar uncertainty as in the models with N DW = 1. The factor of N just comes from the fact that the string tension is proportional to v , which becomes N f A when we write everything in terms of the axion decay constant.If we assume that the PQ symmetry breaking is protected by a discrete Z N symmetry,and Planck-suppressed operators with Λ = M P , the parameter Ξ in Eq. (3.21) is givenby (3.9). In this case, we have the following parameterisation [125],Ξ = | c N | N N − ( √ N (cid:18) f A M P (cid:19) N − . (3.23)It turns out that N = 9 or 10 is favoured if we require that the axions produced by thestring-wall systems should not overclose the Universe and that the operator ∆ V does notshift the minimum of the QCD potential (3.2) away from θ = 0 even for O (1) values of thephase and/or modulus of c N .Since the magnitude of the energy bias (3.23) becomes more suppressed for smallervalues of f A , the lifetime of the string-wall systems becomes longer in the small f A range.For such long-lived defects, the dominant contribution to the DM abundance is given byEq. (3.20), which places a lower bound on f A from the requirement that the axion abundanceshould not exceed the observed cold DM abundance. On the other hand, if f A takes a largervalue, the lifetime of the defects becomes shorter, and the DM abundance is determined bythe realignment and string contributions, which leads to an upper bound on f A . Therefore,the value of f A is constrained to a finite range in the post-inflationary PQ symmetry breakingscenario with N DW >
1. For instance, the allowed value of f A in the models with N DW = 6and N = 9 reads4 . × GeV (cid:46) f A (cid:46) . × GeV ( N DW = 6 and N = 9) , (3.24)which corresponds to the mass range5 . × − eV (cid:46) m A (cid:46) . × − eV ( N DW = 6 and N = 9) . (3.25) The result of numerical simulations for axionic domain walls with N DW > a, dec h [139]. As mentioned before, the realignment contribution shown in Eq. (3.11) can be an overestimate. Insteadof using it, here we adopt Eq. (3.22) with a lower value of K ≈
13 suggested in Sec. 3.3.1 to obtain upperlimits on f A and lower limits on m A shown in Eqs. (3.24)-(3.27). – 25 –imilarly, for the models with N DW = 6 and N = 10, we have1 . × GeV (cid:46) f A (cid:46) . × GeV ( N DW = 6 and N = 10) , (3.26)which corresponds to the mass range5 . × − eV (cid:46) m A (cid:46) . × − eV ( N DW = 6 and N = 10) . (3.27)Since the relic QCD axion abundance depends on the coefficient | c N | in (3.23) in additionto f A , axions can explain the total cold DM abundance in the whole mass ranges describedabove, up to the tuning of the coupling parameter c N and the uncertainty on the productionefficiency K from strings. In particular, axions can account for the whole DM in the parameterrange f A ≈ O (10 –10 ) GeV and m A ≈ O (10 − –10 − ) eV with a mild tuning of the phase ofthe parameter c N [125, 145]. Interestingly, such a parameter range agrees with that indicatedby the anomalous cooling of stars in various evolutionary stages, which will be discussed inthe next section.We plot the parameter range where QCD axions can account for the observed DMabundance for the models with N DW = 1 and for the models with N DW = 6 in Fig. 6.We see that the predicted value for m A can be much larger than the conventional estimation m A ≈ O (10 − ) eV due to the contribution from strings and string-wall systems. In particular,the axion can be DM in the meV mass range both for the models with N DW = 1 and thosewith N DW >
1, and such a parameter range can be decisively probed by IAXO.
According to the standard cosmological model, ΛCDM, the energy budget of the Universetoday consists of (in order of decreasing magnitude): dark energy, dark matter, ordinarybaryonic matter, and relic neutrino and photon radiation. However, the dark sector may besignificantly richer than this phenomenological model suggests. Extensions of the StandardModel commonly include light particles in the dark sector, e.g. axions and hidden photons.Beyond their potential role as cold DM, these light particles may have been produced in thehot Big Bang from thermal processes or decays of heavy particles and would then contributeto the energy density of the Universe as a ‘dark radiation’ component. The observationalsuccess of the ΛCDM model indicates that dark radiation, if it exists, must contribute tothe total energy density today by an even smaller amount than the CMB photons and theC ν B neutrinos. By convention, the dark radiation is parametrised phenomenologically as the‘excess number of neutrino species’, ∆ N eff : ρ d . r = ρ rad − ρ CMB − ρ CνB = 78 (cid:18) (cid:19) / ∆ N eff ρ CMB , (4.1)where ρ rad denotes the total relativistic energy density. The CMB data from Planck (whencombined with other experiments) is consistent with ∆ N eff = 0, but a significant dark radia-tion energy density is still allowed within the 68% error bars of σ (∆ N eff ) = 0 .
23 [146]. Darkradiation is also capable of reducing the persistent tensions between local measurements of H [147] and the value inferred using ΛCDM and CMB data. Dark radiation present atthe time of Big Bang nucleosynthesis (BBN) would increase both the expansion rate andthe freeze-out abundance of neutrons, and consequently affect the light element abundances.Observational determinations of the primordial deuterium abundance give no evidence for– 26 – a (eV)10 − − − − − − | g a γ | ( G e V − ) − − − − − − − − ALP miracle
CAST
BabyIAXOIAXO A x i o n m o d e l s A L P D M m o d e l s K S V Z Haloscopes N D W = N D W = , N = N D W = , N = P o s t - i n fl a t i o n D M m o d e l s Figure 6 . The predicted mass ranges in which QCD axions can account for the observed cold DMabundance in the post-inflationary PQ symmetry breaking scenario. The generic prediction of QCDaxion models is plotted in the yellow region. The sensitivity prospects of IAXO are also plotted.The prediction of the post-inflationary PQ symmetry breaking scenario differs according to the valueof N DW and the structure of the explicit symmetry breaking terms in the models with N DW > N DW = 1 (red), N DW = 6 and N = 9 (gray), and N DW = 6 and N = 10 (blue). dark radiation with error bars comparable to Planck, σ (∆ N eff ) = 0 .
28 [148]. Future CMBexperiments will be significantly more sensitive to dark radiation: the ground based ‘Stage-3’and ‘Stage-4’ CMB polarisation programmes state a projected sensitivity of σ (∆ N eff ) = 0 . σ (∆ N eff ) = 0 .
02, respectively.
The definition of ∆ N eff is chosen so that an additional neutrino species that initially was inthermal equilibrium with the photon-baryon plasma and decoupled simultaneously with theStandard Model neutrinos gives ∆ N eff = 1. More generally, any thermally produced, light,feebly interacting particles that decouple at time T d contribute to the dark radiation by:∆ N eff = (cid:18) g (cid:63) ( T ν ) g (cid:63) ( T d ) (cid:19) / × (cid:26) , Scalar . (4.2)– 27 –ere g (cid:63) = g (cid:63) ( T ) denotes the effective number of relativistic degrees of freedom in thermalequilibrium at temperature T . Assuming that the dark radiation component decoupledearlier than the regular neutrinos, the Standard Model contribution to g (cid:63) ( t ) ranges from g (cid:63) ( T ν ) = 10 .
75 to g (cid:63) ( T (cid:38) m top ) = 106 .
75 (see [88]) the latter figure being applicable at veryearly times when even the top quark was in thermal equilibrium. Beyond the Standard Model, g (cid:63) ( T ) may have received contributions from additional particles, and the Standard Modelcontribution in general only gives a lower bound on g (cid:63) ( T ). Hence, a thermally produced ALPthat decouples before T ∼ m top contributes to the dark radiation by:∆ N eff ≤ . , (4.3)which saturates if g (cid:63) never received any contributions other than those of the Standard Modeland the ALP. Figure 7 shows the value of ∆ N eff as a function of the decoupling temperatureunder the SM-only assumption. �� - � �� - � �� - � �� - � �� - � � �� �� � �� � �� - � �� - � � Figure 7 . Extra radiation in an axion as a function of its decoupling temperature assuming onlythe SM degrees of freedom. Dashed lines show the upper limits (2 σ ) from Planck and the future 4thgeneration CMB-S4. A thermal population of QCD axions can be always established through interactionswith gluons at sufficiently high temperatures, which happens provided the reheating temper-ature satisfies T RH > T d where [149] (see also [150–152]): T d ≈ . · GeV (cid:18) f A GeV (cid:19) . . (4.4)Moreover, the efficiency of the thermal production is enhanced by about 3 orders of magnitudeif there is a direct coupling between axion and top quarks [151], and so in this case the required T RH goes down by the same amount.From analysis of Planck data and BBN constraints, the reheating temperature of thehot Big Bang plasma is only bounded from below, T RH ≥ . · − GeV [107], but it is notknown whether T RH far exceeded this bound. Note that a determination of the energy scale of inflation would not model-independently determine the – 28 –ven if the reheating temperature is too low to fully thermalise the ALPs, small con-tributions to the energy density may arise from (non-equilibrium) ALP interactions with thethermal plasma, generically giving ∆ N eff (cid:28) .
027 [149].If axions have direct couplings to quarks and leptons, some of which are poorly con-strained observationally, they can thermalize at temperatures between about 0.1 GeV and100 GeV, for f below 10 GeV, leading to larger values of ∆ N eff , up to about 0.05 for quarksand even up to 0.5 for leptons [152–154], as shown in Fig. 8. Such values are not upperbounds, but predictions, and should be visible by future CMB ‘Stage-4’ experiments, allow-ing for a very interesting interplay between CMB experiments and IAXO. Large values of∆ N eff might even mitigate the present cosmological tension on Hubble constant measure-ments [154], at about 3.5 σ between CMB and Supernovae measurements [84, 155], andclaimed at 3.7 σ [156], and even 4.4 σ [157] more recently. Non-thermally produced axions may potentially give much larger contributions to the darkradiation energy density of the Universe. In fact, substantial levels of axionic dark radiation(corresponding to ∆ N eff ∼ O (0 . direct decay of the field driving reheating into ALPs willproduce a non-thermal cosmic background of relativistic axions, and hence dark radiation.While not specific to string theory, this scenario is a generic consequence of reheating scenariosinvolving weakly coupled moduli fields, as we will now discuss.Compactifications of string theory include geometric deformation moduli and axions ingeneral in the low-energy theory (cf. section 2). During inflation, the compactification geom-etry becomes slightly distorted due to the energy density of the inflaton, and moduli becomedisplaced from their post-inflationary minima. After the end of inflation, the geometry willrelax back to its final vacuum configuration, and will undergo oscillations as it settles down.Specifically, when the Hubble parameter H has decreased to the same order of the mass ofone of the moduli fields, m φ , the modulus starts oscillating around its minimum with aninitial amplitude set by the displacement acquired during inflation. The energy density ofsuch a coherently oscillating field red-shifts only like matter, ρ φ ∼ R − ( R the scale factorof the Friedman-Robertson-Walker metric), and if the modulus is sufficiently long-lived, itsenergy density comes to dominate over any previous radiation component which red-shiftslike ρ r ∼ R − . This way, the most long-lived modulus field comes to dominate the energydensity of the Universe, and the reheating temperature of the hot Big Bang plasma will beset at its decay. Moduli arise from higher-dimensional components of the metric tensor andgenerically couple with gravitational strength interactions, and their decay rate is typicallygiven by, Γ ∼ π m φ M , (4.5)up to an O (1) constant. Clearly, parametrically lighter moduli are more long-lived thanheavier moduli. Assuming instantaneous reheating, T RH is given by (for the more precise initial temperature of the primordial hot Big Bang plasma, as the period between the end of inflation and BBNmay be complicated and include multiple periods of matter and radiation domination, with T r correspondingto the maximal temperature of the final phase. – 29 – × × × × / c [ GeV ] Δ N e ff σ , CMB - S42 σ , CMB - S4 b t c ( T F = ) c ( T F = ) c ( T F = ) c ( T F = ) f - / / c l [ GeV ] Δ N e ff μ , scattering τ , scattering τ , decay σ , futuristic1 σ , CMB - S42 σ , CMB - S4 Figure 8 . Top: ∆ N eff as a function of f /c i . The green, red and blue lines correspond to thepredictions for the charm, bottom and top particle, respectively. The orange bands represent the 1 σ and 2 σ forecasted contours of a next generation CMB experiment (CMB-S4). See also [158] for similarforecasts combining CMB-S4 and large scale structure. Bottom: Contribution to ∆ N eff from muon(blue) and tau (red) scattering as a function of c (cid:96) /f . Decays are possible for off-diagonal couplings;we show only results for tau decays (magenta), which are the only allowed ones in the above range for c (cid:96)(cid:96) (cid:48) /f . Each process is shown as a band, parametrizing the uncertainty in the number of relativisticdegrees of freedom: the straight line g ∗ is taken from [159] and the dashed line from [88]. We also showthe analytical expectation, ∆ N eff ∝ f − / , for non-thermalized axions. The orange bands representthe 1 σ and 2 σ forecasted contours of CMB-S4, plus a more futuristic 1 σ band, according to [160]. – 30 –quation, see e.g. [161]): T RH ∼ (cid:112) HM Pl ≈ (cid:112) Γ M Pl ∼ m / φ M / = 20 MeV (cid:16) m φ
100 TeV (cid:17) / . (4.6)Hence, the observational requirement of thermalisation before BBN and neutrino decoupling T RH ≥ . · − GeV [107], translates into a generic bound on the mass of the lightestmodulus: m φ (cid:38)
50 TeV. The typical values of m φ from string compactifications is not knownin general, but some scenarios that also predict supersymmetry breaking ‘soft’ terms at theTeV scale give m φ ∼ –10 GeV [62, 63, 162, 163] which gives from (4.6) T RH ∼ m φ (cid:38) GeV for the QCD axion) may produce athermal axion dark radiation, leading, in effect, to a suppression of ∆ N eff .The moduli field φ generically has open decay channels into any sufficiently light degreeof freedom. In particular, φ can decay into axions: φ → a + a . Importantly, weakly coupledaxions do not thermalise, and so are more energetic than the average particle in the hot BigBang plasma, E initial a = m φ (cid:29) m φ (cid:114) m φ M Pl . (4.7)As the Universe expands, the axion energy redshifts with the inverse scale factor. However,the ratio of the axion energy to the plasma/CMB temperature remains large until the presentday [161]: (cid:18) E a T (cid:19) now = (cid:18) (cid:19) / (cid:18) g (cid:63) ( t ν ) g (cid:63) ( t RH ) (cid:19) / (cid:18) E a T (cid:19) initial . (4.8)The levels of axions dark radiation produced directly from modulus decay is then determinedby the branching ratio B a of φ decaying into axions, B a = Γ φ → aa / Γ φ → anything . The relevantvalue of ∆ N eff is given by [50, 51]:∆ N eff = 437 B a − B a (cid:18) g (cid:63) ( T ν ) g (cid:63) ( T RH ) (cid:19) / . (4.9)The exceptionally sensitive CMB Stage-4 experiments will probe sub-percent branching ratiosinto axions: in the Standard Model (i.e. for g (cid:63) ( T RH ) in the range 10.75–106.75), a constrainton ∆ N eff < .
02 corresponds to bounds on the branching ratio into axions B a < O (0 . . E a = m φ /
2, however, the decay of the modulus field is not instantaneous, and, when observedtoday, axions created early will be slightly more red-shifted than axions created at a latertime. This results in a ‘quasi-thermal’ shape [161]. For moduli masses of O (10 ) GeV, thecharacteristic energy of this relativistic ‘Cosmic axion Background’ is around O (0 .
2) keV.Particular realisations of this scenario have been considered for various string compact-ifications [50, 51, 164–167], where also the branching ratio into axions can be estimated.Commonly, O (∆ N eff ) ∼ . Relativistic ALPs can produce a number of additional signals beyond their effect on theenergy density of the Universe. These arise through either direct scattering of the axions offmatter, or through axion-photon conversion in background magnetic fields, as explained insection 7. • Scattering of ALPs off the thermal plasma
Direct scattering of axions off the thermal plasma is a telltale feature of non-thermallyproduced axions which are naturally much more energetic than the plasma tempera-ture, cf. (4.8). These high-energy axions can access dynamical processes with a centre-of-mass energy E CoM (cid:29) T , which in the thermal plasma would be suppressed byexp( − E CoM /T ) (cid:28)
1. As these axions are non-thermal, the scattering rate is alwaysΓ (cid:28) H , yet rare scattering events can cause significant deviations from standard cos-mology. For example, scattering off photons during BBN is constrained by the ob-servationally inferred primordial helium abundance [161]. However, scattering off thepre-recombination thermal plasma in the red-shift range z ∼ · will producevery small spectral distortions of the CMB that will be hard to detect [170]. Highlyenergetic ALP scattering may furthermore produce dark matter [161]. • ALP-photon conversion in primordial magnetic fields
ALP dark radiation may also be detected through ALP-photon conversion in astrophys-ical magnetic fields [171]. Particularly interesting are hypothetical primordial cosmicmagnetic fields spanning Mpc distances, and galaxy cluster magnetic fields, which tendto be coherent over kpc scales.The magnitude of primordial intergalactic fields is unknown, but lies between 10 − and10 − G [172]. In the presence of a substantial primordial, cosmic magnetic field, ahighly relativistic Cosmic ALP Background can convert into energetic photons thatwill contribute to the reionization of the Universe. Such dark radiation can be con-strained by its contribution to the optical depth to recombination, τ . Conservativeestimates using Planck constraints on τ and ∆ N eff indicate that this could lead to acombined constraint of g aγ B ≤ − GeV − nG [173], where B denotes the cosmicmagnetic field, which is observationally constrained to B ≤ nG [174]. Hence, if a Cos-mic ALP Background is detected by other means, it will provide strong constraints onthe strength of cosmological magnetic fields, and vice-versa. • Soft X-ray excess from galaxy clusters
The magnetic fields of galaxy clusters are even more interesting, as they are well con-strained observationally. The fields of galaxy clusters are typically O (1 − µG instrength (e.g. see [175]). These magnetic fields are measured via observations of Fara-day rotation of radio sources located in or behind clusters. On passing through amagnetic field, the polarisation angle φ rotates with wavelength, λ , as, φ = φ + RM λ , (4.10)– 32 –here the rotation measure is given by: RM = 812 (cid:90) (cid:16) n e − (cid:17) (cid:18) B (cid:107) µ G (cid:19) (cid:18) dl kpc (cid:19) (4.11)along the line of sight. As the electron density in a cluster n e ( r ) is well-determined fromX-ray measurements, radio observations of the Rotation Measure lead to a statisticalcharacterisation of the magnetic field strength (although as the magnetic field reversesdirection many times along the line of sight, it is not possible to know the exact 3-dimensional field configuration). The rate of variation in the Rotation Measure alsoallows the typical coherence lengths of the magnetic field to be estimated as 1 - 10 kpc,with the field extending throughout the ∼ n e inside a galaxy cluster is between 10 − and 10 − cm − and the typical extent of a galaxy cluster is around 1 Mpc. For ALP masses less thanaround 10 − eV (the plasma mass within galaxy clusters), this implies that at X-rayenergies galaxy clusters provide ideal environments for ALP-photon conversion. Indeed,at X-ray energies the ALP-photon interconversion probability on passing through a clus-ter is O (1) when g aγ ∼ − GeV − . The conversion probability is energy-dependent,and maximises at E (cid:38) keV energies in the galaxy cluster environment. Detailed simu-lations of ALP-photon conversion passing through the Coma cluster at various energiescan be found in [176]. These both confirm these general results and also show thepresence of a threshold energy for conversion to occur efficiently (for the Coma cluster,this is around 50 eV).Non-thermally produced ALP dark radiation arising from the decay of a modulus with m φ = 10 GeV would currently have a mean energy of around 200 eV [161]. For g aγ inthe range 10 − GeV − to 10 − GeV − this would lead via ALP-photon conversion toan additional contribution to cluster X-ray spectra in the O (0 . O ( ≤ . Significant conversion probabilities can also be obtained by ‘resonant’ conversion of lower-energy ALPsinto photons if the ALP mass is very close to the local plasma frequency. – 33 – (cid:45) (cid:45) (cid:45) (cid:45) E (cid:42) (cid:64) eV (cid:68) g a Γ (cid:64) G e V (cid:45) (cid:68) Figure 9 . IAXO sensitivity to the axion/ALP two-photon coupling of cosmological dark radiation(DR). Lines correspond to 1 event/year for ∆ N eff = 0 . , . , . ALP dark radiation may explain this galaxy cluster soft X-ray excess in the parameterspace region with m a < − eV and g aγ ∼ − GeV − which is within the reach ofdetectability of IAXO. • Direct detection with IAXO
For obtaining the sensitivity of IAXO to dark radiation we parametrize the flux pro-duced by the modulus φ partially decaying to ALPs as d Φ dE = 2 × × ∆ N eff × EE ∗ e − ( E/E ∗ ) (cid:20) s keV (cid:21) , (4.12)where E ∗ is the average ALP energy. Under the assumption that negligible backgroundis achievable down to a threshold of 0.2 keV, the sensitivity of IAXO to dark radiationaxions/ALPS would be the one shown in Fig. 9, as a function of E ∗ . More realisticprospects depend on the experimental parameters (background and threshold) actuallyachieved with the technologies of choice to extend IAXO energy window to lower values,and will be studied in the near future. The precise measurements of the CMB temperature and polarization anisotropies providestrong support for the slow-roll inflation paradigm [180]. Successful inflation requires anextremely flat inflaton potential for minimally coupled scalars, which is naturally realised ifthe inflaton is an axion. – 34 –he simplest axion inflation is the natural inflation with the following potential [181,182], V inf ( a ) = Λ (cid:18) − cos (cid:18) af a (cid:19)(cid:19) , (5.1)where the inflation scale is set by the energy scale Λ. The above potential leads to the large-field inflation where the inflaton field excursion exceeds the Planck mass. In fact, the decayconstant f a is constrained to be f a (cid:38) M Pl [180]. Such a super-Planckian decay constanthas been questioned from the perspective of a string theory [183–187] and quantum gravityin general . Even if such a large decay constant is justified, the predicted spectral index n s and tensor-to-scalar ratio r are not favoured by the current CMB observations.From the effective field theory point of view, there are two possible simple ways toovercome both the trans-Planckian values of the decaying constant and the incompatibilitieswith current CMB data. The first one invokes the presence of multiple non-abelian gaugefields coupled to the axion in a shift invariant way. [188–191]. Here we assume that the axionpotential consists of multiple cosine terms with different height and period.In the minimal case, the potential consists of two cosine terms [188]: V inf ( a ) = Λ (cid:18) cos (cid:18) af a + ϑ (cid:19) − κn cos (cid:18) naf a (cid:19)(cid:19) + const ., (5.2)where n ( >
1) is a rational number, κ is a numerical coefficient, ϑ is a relative phase, andthe last term is a constant to realize the tiny cosmological constant at present. The quartichilltop inflation is realized for ϑ ≈ κ ≈
1. The inflation takes place in the vicinityof the origin where the curvature, or the effective mass, is much smaller than the Hubbleparameter during inflation. After inflation the inflaton oscillates about one of the potentialminima, a min . Interestingly, if n is an odd integer, the axion mass at the potential maximumand minimum is equal in magnitude but has an opposite sign, i.e., the axion has a flat-topand flat-bottomed potential. As a result, the axion is very light and cosmologically stable inthe present vacuum, and it may also account for DM. This opens up a possibility to unifyinflation and DM in terms of an axion which can be searched for by experiments. In thefollowing subsections we briefly summarize implications of the axionic unification of inflationand DM [192].The second possibility is to change the axion propagator by non-minimally couplingto gravity. The only gravity-axion interaction conserving the tree-level shift invariance isEinstein tensor coupling [193] − G αβ M ∂ µ a∂ ν a , (5.3)where M is a new mass scale . In the limit of ϑ = 0 and κ = 1, the inflation model with (5.2) reduces to the quartic hilltopinflation, where the inflaton is massless both at the potential maximum and minimum. Thiscase is ruled out by current observations implying a spectral index of order n s = 0 . ± .
006 [180]. The cosmological constant cannot be larger than M P . Note that this is not the unitarity violation scale [193]. – 35 –he predicted spectral index can be increased to give a better fit to the observation byallowing a non-zero ϑ in the small field scenario [188, 194]. Interestingly, this implies a smallaxion mass, m a ≡ V (cid:48)(cid:48) ( a min ) (cid:39) (cid:32)(cid:18) n − (cid:19) ϑ Λ f (cid:33) , (5.4)where we have approximated | ϑ | (cid:28)
1. Then the observed spectral index fixes the inflatonmass as m a H I ∼ | V (cid:48)(cid:48) ( a ∗ ) | H I = O (0 . , (5.5)where H I is the Hubble parameter during inflation, and a ∗ is the axion field value at thehorizon exit of cosmological scales. We emphasize that the second equality of (5.5) holdsgenerally in small-field inflation. A similar equality holds when n s is increased also by allowing κ (cid:54) = 1. Thus, the Planck normalization of curvature perturbation and the observed spectralindex determine the relation between the decay constant and the axion mass [192, 195] f a (cid:39) × GeV (cid:16) n (cid:17) (cid:16) m a (cid:17) . (5.6)It seems that the QCD axion cannot behave as described here without changing radically thebehaviour of the QCD potential and compromising the solution to the strong CP problem,however an ALP can very well realise this model. Moreover, if we couple the inflaton tophotons we can obtain successful reheating. The coupling to photons is thus given as afunction of the ALP mass as, g aγ (cid:39) × − C aγ (cid:16) n (cid:17) − (cid:16) m a (cid:17) − GeV − . (5.7) After inflation, the ALP oscillates about the potential minimum, and the Universe is dom-inated by coherent oscillations of the ALP. Since the ALP mass is much smaller than thetypical curvature ∼ Λ /f , the potential is well approximated by a quartic one. As theUniverse expands, the averaged oscillation amplitude, a ( t ), gradually decreases inverselyproportional to the scale factor.The ALP decays and dissipates into plasma through its coupling to photons. First theALP decays into two photons with the rate,Γ( a → γγ ) = g aγ π m eff ( t ) , (5.8)where m eff ( t ) ≡ | V (cid:48)(cid:48) ( a ( t )) | / is the effective mass of the ALP when the oscillation amplitudeis large. The decay proceeds until it is kinematically blocked by the thermal mass of photons.Afterwards, the ALP gradually dissipates into plasma through scatterings of photons andelectrons off the ALP condensate. The dissipation rate is suppressed by the effective ALPmass [196], and given by, Γ dis ,γ = C dis ,γ g aγ T (cid:18) m e T (cid:19) (5.9)– 36 –here C dis ,γ is a numerical factor of O (1 −
10) that represents uncertainties of the order-of-magnitude estimate as well as the effects of tachyonic preheating and the subsequentself-resonance [197–200]. When the plasma temperature exceeds the weak scale, one needsto take into account the dissipation through similar couplings to the weak gauge bosons.The effective ALP mass decreases as the averaged oscillation amplitude becomes smaller.At a certain point, the dissipation becomes ineffective, and the reheating ends. To quantifythe efficiency of the reheating, we define ξ as the ratio of the energy density of the remnantALP to the total energy density just after the reheating. For instance, if ξ = 0 .
1, 90 % ofthe initial ALP energy dissipates into plasma, while the rest becomes the remnant.Once the decay and dissipation rates are given, one can easily solve the Boltzmannequation to evaluate ξ . In order for the ALP to dissipate most of its energy, say, ξ (cid:46) O (0 . g aγ must be larger than O (10 − ) GeV − , taking account of theabove mentioned uncertainties. In this case, the reheating ends in several Hubble times afterinflation, and the reheating temperature is estimated to be T RH (cid:39) (cid:18) g (cid:63) ( T RH )106 . (cid:19) − (cid:16) m a (cid:17) TeV . (5.10)In addition, ALPs are also thermalized in plasma, and its abundance is given by∆ N eff (cid:39) . , (5.11)where it decouples at a temperature slightly below the top quark mass. Thermalized ALPsbehave as dark radiation during nucleosynthesis, and they become hot DM at a later time,suppressing the matter power spectrum at small scales. Both effects can be searched for bythe future CMB and large-scale structure observations. After the reheating becomes ineffective, a small amount of the ALP is left over. The remnantALP is stable on cosmological time scales, and so, it contributes to DM. The energy density ofthe ALP remnant, ρ a , first decreases like radiation since the potential is well approximated tobe a quartic potential. Then, ρ a starts to decrease like matter when its oscillation amplitudebecomes so small that the ALP mass becomes non-negligible. Assuming that the ALPremnant accounts for the observed DM density, the transition temperature is given by T c (cid:39) . ξ − (cid:18) g (cid:63)s ( T RH ) g (cid:63)s ( T c ) (cid:19) eV , (5.12)where T c is the temperature at the transition. Since the ALP condensate behaves like darkradiation at T > T c , it suppresses the small-scale matter power spectrum, which is constrainedby the SDSS and/or Lyman- α data. In order to be consistent with the observed data, thetransition should take place no later than the redshift z c = O (10 ) [201]. Then the ratio ξ isbounded above: ξ (cid:46) . (cid:18) × z c (cid:19) , (5.13)where we substituted g (cid:63) ( t RH ) = 106 .
75 and g (cid:63)S ( T c ) (cid:39) . g aγ (cid:38) O (10 − ) GeV − . (5.14)– 37 – bservable Stellar System Proposed Solution(s) References Rate of period WD Variables axion or ALP coupled to electrons; [205–209]change neutrino magnetic moment;Shape of WDLF WDs axion or ALP coupled to electrons; [210, 211]Luminosity of Globular Clusters axion or ALP coupled to electrons; [212–214]the RGB tip (M5, ω -Centauri) neutrino magnetic moment;R-parameter Globular Clusters axion or ALP coupled to photons; [215, 216]B/R Open Clusters axion or ALP coupled to photons; [217–220]Neutron stars CAS A axion or ALP coupled to neutrons. [221] Table 1 . Summary of anomalous cooling observations [222]. All the anomalies have, individually,about 1-2 σ statistical significance, except for the last two for which the significance has not beenquantified. After the transition, the ALP remnant behaves like CDM. The observed DM abundanceis explained if the ALP mass is given by m a ∼ . x − (cid:18) ξ . (cid:19) − eV , (5.15)where x is a numerical factor of order unity which parametrizes the typical oscillation am-plitude of the ALP remnant as a (rem)0 = x ξ / f .Combining (5.7), (5.13), (5.14), and (5.15), we arrive at the unique parameter region,0 .
01 eV (cid:46) m a (cid:46) , (5.16) g aγ = O (10 − ) GeV − , (5.17)where both inflation and DM are simultaneously explained by the ALP. It is highly non-trivial that such a viable region exists without running afoul of the current experimentaland observational limits. We refer to this coincidence as the ALP miracle [192]. The regionof the ALP parameter space consistent with the CMB observation has been more carefullyevaluated in a recent work [195], and it is shown in Fig. 6. Interestingly, the predicted sweetspot of the ALP miracle significantly overlaps with the sensitivity reach of IAXO.
Independent observations of diverse stellar systems have shown deviations from the predictedbehaviour, indicating in all cases an over-efficient cooling [145, 202–204]. These deviations,often referred to as cooling anomalies , have been observed in: several pulsating whitedwarfs (WDs), in which the cooling efficiency was extracted from the rate of the periodchange; the WD luminosity function (WDLF), which describes the distribution of WD asa function of their brightness; red giants branch (RGB) stars, in particular the luminosityof the tip of the branch; horizontal branch stars (HB) or, more precisely, the R-parameter,that is the ratio of the number of HB over RGB stars; helium burning supergiants,more specifically the ratio B/R of blue and red supergiants; and neutron stars. Table 1summarizes the results. – 38 –urthermore, axions may have observable effects on the late evolutionary stages ofmassive stars, for example influencing the nucleosynthesis [223]. Axions with couplings atreach of IAXO would also modify the threshold initial mass for which carbon is ignited inthe stellar core (Mup ), ultimately shifting the required mass to explode as core collapse SNby 1-2 M (cid:12) , depending on the couplings [224, 225]. Finally, a recent study indicates thataxions reduce the final (pre-SN) luminosity expected for a star of a given initial mass [226].Presently, observations are sparse but they do seem to indicate a preference for the axionscenario, with couplings somewhat below the recent CAST bound and likely accessible toIAXO and possibly BabyIAXO.It is quite remarkable that such different stellar systems show, systematically, an ex-cessive amount of energy loss, indicating a lack of understanding in the current modelingof stellar cooling. Notice that the amount of anomalous energy loss is different in differentobjects. For example, main sequence stars, such as our Sun, do not show this anomalousbehaviour and, in general, the amount of exotic cooling indicates some correlation with theinterior temperature of the stellar object. The accumulation in the recent years of data fromsuch different stellar objects offers a unique possibility to revise our understanding of thecooling mechanisms in stars.An appealing explanation to these anomalous observations is to assume that the sourceof the exotic cooling is a new light, weakly interacting particle, produced in the stellar coreand able to stream freely outside, carrying energy away, in much the same way as neutrinosdo. Interestingly, the peculiar dependence of the observed additional cooling on temperatureand density is extremely selective on the possible particles which could account for all theobservations, and indicate a clear preference for a pseudoscalar (axion-like) particle coupledto photons and matter [203]. Indeed, given the large variations in temperature and densityof the systems considered, it is quite remarkable that one particle alone can simultaneouslyaddress all the observed problems, a fact that, in our opinion, strengthens considerably thephysics case for axions and ALPs. Original hints to cooling anomalies were derived from the observations, starting from Kepleret. al. [227] in 1991, that the rate of period change ˙
P /P of G117 - B15A, a pulsating WD,was larger than predicted by the standard pulsation theory [228].After over 20 years of additional observations, the hint from G117 - B15A remains [205].Additionally, the WD variables R548 [206], PG 1351+489 [209], L 19-2 (113) [208] and L19-2 (192) [208] have all shown similar anomalies, with the observed rate of period changebeing always larger than expected. The current results are reported in Table 2. Since ˙
P /P is practically proportional to the cooling rate ˙
T /T , the results seem toindicate that these WDs are cooling substantially faster than expected. The unaccountedenergy loss could be due to a novel particle, efficiently produced in the dense core of a WDand freely escaping carrying energy away. Examples considered in the literature are axions The discrepancy has been calculated using the data for ˙
P /P in the quoted references and does not accountfor possible unknown systematics. In particular, the high significance of the discrepancy for G117 - B15A couldbe due to the hypothesis that the particular oscillating mode examined (with period about 215 s) is trapped inthe envelope (see, e.g., [205]), an assumption which could be incorrect [211]. Relaxing this hypothesis wouldsignificantly reduce the discrepancy [205] (see also discussion in [203]). – 39 –D class P [s] ˙ P obs [s/s] ˙ P th [s/s] discrepancyG117 - B15A DA 215 (4 . ± . × − (1 . ± . × − σ R548 DA 213 (3 . ± . × − (1 . ± . × − σ PG 1351+489 DB 489 (2 . ± . × − (0 . ± . × − . σ L 19-2 (113) DA 113 (3 . ± . × − (1 . ± . × − . σ L 19-2 (192) DA 192 (3 . ± . × − (2 . ± . × − . σ Table 2 . Results for ˙ P measured and expected in WD variables. produced through electron bremsstrahlung [205, 208] (see sec. 6.2) and neutrinos with ananomalous large magnetic moment produced through plasmon decay [207]. An anomalous behaviour was also observed in the WD luminosity function (WDLF), whichdescribes the number distribution of WDs in brightness intervals. The particular shape of thisdistribution depends on the lifetime of WDs in a specific luminosity bin and, consequently, onthe efficiency of the cooling mechanisms. A number of studies (see, e.g., [211]) showed thatadditional cooling provided by axions/ALPs coupled to electrons could improve considerablythe fit, while even a large neutrino magnetic moment would not have any substantial effecton the WDLF [210]. This result can be attributed to the too steep temperature dependenceof the plasmon decay into neutrinos [203].A more recent study of the hot part of the WDLF [229] did not confirm this anomalousbehaviour. However, the hotter section of the WDLF has much larger observational errorsand the ALP production would be almost completely hidden by standard neutrino coolingin the hottest WDs. Whatever the case, a decisive improvement in our understanding ofthe WDLF is expected in the next decade or so. Observations from the Gaia satellite havealready increased the catalog of WDs by an order of magnitude with respect to SDSS, andLSST is expected to ultimately increase the census of the WDs to tens of millions [230].
Further hints to anomalous energy loss emerged in the recent analyses of Red Giant Branch(RGB) stars in the globular cluster M5 [212, 213] and, with less significance, in the globularcluster ω -Centauri [214].The red giant is the evolutionary stage of low mass stars that follows the main sequence.Stars in this stage have a He core and burn H in a shell. During their evolution in the RGB,stars become brighter and brighter until they reach a tip in the color magnitude diagram atthe time of the He-flash, corresponding to the He ignition in their core. After the He-flash,the luminosity decreases and the stars move into the Horizontal Branch (HB) stage.Additional cooling of the core during the RGB stage would delay the He ignition allowingthe star to become brighter before moving into the HB phase. The tip of the RGB is,therefore, a measure of the cooling efficiency during the RGB evolutionary stage.The studies in [212–214] showed a brighter than expected tip of the RGB, indicatinga somewhat over-efficient cooling during the evolutionary phase preceding the helium flash.In all cases, the significance is fairly low, ∼ . σ in the M5 analysis and less than 1 σ in ω -Centauri. In spite of the low significance of these results, however, it is remarkable thatRGB observations seem to confirm the need for additional cooling in agreement with the– 40 –ompletely unrelated results from WDs. These results can be improved using multi-bandphotometry of multiple globular clusters [231]. A considerable reduction of the observationaluncertainties, particularly those related to the clusters distances, are expected from the dataof the Gaia satellite mission [232].Moreover, an independent analysis [215] showed a disagreement between the observedand expected R -parameter, R = N HB /N RGB , which compares the numbers of stars in theHB ( N HB ) and in the upper portion of the RGB ( N RGB ). Assuming Gaussian errors, thediscrepancy between the observed, R = 1 . ± .
03, and the expected, R = 1 . ± .
03, valuesis about 2 σ . The result indicates a surplus of RGB with respect to the numerical predictionand can be interpreted as an anomalous cooling with a different degree of efficiency in thetwo evolutionary stages. This anomaly is known as the R − parameter or HB hint.Assuming the result is due to new physics, the ideal candidate to explain the low numberof HB would be an ALP coupled to photons and produced through the Primakoff mechanismin the stellar core. This process is fairly inefficient in the high density environment of theRG core and could have the effect of accelerating only the following HB stage (hence, HBhint), reducing therefore the number of HB versus RGB stars and explaining the result for R [215, 216]. A cooling mechanism efficient during the RGB stage could, however, also explainthe discrepancy in the predicted and measured R − parameter. In particular, an ALP coupledto electrons and produced through electron bremsstrahlung or Compton could produce asimilar effect [203]. In general, we could expect a combination of the two mechanisms, asdiscussed in sec. 6.2. An additional deviation from the standard cooling theory was observed in core He-burningstars of intermediate mass ( M ∼ M (cid:12) ). The problem, in this case, is that numericalsimulations predict a larger number ratio of blue (hot) over red (cold) supergiants (B/R),with respect to what is actually observed [217, 218]. The predicted number would be lowered(alleviating or, perhaps, solving the B/R problem) in the hypothesis of an additional coolingchannel efficient in the stellar core but not in the H-burning shell [233], which could beprovided by axions coupled to photons and produced through the Primakoff process in a wayanalogous to what discussed for HB stars [219, 220, 234]. An exact prediction of the requiredadditional cooling is, however, presently unavailable. Finally, X-ray observations of the surface temperature of a neutron star in Cassiopeia A [235–237] showed a cooling rate considerably faster than expected. The effect seems to indicate theneed for an additional energy loss roughly equal to the standard one. This was interpretedin terms of an axion-neutron coupling [221] g an (cid:39) × − . (6.1)Although this result is compatible with well established limits from NS cooling [238, 239],new results based on different assumptions of the NS micro-physics have challenged this pic-ture [240, 241]. Moreover, the observed anomaly may also have origin in the phase transitionof the neutron condensate into a multicomponent state [242]. Given this controversy, and thegeneral difficulty in the modelling of NS, we decided not to include this result in our globalanalysis of the cooling anomalies presented here.– 41 – .2 Cooling anomalies, axions and IAXO As discussed above, axions/ALPs are favourite candidates, among the various new physics op-tions, to explain the cooling anomalies [203]. More recently, these have also been interpretedin terms of concrete QCD axions models [145] such as KSVZ, DFSZ, and Axi-Majoron (A/J).Here we present the regions of the ALP and axion parameter space hinted by the anomalousobservations and the IAXO potential to probe these areas.We consider first the case, often discussed in the literature, of an ALP interacting onlywith photons. In this case, the most relevant axion production mechanism is the Primakoffprocess, γ + Ze → Ze + a , (6.2)which consists in the conversion of a photon into an ALP in the electric field of nuclei andelectrons in the stellar core. This process depends strongly on the environment temperatureand is suppressed at high density (e.g., those characterizing the core of WDs and RGBstars) by the plasma frequency and degeneracy effects (see, e.g., [243]). Thus, if we ignorethe interaction with electrons, axions cannot provide a solution for the excessive coolingobserved in WDs and RGB. Instead, their main effect on the evolution of low mass starswould be to accelerate the HB phase while leaving essentially unchanged the RGB stage, andtherefore to reduce the expected R-parameter.This property has been used to constrain the axion-photon coupling [215, 216, 243].The current 2 σ bound, g aγ < . × − GeV − [215, 216], is shown in Fig. 10 (left) by thesolid black line labeled “HB”.The red-hashed region in the left panel of Fig. 10 is the hint from the R − parameter (HB-hint) [215, 216, 244] at 1 σ confidence level: g aγ = (0 . ± . × − GeV − . Notice that thebound and the hinted regions are obtained assuming a vanishing ALP-electron interaction.The more general case will be discussed below. As evident from the figure, IAXO is expectedto have sufficient sensitivity to detect ALPs in this region, for masses below m a ∼ . σ intervals on the axion/ALPs coupling with electrons and photons derived fromthe observations of individual stellar systems are shown in Tab. 3. The most relevant axionproduction mechanism at high density is the bremsstrahlung process, e + Ze → Ze + e + a , (6.3)which induces an additional energy loss rate proportional to T . As shown in [203], this tem-perature dependence is optimal to fit the WDLF and provides a reasonably good explanationfor the observed excess cooling in DA and DB WD variables, whose internal temperaturesdiffer by a factor of a few. The combined analysis of all the observed WD variables gives afairly good fit, χ / d.o.f= 1 .
1, for g ae = 2 . × − and favours the axion solution at 2 σ . The confidence intervals for the WD variables shown in Tab. 3 have been derived from a likelihood analysisof the data in [206–208] and are not given in the original references. The hint from the WDLF is derivedfrom the data in [211]. The RGB hint refers to the M5 data [213] and is calculated from a likelihood analysiscombining the observational and computational 1 σ errors in quadrature. It has a slightly smaller error thanthe one reported in [203], which was based on a conservative interpretation of Fig. 2 in [213]. – 42 – a (eV)10 − − − − | g a γ | ( G e V − ) − − − Q C D a x i o n m o d e l s CASTBabyIAXO
IAXO
IAXO+ K S V Z HB HB hint D S F Z II & A / J E / N = / D S F Z I & A / J E / N = / m a ( eV )10 − − − − | g a e g a γ | / ( G e V − / ) − − − − CAST
IAXO
IAXO+
Figure 10 . Summary of astrophysical hints in the ( g aγ , m a ) and ( √ g aγ g ae , m a ) planes compared withthe expected sensitivity of IAXO. The 2 σ regions corresponding to five explicit QCD axion models(DFSZ I, DFSZ II and A/J with E / N = 2/3, 5/3 and 8/3) that can account for the WD/RGB/Ranomalies following the work in [145] are shown. Only the fraction of the regions for which solaraxion production through the g aγ or g ae couplings is dominant, is shown on the left and right plotsrespectively. The unitarity constraints have also been imposed. Left: The red-hashed region is thehint from HB stars [215, 216, 244], obtained in the assumption of interaction with photons only. Theyellow region shows the customary band for QCD axion models, see text for details. The mentionedhinted regions are shown as red segments. Note that the DFSZ lines have been displaced 5% upwardsfor visibility as they overlap with A/J models. The A/J model with E / N = 5/3 does not appear herebecause solar production via g ae is dominant for all this region. Right: The IAXO sensitivity assumesonly solar production through the electron coupling. The 2 σ hinted regions appear as diamond-shaped regions in yellow (A/J E / N = 5/3), red (A/J E / N = 8/3), orange (A/J E / N = 2/3), lightgreen (DFSZ II) and dark green (DFSZ I). The wide yellow band encompasses all possible DFSZmodels within unitarity constraints. As shown, the combined prospects of IAXO via the g aγ and g aγ g ae channels will probe most of the hinted region, with the only exception of the A/J E / N = 5/3model, due to the extremely suppressed g aγ coupling. observable hint (1 σ ) observable hint (1 σ ) WD G117 - B15A α = 1 . ± .
47 WD R548 α = 1 . ± . α = 0 . ± .
38 WD L19-2 (113 s mode) α = 2 . ± . α = 0 . ± . α = 0 . +0 . − . luminosity of RGB tip α = 0 . +0 . − . R-parameter g = 0 . ± . Table 3 . Hints at 1 σ from stellar cooling anomalies (from ref. [222]). Here, α = ( g ae × ) / π and g = g aγ × GeV. The hint on the R-parameter shown here assumes no ALP-electron interaction.
Moreover, the peculiar temperature dependence of the axion bremsstrahlung rate allowsto account for the excessive cooling observed in RGB stars [213] (which have a considerablylarger internal temperature than WDs) with a comparable axion-electron coupling. Thecombination of the hints from the WD pulsation, the WDLF and RGB stars gives g ae =– 43 – - B15AR548L19 - ( ) L19 - ( ) PG 1351 + - tip - α σ σ σ α χ / d . o .f. Figure 11 . Hinted regions for the axion-electron coupling, with α = ( g ae × ) / π . The segmentsin the left side plot show the 1 σ intervals. On the right we show the fit of all the hints combined,excluding the G117-B15A hint, as discussed in [203]. . +0 . − . × − with χ / d.o.f= 1 . σ , see Fig. 11.A more recent independent analysis of the RGB luminosity tip in ω -Centauri [214] con-firms the hint to additional cooling, though this study has so far been carried out consideringonly the neutrino magnetic moment as source of exotic cooling.Finally, if we add also the analysis from the R − parameter anomaly, we find the re-sults presented in Fig. 12, where the light brown region shows the 1 σ hinted area. Noticethat the R -parameter depends, in general, on both the axion-electron and axion-photon cou-plings [203]. This explains the slight left bending shape of the R − parameter hinted region.In particular, the quantitative analysis with the data at hand shows that it is possible toexplain all the observed cooling hints, including the R -parameter anomaly, even neglectingthe axion-photon coupling, though there is a preference for non-vanishing couplings withboth electrons and photons. The best fit values, g ae = 1 . × − and g aγ = 0 . × − GeV − , indicated in the figure with a red dot, are well within reach of IAXO though the 1 σ hinted region extends, in this case, to lower axion-photon couplings.The sensitivity of IAXO to axions coupled to both electrons and photons is shown inFig. 10 (right). Here it is assumed that production of solar axions proceeds through theaxion-electron coupling and detection through the photon coupling. If production via thePrimakoff process dominates the solar flux the relevant graphical reference is Fig. 10 (left).We can already note that the best fit value g ae = 1 . × − with g aγ = 0 . × − GeV − would be detectable by IAXO up to ALP masses of 0 . σ hinted region, the coupling to photons could be zero and this would preventany possibility of detection by IAXO.The situation is much better defined for QCD axion models, where we know that onlyfine-tuning can provide a small coupling to photons. Indeed, it is more generic to have aphoton coupling than a sizeable electron coupling. Moreover, axions couple generically toprotons and neutrons and this implies a strong constraint from the measured duration ofthe neutrino pulse of SN1987A (see, e.g., [245–247]). Reference [145] studied three classesof axion models, KSVZ, DFSZ and A/J, which differ on the origin of the axion coupling toelectrons. In KSVZ, this arises from a loop involving the photon coupling and turns out to bevery small. The interplay of the direct constraints on the photon coupling and the SN1987Abound does not allow a KSVZ solution to the cooling anomalies [145]. The situation is verydifferent in DFSZ I and DFSZ II models, which are two Higgs doublet models extended– 44 – ae − − − − − − | g a γ | ( G e V − ) − − − − − CAST
IAXO
BabyIAXOIAXO+
Figure 12 . Combined analysis of the hints on g ae and g aγ . The light brown shaded region is the1 σ hint, with the best fit value g ae = 1 . × − , g aγ = 0 . × − GeV − . Also shown is thesensitivity of IAXO, assuming m a < .
01 eV] for which the axion photon oscillation probability isindependent of the mass. with an additional SM singlet scalar that allows a PQ symmetry spontaneously broken athigh f a (see for instance [77]). DFSZ II has a larger C aγ = E/ N − .
92 = 2 / − .
92 andwill be thus easier to discover with IAXO than DFSZ I, for which E / N = 8 /
3, implying alarger cancellation between the model dependent and model independent contributions tothe photon coupling. The regions where these models can account for the WB/RGB/HBhints are depicted in Fig. 10 (right) and Fig. 13 (top panels). In Fig. 10, we are assumingproduction solely through the axion-electron coupling. This condition is relaxed in Fig. 13.In Axi-Majoron models, the electron coupling arises through a loop involving right-handed neutrinos and can be large if the involved Yukawa couplings are large. The threedifferent Axi-Majoron models considered consists in the addition of a new heavy quark anda SM scalar singlet and can be considered as variations of KSVZ where the PQ scalar fieldgives mass to RH neutrinos and breaks lepton number spontaneously. They differ in the SMcharges of the heavy quark that give different couplings to photons. The hinted regions forthese models are shown in the right panel of Fig. 10 and in Fig. 13 (bottom panels). Justlike in the case of DFSZ axions, in Fig. 10 we are assuming production solely through theaxion-electron coupling while we relaxed this conditions in Fig. 13.As shown in [145], DFSZ and Axi-Mjoron models can well explain the stellar hints evenwhen the constraints from SN 1987A are accounted for. However, some tension does existin these models between the hinted values for the axion coupling with electrons and photons(from WD and globular cluster stars), and the bound on the axion-nucleus coupling extractedfrom the observed neutrino signal of SN 1987A. Recent astrophobic models [70] relax furtherthis tension and promise even better fits. An quantitative analysis of these models in thecontest of the cooling anomalies is in preparation.In general, IAXO will be capable to find the hinted QCD axions in a sizable part ofthe parameter space, although it is with the upgraded IAXO configuration that most of the– 45 – σ σ σ σ DFSZ I; WD, RGB, HB σ σ σ σ DFSZ I; WD, RGB, HB
IAXOIAXO + ARIADNE �� - � �� - � �� - � ��� - � ����� � �� �� �� � �� � �� � σ σ σ σ DFSZ I; WD, RGB, HB σ σ σ σ DFSZ I; WD, RGB, HB
IAXOIAXO + ARIADNE �� - � �� - � �� - � ��� - � ����� � �� �� �� � �� � �� � σ σ σ σ WD, RGB, HB A / J ( R Q =( - / )) σ σ σ σ WD, RGB, HB A / J ( R Q =( - / )) IAXOIAXO + �� - � �� - � �� - � ��� - � ����� � �� �� �� � �� � �� � σ σ σ σ WD, RGB, HB A / J ( R Q =( + / )) σ σ σ σ WD, RGB, HB A / J ( R Q =( + / )) IAXOIAXO + �� - � �� - � �� - � ��� �� �� � �� � �� � σ σ σ σ WD, RGB, HB A / J ( R Q =( + / )) σ σ σ σ WD, RGB, HB A / J ( R Q =( + / )) IAXOIAXO + �� - � �� - � �� - � ��� �� �� � �� � �� � �� - � ����� � Figure 13 . Isocontours of the χ fit to the WD, RGB and HB anomalies in different axion modelsfrom [145]. In DFSZ tan β is the ratio of the 2 Higgs doublet VEVs, which is constrained by unitarityin the Yukawa couplings to be within the dashed lines. The parameter entering in the axion-electroncoupling in the A/J models is the trace over RH Yukawa shown as | tr κ − κ ee | and it also has an upperbound from unitarity. The green regions are the sensitivities of IAXO and ARIADNE. See [145] fordetails. The lower axis gives the axion mass in units of eV. hinted parameter space will be covered. The A/J model R Q = (3 , , +1 /
6) is an exceptionbut it is already quite a tuned solution very close to the unitarity constraint [145]. In theregion accessible by IAXO the solar axion emission happens mostly through the axion-electroncoupling. This is particularly true for the DFSZ I model. In this case, the production rateinduced by the axion-photon coupling would dominate for low values of tan β , in a regionalmost entirely excluded by the unitarity constraints.The regions hinted by the cooling anomalies, shown in Figs. 10 and 12, are phenomeno-logically quite interesting. They are largely accessible to IAXO and partially to ALPS II [35].At high mass there is room for a QCD axion solution, in a range of parameters still partiallyaccessible to IAXO, while at lower masses they point to ALPs which could provide the re-quired CDM [43]. Finally, at slightly lower masses there is an overlap with the parametersrequired to explain the transparency hints [248]. In all cases, IAXO shows a high potentialto explore the relevant parameter space. – 46 – Axion-like particles in the propagation of photons over astronomicaldistances
IAXO will be capable to probe the ALP parameter space which is currently only accessibleto astrophysical observations. Astrophysical magnetic fields can range from values below10 − G in intergalactic space to (cid:38) G on the surface of magnetars (highly magnetizedneutron stars) on a large range of spatial scales. As the photon-ALP conversion probabilityscales with the product of the field strength B and the spatial extend of the field L , in someastrophysical environments a large conversion probability could be possible [249–251]. For ahomogeneous magnetic field, the conversion probability becomes maximal and independentof energy for energies E crit (cid:46) E (cid:46) E max , with [e.g. 249, 252] (neglecting dispersion due to,e.g., the CMB [253]) E crit ≈ | m a − ω | g aγ B ∼ . | m − . × − n cm − | g − B − µ G , (7.1) E max ≈ π α B g aγ B ∼ . × GeV g B − µ G . (7.2)In the above equations, we have introduced the notation B X = B/X , m X = m a /X , and g X = g aγ × X / GeV − . The plasma frequency is connected to the electron density of themedium through ω pl ∼ . √ n cm − neV. Above E max , the oscillations are damped due tothe QED vacuum polarisation. The chosen units in the above equations already indicate thatfor magnetic fields with a strength of µ G and ALP masses around neV, a strong mixing canbe expected at gamma-ray energies. Such magnetic fields are commonly observed in the intra-cluster medium of galaxy clusters, in galaxies themselves, and in the lobes of jets launchedby active galactic nuclei (AGNs) [e.g. 254, 255]. For lower ALP masses m a (cid:46) − eV, E crit shifts to X-ray energies. For X-rays or gamma rays produced in AGN, Tab. 4 displaysthe typical strength of the magnetic field B , its coherence length L , and the photon-ALPoscillation length λ a of magnetic fields traversed by the photons.Below, we summarize current hints, constraints, and future sensitivities of gamma-rayand X-ray observations. In comparison to future IAXO observations, the X-ray and gamma-ray measurements suffer from the unknown exact conditions of the involved magnetic fieldsand photon fluxes. In this respect, IAXO offers the unique opportunity to test any hintsfound in astrophysical observations in a more controlled setting. For gamma rays, one expects two possible observables as a result of photon-ALP mixing. Onthe one hand, the conversions should lead to oscillatory features around E crit and E max assome photons will oscillate into ALPs thereby reducing the photon flux. The shape of theseoscillations will depend on the morphology of the magnetic field, the plasma frequency, theALP parameters, and the gamma-ray energy. Since astrophysical magnetic fields are oftenturbulent in nature, the spectral features are expected to show a complicated dependence onenergy.On the other hand, photon-ALP oscillations could lead to a boost in the photon flux dueto an astrophysical version of the “light shining through a wall” experiment, in which photonsconvert to axions or ALPs and back on either side of an opaque wall. For gamma rays emittedby AGNs, the opacity is caused by the interaction of gamma rays with lower-energy photonsfrom infrared (IR) to ultraviolet (UV), which produces electron-positron pairs [260, 261]. In– 47 –stronomical object B L λ a AGN jet 0 . − O (0 . − . ∗ Radio lobes 10 µ G 10 kpc O (10 kpc)Spiral galaxies 7 µ G 10 kpc O (10 kpc)Starburst galaxies 50 µ G 10 kpc O (1 kpc)Elliptical galaxies 5 µ G 150 pc O (10 kpc)Cluster 5 µ G 10 −
100 kpc O (10 kpc)IGM < . . −
10 Mpc O (50 Mpc) † Milky Way 5 µ G 10 kpc O (10 kpc) Table 4 . Various astronomical objects with typical values of average magnetic field B , domain length L and corresponding photon-ALP oscillation length λ a traversed by a photon beam originating fromAGNs. IGM stands for intergalactic medium. The magnetic field in the AGN jet does not possessa domain-like structure [256]. The values of λ a are calculated in the strong mixing regime wherethe plasma and ALP mass effects at low energies, the QED vacuum polarization and CMB photondispersion effects at high energies can be neglected. In the case where λ a (cid:46) L the simple sharp edgesmodel model to describe domain-like magnetic fields gives unphysical results because the photon/ALPbeam becomes sensitive to the discontinuities of the magnetic field components: in this case, moresophisticated models must be used [e.g. 253, 257, 258]. ∗ In the case of the jet QED effects are quiteimportant and for energies E (cid:38) O (100 GeV) they strongly reduce the reported value of λ a . ∗∗ Asimilar fact happens in the extragalactic space because of the ALP mass term for E (cid:46) O (100 GeV)and because of the CMB photon dispersion on the CMB for E (cid:38) O (5 TeV) [259]. intergalactic space, these lower-energy photons originate from the extragalactic backgroundlight (EBL), the background radiation field which encompasses the stellar emission integratedover the age of the Universe, and the emission absorbed and re-emitted by dust [see 262, 263,for reviews]. Due to strong foreground emission in the solar system in the EBL wavelengthrange, the EBL is extremely difficult to measure directly [264]. Within the AGN jet, theradiation fields from the dusty torus, the accretion disk, or optical emission from ionizedclouds (the so-called broad line region, BLR) can cause the attenuation [e.g. 265]. Thepresence of the BLR is usually associated with a sub-class of AGNs, so called flat-spectrumradio quasars (FSRQs). Basically, because of photon-ALP oscillations, the photon acquiresa “split personality”: sometimes it travels as an ordinary photon and gets absorbed by theEBL, but sometimes it travels as an ALP which does not interact with EBL photons. As aconsequence, the effective optical depth, τ eff is smaller than the optical depth as evaluated byconventional physics. Since the photon survival probability is given by P ALP γ ← γ = exp[ − τ eff ],even a small decrease of the effective optical depth can give rise to a large enhancement inphoton flux.The exponential attenuation of the initial gamma-ray flux is described with the opticaldepth τ , which increases monotonically with the primary gamma-ray energy, the sourcedistance (in the case of absorption on the EBL), and the photon density of the radiationfield. ALPs are not absorbed during their propagation and if a sizeable amount of photonsconverts into ALPs, which then reconvert back to photons close to Earth, a boost of theexpected gamma-ray flux is expected [e.g. 250, 259, 266, 267].– 48 – .2 Hints for anomalies in the gamma-ray opacity? Several authors have found evidence that state-of-the-art EBL models over-predict the at-tenuation of gamma rays using published data points of AGN spectra obtained with imagingair Cherenkov telescopes (IACTs), which measure gamma rays above energies of ∼
50 GeV.Such an over-prediction would manifest itself through a hardening of the AGN spectra.For example, if the observed spectrum at low energies can be described with a power law, dN/dE ∝ E − Γ low a spectral hardening would mean that at high energies dN/dE ∝ E − Γ high with Γ high < Γ low . Such a behavior is in general not expected from standard gamma-rayemission scenarios (although specific models can produce such spectra [e.g. 268]), especiallyif the hardening ∆Γ = Γ low − Γ high correlates with increasing optical depth for several sources.Such a correlation has indeed been found and ALPs have been proposed as a possible expla-nation [250, 269–276]. An over-predicted EBL attenuation should also lead to a correlationbetween fit residuals and the optical depth when smooth concave, i.e. non-hardening, func-tions are assumed for the emitted AGN spectra. A 4 σ indication for this effect has beenfound [277, 278], and ALP parameters reducing this tension were derived [279] (see the “T-Hint” labelled region in Fig. 14). Furthermore, using recent EBL measurements with theCIBER experiment, which suggest a larger attenuation than current EBL models, the au-thors of Ref. [280] found that ALPs can improve the fits of IACT spectra when again concaveintrinsic spectra are assumed (region on top of the “T-Hint” region in Fig. 14). Instead ofALPs, these evidences have also been interpreted as evidence for particle cascades initiatedby ultra-high energy cosmic rays [281] or a correlation between AGN lines of sight with cosmicvoids [282].However, recent analyses could not confirm the above correlations for a spectral hard-ening. Extending the IACT data sample of Ref. [279], the authors of Ref. [283] did not finda correlation of fit residuals with the optical depth. Furthermore, when including system-atic uncertainties such as the IACT energy resolution, no spectral hardening between IACTspectra and spectra measured at lower gamma-ray energies with the Large Area Telescope(LAT) on board the Fermi satellite could be found [284]. Using
Fermi -LAT data alone, nospectral hardening as a function of redshift (or equivalently the optical depth) was found ina recent analysis [285].It has also been hypothesized that photon-ALP conversions could be responsible forIACT observations of FSRQs above 100 GeV. If gamma rays are produced close to the centralsuper massive black hole, their flux should be severely attenuated due to the interactionwith the radiation fields within the jet as mentioned above. Yet, a number of FSRQs hasindeed been observed [e.g. 286–288] and it has been shown that the inclusion of ALPs canreproduce the observed spectra for ALP parameters g ∼ m neV (cid:46)
10 [267]. One hasto keep in mind, though, that astrophysical mechanisms could also produce the gamma-rayemission beyond the BLR and thus circumvent the pair production and the attenuation [e.g.289, 290]. But it was the dissatisfaction of the ad hoc nature of these astrophysical attemptedexplanations that led to the above ALP-based proposal. In addition for BL Lacs (a sub-classof AGNs), by combining all the magnetic environments crossed by the photon/ALP beam –namely BL Lac jet, host galaxy, extragalactic space and Milky Way – it is possible to inferimportant predictions for BL Lacs spectra which present peculiar observable features inducedby photon-ALP oscillations: (i) the oscillatory behavior of the energy spectrum (ii) photonexcess above 20 TeV [291]. See also http://tevcat.uchicago.edu/ . – 49 –s shown in Fig. 14, IAXO is sensitive to the entire parameter space where ALPs havebeen proposed to alter the gamma-ray transparency. Therefore, IAXO measurements couldgive the definite answer on this matter. While this part does not deal specifically with opacity, we mention this effect here as anotherpossible hint for a light ALP ( ∼
10 neV) with the same coupling as what is discussed inthis section (10 − GeV − ). In the strong mixing regime, it is easy to show that the averageluminosity of a photon beam is 2/3 that without mixing [292] . It means that the luminosityof X-ray sources would be actually higher than what we observe. As all source would beconcerned, this effect based on average luminosities would not be observable. However,photon-ALP mixing would also affect higher moments of the luminosity distribution of thesources. By studying these higher moments, the authors of Ref. [293] showed an anomalycompatible with ALPs in the same parameter space region as mentioned in the previoussubsection. It has later been noted that this effect could be due to outliers in the used sourcecatalog [294] so it could be a selection-bias-induced fake signal. Searches for spectral distortions in gamma-ray spectra have already constrained the parame-ter space where ALPs could explain a reduced transparency. Observations with the H.E.S.S.telescopes of one blazar, i.e. an AGN with its jet closely aligned to the line of sight, haveled to the exclusions labelled “H.E.S.S.” in Fig. 14 under the assumption that the blazaris located in a Galaxy group that harbours a magnetic field of B = 1 µ G [295]. Furtherconstraints were derived using
Fermi -LAT observations of NGC 1275, the central AGN ofthe Perseus galaxy cluster [296]. This galaxy cluster could have a central magnetic field aslarge as 25 µ G [297]. The constraints are the strongest to date in the mass range between0 . (cid:46) m neV (cid:46)
20 (see Fig. 14). Similar analyses using X-ray observations of AGN in galaxyclusters have constrained lower mass ALPs (see below).For lower ALP masses, strong constraints where derived from the non-observation ofa gamma-ray burst from the core-collapse supernova (SN) SN1987A [298–300]. During thecore collapse, gamma rays in the core could convert to ALPs in the electrostatic fields of ionsand escape the explosion. If they convert back into gamma rays in the Galactic magneticfield, a gamma ray burst lasting tens of seconds could be observed in temporal co-incidencewith the SN neutrino burst. Interestingly, if a Galactic SN occurred in the field of view ofthe
Fermi
LAT, a wide range of photon-ALP couplings could be probed for m neV (cid:46)
100 (seethe red dashed line in Fig 14). These prospects for a detection rely on a simplified pictureof the proton-neutron star nuclear medium [300] and encourage future refinements of thetheoretically expected ALP flux from such objects [259, 301].The next generation of gamma-ray observatories will allow deeper studies of ALP-induced effects and will have enough sensitivity to probe the whole parameter space relevantfor the gamma ray transparency hints. In particular, analyses of Cherenkov Telescope Array(CTA) data will consist of searches for the spectral irregularities [see 302, for preliminaryresults], the reduced opacity [303], or spatial correlations between blazar spectra with themagnetic field of the Milky Way [304]. A similar correlation study could also be conductedwith future gamma-ray observations of the HAWC or LHAASO observatories [305]. However,CTA will be fully operational within a decade, and the necessary accumulation of data couldtake years. – 50 –n the other hand, as shown on Fig. 14, IAXO will be sensitive to the relevant massesand coupling. In conclusion, light ALPs can have a strong impact on gamma-ray astronomyand IAXO will cover the whole relevant parameter space. m a (eV)10 − − − − − − − − − | g a γ | ( G e V − ) − − − − − CASTBabyIAXO
IAXO
IAXO+
ALPSII HESST Hint F e r m i S N p r o s p ec t s Fermi NG1275Chandra Hydra A SN1987A
M87 NGC1275
Soft x-ray excess& 3.5 keV line HB hint
Figure 14 . The IAXO potential in the low ALP mass part of the g aγ vs. m a plane. The hashedregion indicates the R-parameter hint, discussed in section 6.1.3, in the case of ALPs interacting onlywith photons. In this case, this is known also as the HB-hint [215, 216, 222]. The region indicatedwith ”T-hint” is the transparency region, discussed in the section 7. ALPs with parameters in thisregion have been invoked to address the unexpected transparency of the Universe to very high energyphotons [279, 280] and some anomalous redshift-dependence of AGN gamma-ray spectra [306]. Noticethat the lower mass section of the hinted region has been excluded by the non-observation of gammarays from SN 1987A [300] and, more recently, by the search for spectral irregularities in the gammaray spectrum of NGC 1275 [296]. The region enclosed within the dashed red line, labelled Fermi SNprospects, shows the Fermi LAT potential to probe the ALP parameter space in case of a new nearby(galactic) SN explosion [307]. Finally, the figure shows the regions excluded by the analysis of thespectral distortions of X-ray point sources in galaxy clusters: the Chandra observations of the AGNin Hydra A [308], the Perseus cluster NGC1275 [309] (indicated in the figure as NGC1275 and to bedistinguished from the region labelled Fermi NG1275), and the M87 AGN of the Virgo cluster [310].Refer to the text for more details. – 51 – .4 Conversion between X-ray photons and ALPs Let us now describe further aspects of ALP physics that rely on the interconversion of ALPsand X-ray photons in the magnetic field of galaxy clusters. • Spectral distortions of X-ray point sources in galaxy clusters
Just as with gamma-ray observations, it is possible to search for ALPs by looking foroscillatory features in the spectra of arriving X-rays from AGN or quasars in or behindgalaxy clusters – and conversely, the absence of such features can be used to placebounds on ALP parameter space.The main disadvantage of this method is that the actual magnetic field along the lineof sight is unknown. For any one source, it is then never possible to exclude thepossibility that the magnetic field configuration along the line of sight is particularlyunfavourable for ALP-photon conversion. This motivates considering sufficiently broadenergy ranges so that multiple features are expected, and observing multiple X-raysources within distinct astrophysical environments.In X-rays, this method has been applied for the AGN in Hydra A [308], the centralAGN of the Perseus cluster NGC1275 [309] (see above for gamma rays), the centralM87 AGN of the Virgo cluster [310], and a variety of weaker quasars and AGNs in andbehind clusters [311]. Of these sources, the two bright local AGNs NGC1275 and M87are the most constraining, as both are bright AGNs with deep exposures at the heartof large cool-core galaxy clusters (which tend to have the highest magnetic fields). Forreasonable and observationally supported values for the magnetic field structure withinthe Perseus and Virgo cluster, the absence of large spectral modulations was used in[309] and [310] to constrain g aγ (cid:46) . × − GeV − for ALP masses m a (cid:46) − eV.These bounds depend on the magnetic field model for the cluster. For magnetic fieldsweaker or stronger than assumed, the constraint scales inversely with the magneticfield. • Spectral distortions of the continuum thermal bremsstrahlung emission ofgalaxy clusters
The advantage of bright point sources is that they are sensitive to the magnetic fieldalong a single line of sight, and so avoid effects of destructive interference when av-eraging over many different sightlines. However bright point sources are also ratherrare.Another approach to searching for ALPs is to use instead the continuum emission fromgalaxy clusters. This arises as thermal bremsstrahlung from the intracluster medium,the hot ( T ∼ − g aγ ∼ GeV, this thermalemission will experience significant spectral modulation along a single line of sight (asfor point sources). Averaged over a large region of the cluster, the modulations fromindividual sightlines will average out. However, in the presence of significant ALP-photon conversion significant spectral distortions will be present on small scales [312].– 52 –
The 3.5 keV Line
One of the most interesting recent results in particle astrophysics has been the obser-vation of an unexplained line at E ∼ .
55 keV [313, 314]. The line was found originallyin observations of stacked samples of galaxy clusters and is not at an energy that cor-responds to a known atomic line. There has been considerable interest around thepossibility that this could arise from dark matter [315–323], while also various possibleastrophysical explanations have been proposed.If the line arises from dark matter, one difficulty with ‘standard’ interpretations (suchas a sterile neutrino) is that the line is much stronger in clusters than in galaxies, andin particular is much stronger at the centre of the Perseus cluster than at any otherlocation. This implies that the line is not sensitive only to the dark matter content,but also to some aspect of the astrophysical environment. One way this can arise isin models where the dark matter decays originally to a relativistic ALP with energy E = 3 . − GeV − (cid:46) g aγ (cid:46) − GeV − for ALP masses m a < − eV. Interestingly, IAXO will be able to testa relatively large region of this parameter space.The morphology of the 3.5 keV signal can be explained by ALP-photon conversion.The stronger signal in clusters would then arise from the larger and more extendedmagnetic fields present in clusters compared to galaxies. As Perseus is a close cool-corecluster, observations of Perseus only cover the central region with a large magnetic field(as B ∝ n e ( r ) , the magnetic field is significantly enhanced in the central high- n e coolcore). The extremely strong signal in the centre of Perseus would then be a consequenceof the efficiency of ALP-photon conversion in strong magnetic fields.If this line is found to arise from new physics, ALPs then offer a way to reproduce theunusual morphology. The International Axion Observatory (IAXO) is a next generation axion helioscope thataims at a substantial step forward, of more than one order of magnitude, in sensitivity to g aγ with respect to current best limits. The baseline layout of the experiment is based on theenhanced axion helioscope studied in [325] and sketched in Fig. 15. In this configuration theentire cross sectional area of the magnet is equipped with X-ray focusing optics to increasethe signal-to-noise ratio. When the magnet is pointing to the Sun, solar axions are convertedinto photons, that are focused and detected by low background X-ray detectors placed atthe focal point of the telescopes. In this way a larger magnet aperture A translates directlyinto the figure of merit of the experiment, as a larger signal is expected while the detectorbackground remains low. This opens the way for new large-volume magnet configurations,like the ones of superconducting detector magnets typically developed for high energy physics.A useful figure of merit (FOM) was introduced in [325] to easily gauge the relativeimportance of the various experimental parameters affecting the sensitivity of a helioscope: f ≡ f M f DO f T (8.1)– 53 – AGNET COILMAGNET COIL B field A L
Solar axion flux γ X-ray detectorsShielding
X-ray op!cs
Movable pla"orm
Figure 15 . Conceptual arrangement of an enhanced axion helioscope with X-ray focalization.Solar axions are converted into photons by the transverse magnetic field inside the bore of a powerfulmagnet. The resulting quasi-parallel beam of photons of cross sectional area A is concentrated byan appropriate X-ray optics onto a small spot area a in a low background detector. The envisageddesign for IAXO, shown in Fig. 16, includes eight such magnet bores, with their respective optics anddetectors. where we have factored the FOM to explicitly show the contributions from various experi-mental subsystems: magnet, detectors and optics, and tracking (effective exposure time ofthe experiment) f M = B L A f DO = (cid:15) d (cid:15) o √ b a f T = √ (cid:15) t t , (8.2)where B , L and A are the magnet field, length and cross sectional area, respectively. Theefficiency (cid:15) = (cid:15) d (cid:15) o (cid:15) t , being (cid:15) d the detectors’ efficiency, (cid:15) o the optics throughput or focusingefficiency (it is assumed that the optics covers the entire area A ), and (cid:15) t the data-takingefficiency, i. e. the fraction of time the magnet tracks the Sun (a parameter that dependson the extent of the platform movements). Finally, b is the normalized (in area and time)background of the detector, a the total focusing spot area and t the duration of the datataking campaign. The expressions in (8.2) assume some simplifications, like that B is constantin all the volume of the magnet, or that b and (cid:15) are constant throughout the energy range ofinterest. Generalizations of the figure of merit for arbitrary distributions of the parametersare straighforward, and the simplified versions quoted here are in any case useful to see themain dependencies.Following this metric, IAXO thus aims at a f more than a factor 10 larger that its pre-decessor CAST. The conceptual design report (CDR) of the experiment [6, 7] demonstratesthe technical feasibility of this step in sensitivity, and its main parameters will be reviewedbelow. Previous implementations of axion helioscopes have relied on existing equipment thatwere originally built for other experimental purposes. On the contrary, IAXO subsystems(magnet, optics and detectors) are entirely conceived and optimized for solar axion detec-tion. The design prescriptions have been to rely on state-of-the-art technologies scaled upwithin realistic limits, i.e. no R&D is needed to reach the stated experimental parameters.The baseline sensitivity projections studied below will refer to the experimental parame-ters anticipated in the CDR. More recently, the realization of an intermediate experimental– 54 – igure 16 . Schematic view of IAXO. Shown are the cryostat, eight telescopes+detector lines, theflexible lines guiding services into the magnet, cryogenics and powering services units, inclinationsystem and the rotating platform for horizontal movement. The dimensions of the system can beappreciated by a comparison to the human figure positioned by the rotating table [7]. stage, BabyIAXO, featuring a scaled-down prototype version of the magnet, optics and de-tectors, is being considered. This new activity brings the opportunity to explore potentialimprovements over the CDR figure of merit of the full infrastructure, potentially resulting inan upgraded sensitivity scenario for IAXO. The sensitivity projections shown below and inthe plots all throughout this paper refer to these three experimental scenarios: BabyIAXO,IAXO-baseline and IAXO-upgraded. The experimental parameters corresponding to each ofthem are listed in table 5, and the prescriptions to define them are discussed in the following.The central component of IAXO is therefore a large superconducting magnet. Contraryto previous helioscopes, IAXO’s magnet will follow a toroidal multibore configuration [326],to efficiently produce an intense magnetic field over a large volume. The baseline layout ofthe IAXO magnet is a 25 m long and 5.2 m diameter toroid assembled from 8 coils, andgenerating effectively 2.5 T average (5 T maximum) in 8 bores of 600 mm diameter. Thetoroid’s stored energy is 500 MJ. The design is inspired by the ATLAS barrel and end-captoroids [327, 328], the largest superconducting toroids built and presently in operation atCERN. The superconductor used is a NbTi/Cu based Rutherford cable co-extruded withAluminum, a successful technology common to most modern detector magnets. Figure 16shows the conceptual design of the overall infrastructure [7]. IAXO needs to track the Sunfor the longest possible period. For the rotation around the two axes to happen, the 250 tonsmagnet is supported at the centre of mass by a system also used for very large telescopes.The necessary magnet services for vacuum, helium supply, current and controls are rotatingalong with the magnet.Each of the eight magnet bores is equipped with X-ray telescopes that rely on the high X-ray reflectivity on multi-layer surfaces at grazing angles. By means of nesting, that is, placingconcentric co-focal X-ray mirrors inside one another, large surface of high-throughput optics– 55 –an be built. The IAXO collaboration envisions using optics similar to those used on NASA’sNuSTAR [329], an X-ray astrophysics satellite with two focusing telescopes that operate in the3 - 79 keV band. The NuSTAR’s optics, shown in Fig. 17, consists of thousands of thermally-formed glass substrates deposited with multilayer coatings to enhance the reflectivity above10 keV. For IAXO, the mirror arrangement and coatings are designed to match the solaraxion spectrum. The conceptual design of the IAXO telescopes [330] can be seen on the rightof Fig. 17. As proven in [6, 7, 330], this technology can equip the aperture area of IAXOmagnet bores with focusing efficiency of around 0.6 and focal spot areas of about 0.2 cm .At the focal plane in each of the optics, IAXO will have low-background X-ray de-tectors. The baseline technology for these detectors are small gaseous chambers read bypixelated planes of micro-mesh gas structure (Micromegas) [331] manufactured with the mi-crobulk technique [332]. These detectors have been successfully used and developed in CASTand other low background applications [333]. The latest CAST detectors have achieved back-ground levels of 10 − counts keV − cm − s − with prospects for improvement down to 10 − oreven 10 − counts keV − cm − s − [334]. These background levels are achieved by the use ofradiopure detector components, appropriate shielding, and offline discrimination-algorithmson the 3D event topology in the gas registered by the pixelised readout. A pathfinder systemcombining an X-ray optics of the same type as proposed for IAXO and a Micromegas detec-tor has been operated in CAST during 2014 and 2015 with the expected performance [335].Alternative or additional technologies are being considered to complement the capabilitiesof Micromegas detectors or to extent them in specific aspects. GridPix detectors, similarto Micromegas detectors but built on a small CMOS pixelized readout [336], enjoy very lowenergy threshold down to the tens of eV, and thus are of interest for the search of specificsolar axion production channels lying at lower energies, like the ones mediated by the axion-electron coupling. Silicon Drift Detectors (SDD) offer better energy resolution with flexibleand cost-effective implementations [337, 338]. Finally, bolometric detectors like Magnetic Figure 17 . Left: the NuSTAR X-ray telescope, with optics very similar to that proposed for IAXO.Right: conceptual design of the X-ray optics needed for IAXO. – 56 – igure 18 . Conceptual design of BabyIAXO.
Metallic Calorimeters (MMC) [339], or Transition Edge Sensors (TES) [340] , enjoying muchlower energy threshold and energy resolution, are also under consideration. An R&D ac-tivity is ongoing to assess the low-background capabilities of all this technologies, and theirsuitability as detectors for IAXO. Low energy threshold and resolution are also of interest toextract model parameters in case of a positive signal (see next section).The values considered for the main experimental parameters are listed in table 5 for threedifferent scenarios. The column labelled
IAXO baseline shows the set of values anticipatedin our CDR for IAXO, and are considered a realistic estimation within current state of theart. We refer to [7] for a careful justification of those values. However, we now envision, as afirst step, the realization of an intermediate stage, called BabyIAXO, featuring a scaled-downprototype magnet, as well as prototype optics and detectors, all representative of the finalsystems. BabyIAXO will serve as a testbed for all technologies in the full IAXO, but at thesame time will deliver competitive physics. The BabyIAXO magnet is conceived as a common-coil dipole magnet, with two bores placed in between the coils. The two superconducting coilsare 10-m long, but otherwise they enjoy quite similar engineering parameters than the onesproposed for the final IAXO toroid. The BabyIAXO magnet bores will have a diameter of70 cm and each one will be equipped with one full detection line, optics and detector, of similardimensions than the final IAXO systems. In the baseline configuration of the experiment,one of the BabyIAXO lines will host a newly built IAXO optics prototype similar to the onedescribed above (potentially extended to 70 cm diameter), while the second one is expectedto host an existing XMM spare optics [341]. A technical description of the BabyIAXO systemwill be object of another dedicated publication. Figure 18 shows a conceptual design of theBabyIAXO setup, whose expected experimental parameters are listed in the first column oftable 5.In addition, the experience with BabyIAXO is expected to test enhanced design choicesthat eventually lead to improved IAXO FOM values, especially f M , beyond the ones antic-ipated in our CDR. The column labelled IAXO upgraded represent this possible improvedscenario. Collectively they constitute a factor ∼
10 better f (of which a factor 4 better f M )than the baseline scenario. However, the extent to which those possible improvements mayget eventually realized is tentative, and this scenario must be considered as a desirable target– 57 –hose feasibility will be studied as part of the BabyIAXO stage.The IAXO sensitivity projections shown in Fig. 19 as well as in all plots throughout thispaper refer to the three scenarios of table 5. They have been computed by means of MonteCarlo simulation of the expected background counts in the optics spot area, computation ofthe likelihood function and subsequent derivation of the 95% upper limit on the g aγ assumingno detected signal. The calculation is repeated for a range of m a values in order to build fullsensitivity lines in the ( g aγ , m a ) − plane. For the purpose of this analysis, and following similarprescriptions as in [6], detector background and efficiency are assumed flat with energy downto arbitrarily low energies. The axion-photon conversion in the magnet is approximated tothe conversion in an homogeneous field, and the focusing effect is reduced to the equivalenteffect of enhancing the signal-to-noise ratio due to the fact of confining the signal counts intothe spot area. For all scenarios, an additional buffer gas data taking phase is considered,giving rise to the extended step-wise sensitivity line at high masses (0.01-0.25 eV). This datataking phase is composed by a number of overlapping gas density steps spanning the desiredmass range. While in previous projections [325] an equal exposure time is assigned to eachstep, resulting in a exclusion line more or less horizontal in g aγ , in this case the exposure timeis different for every step and adjusted to obtain a sensitivity down to the DFSZ g aγ (KSVZfor the BabyIAXO case) for every m a value. Of course different prescriptions to distributethe total exposure time among the steps are possible, depending on the motivation, e.g. togo for larger m a or lower g aγ . The total exposure of this second phase is the same as thevacuum phase ( t in Table 5). In the case of a positive signal, and depending on the axion parameters, IAXO will be ableto extract information on its mass m a and relative coupling with electrons and photons,potentially providing invaluable information on the underlying theoretical model. If theaxion mass is above around 0.02 eV, axion-photon oscillations destroy the coherence of theconversion along the magnet length. This coherence can be restored if the conversion takesplace in a buffer gas with density matching the axion mass, this being the rationale of the gasphase of both IAXO and BabyIAXO. If a positive detection happens during the gas scanning Parameter Units BabyIAXO IAXO baseline IAXO upgraded B T ∼ ∼ ∼ L m 10 20 22 A m f M T m ∼ ∼ ∼ b keV − cm − s − × − − − (cid:15) d (cid:15) o a cm × × × (cid:15) t t year 1.5 3 5 Table 5 . Indicative values of the relevant experimental parameters representative of BabyIAXO aswell as IAXO, both the baseline and upgraded scenarios, based on the considerations explained in thetext. – 58 –hase, the gas density provides the axion mass. But also in the vacuum phase and for lowermass values the onset of these spectral oscillations can be observed and used to determinethe axion mass. As studied in [79], provided the signal is measured with sufficient statistics, m a values as low as 3 × − eV could be determined by this method.Moreover, if the axion signal is composed by significant fractions of Primakoff and ABCsolar axions, the combined spectral fitting can provide independent estimations of g aγ and g ae [80]. This combined determination works in areas of parameter space particularly wellmotivated by the stellar cooling anomalies. To exploit these capabilities high-resolution andlow-threshold detectors are preferred, because part of the ABC solar axion spectrum liesat lower ( < Although the focus of this paper has been the physics potential of IAXO in its baseline con-figuration as a helioscope, the (Baby)IAXO magnet constitutes a remarkable infrastructureto implement additional setups to search for axions/ALPs in alternative ways. A most ap- m a (eV)10 − − − − − − − − − | g a γ | ( G e V − ) − − − − − − − − − − − A x i o n m o d e l s ALPS-II
CASTIAXO
BabyIAXO
OSQAR P V L A S Haloscopes
HEγ K S V Z T e l e s c o p e s HBSun H D M Figure 19 . Sensitivity prospects of BabyIAXO and IAXO (semitransparent regions) in the overallcontext of other experimental and observational bounds. We refer to [5] for details on the latter. – 59 –ealing option is to implement DM axion detectors that could exploit the particular featuresof the IAXO magnet. The basic idea of the axion-haloscope-technique proposed in [29] is tocombine a high- Q microwave cavity inside a magnetic field to trigger the conversion of axionsof the DM halo into photons. A straightforward implementation of this concept in one boreof the IAXO magnet gives competitive sensitivity [342] even using relatively conservativevalues for detection parameters, thanks to the large B V available, in the approximate massrange of 0.6 – 2 µ eV. Another recent concept proposes the use of a carefully placed pick-upcoil to sense the small oscillating magnetic field that could be produced by the DM axioninteracting in a large magnetic volume [343]. This technique is best suited for even lower m a and is better implemented in toroidal magnets [344, 345]. The large size and toroidalgeometry of IAXO suggest that it could host a very competitive version of this detectiontechnique.However, as can be seen from Fig.6, there is a strong motivation to explore the massrange for QCD axion DM well above 10 − eV. The difficulty in searching in the “high-mass”range can be understood from the fact that the figure of merit of scanning with haloscopesfor axion DM scales with the square of the cavity volume times the Q factor of the cavity.Most existing setups use solenoidal magnets and cylindrical cavities. The diameter of thecylinder sets the frequency scale of the resonance and thus the axion mass scale which theexperiment is sensitive to. Thus going to high mass means going to small diameters. Inaddition, the cavity quality factor Q typically decreases for smaller cavities.To tackle higher masses, different strategies are being developed in the community [5].For instance, to compensate the loss in V one can go to very strong magnetic fields and/or usesuperconducting cavities to keep Q large. Another avenue is to decouple V from the resonantfrequency and go for large V structures resonating at high frequency. Several strategies inthis direction are being explored. One of them, currently being tested in exploratory set-ups at the CAST experiment at CERN [346, 347], employ long rectangular cavities [348].The particular advantage is that the volume can be kept very large (long cavity), while theresonance frequency can be rather high, through the usage of relatively thin cavities. Giventhe size of the (Baby)IAXO magnet, it could host a multitude of rectangular cavities.The implementation of one those concepts in IAXO is under consideration, but is outof the scope of the present paper. It clearly deserves serious study and will be the object ofone or more future publications. Axions and more generic ALPs are recently receiving an increasing attention from the ex-perimental community. This is in part due to the lack of experimental confirmation of theWIMP dark matter paradigm in recent searches both at colliders and direct detection ex-periments. Axions are known to be attractive dark matter candidates, but despite continuedexperimental activity almost since their proposal more than 30 years ago, most of the allowedparameter space remains largely unexplored so far. Recently a plethora of ideas and newsmall-scale initiatives are being put forward [5], under the assumption that dark matter isentirely made of axions, and mostly addressed to the to low-mass range m a ∼ − − − eV.However, as described in section 3, axion (or ALP) dark matter can be also realized in othermass ranges, in particular at higher mass values, for which dark matter experiments areincreasingly difficult to realize. In addition, axions/ALPs may compose only a subdominantfraction of the galactic dark matter density, a possibility that may have a particular theoret-– 60 –cal interest [349]. In general, a signal in a dark matter experiment is proportional to g aγ ρ a ,being the local galactic axion density ρ a degenerate with g aγ . Experiments not relying on theaxion being the dark matter are needed to break such degeneracy. In any case, axions arestrongly motivated by theory, without relying on the dark matter question. They constituteour only compelling solution to the strong-CP problem in the Standard Model. As reviewedin section 2, axions and ALPs are also very generic low energy signatures of high-energycompletions of the SM in extra dimensions.Experiments that invoke both the production and detection of ALPs entirely in thelaboratory are the least model-dependent. Among these experiments, only the ALPS-IIproject will reach a sensitivity beyond current astrophysical and experimental bounds on g aγ , for ALP masses below m a < − eV. In this context, the search for solar axionsconstitutes a good compromise between model-independency (as the emission of axions bythe Sun is a robust prediction of any axion model), and intensity of the source axion flux.The IAXO project embodies the technical know-how accumulated in previous realizations ofthe axion helioscope concept, most in particular the CAST experiment at CERN, extendedto a much larger scale. While CAST has been the first axion helioscope reaching a sensitivityto g aγ comparable to astrophysical bounds, IAXO will largely advance well beyond them. Inparticular, and as a summary of the material reviewed in previous sections: • IAXO will cover a large fraction of unexplored ALP parameter. In particular, it willprobe a large range of QCD axion models in the mass range m a ∼ • Most of this region is not attainable by any other experimental technique, stressing thecomplementarity of IAXO in the wider axion experimental landscape. • IAXO will comfortably cover the range of g aγ invoked as ALP solutions to the pos-sible anomalies observed in the propagation of high energy photon over astronomicaldistances, fully testing this hypothesis. • IAXO will cover a large fraction of the region of parameter space invoked as possi-ble solutions to the anomalies observed in the cooling of several stellar systems, bothinvolving g aγ and g ae . • The parameter space to be covered by IAXO could contain a viable dark matter can-didate. In particular, post-inflation models with N DM > m a in this range to account for the totality of the DM density. In addition, that very( g aγ , m a ) region is also suggested by ALP-miracle models recently proposed in whichALPs can account for both DM and inflation. • Combined with a positive detection in a haloscope, it will break the g aγ ρ a degeneracyand determine the local density of the galactic axionic dark matter. • Although not covered in this paper, the IAXO infrastructure could be used to hostadditional experimental setups, in particular microwave cavities or other devices ableto directly search for DM axions.IAXO will play a prominent role in the new generation of axion experiments currentlyunder proposal or preparation. It is highly complementary with the other experimentalfrontiers (laboratory and relic axions). Collectively a relatively large fraction of the mostmotivated parameter space for axions will be explored, potentially leading to substantialprogress in the low energy frontier in the coming years.– 61 – cknowledgments
We acknowledge support from the the European Research Council (ERC) under the EuropeanUnion’s Horizon 2020 research and innovation programme, grant agreement ERC-2017-AdG-788781 (IAXO+) as well as ERC-2018-StG-802836 (AxScale), and from the Spanish MINECOunder grant FPA2016-76978-C3-1-P. Part of this work was performed under the auspices ofthe U.S. Department of Energy by Lawrence Livermore National Laboratory under ContractDE-AC52-07NA27344. JR is supported by the Ramon y Cajal Fellowship 2012-10597, thegrant FPA2015-65745-P (MINECO/FEDER), the EU through the ITN “Elusives” H2020-MSCA-ITN-2015/674896 and the Deutsche Forschungsgemeinschaft under grant SFB-1258as a Mercator Fellow. We also acknowledge partial support of RFBR (grants 17-02-00305A,16-29-13014 ofi-m).
References [1] J. Jaeckel and A. Ringwald,
The Low-Energy Frontier of Particle Physics , Ann. Rev. Nucl.Part. Sci. (2010) 405–437, [ ].[2] A. Ringwald, Exploring the Role of Axions and Other WISPs in the Dark Universe , Phys.Dark Univ. (2012) 116–135, [ ].[3] Fundamental Physics at the Intensity Frontier , 2012. 10.2172/1042577.[4] P. W. Graham, I. G. Irastorza, S. K. Lamoreaux, A. Lindner and K. A. van Bibber,
Experimental Searches for the Axion and Axion-Like Particles , Ann. Rev. Nucl. Part. Sci. (2015) 485–514, [ ].[5] I. G. Irastorza and J. Redondo, New experimental approaches in the search for axion-likeparticles , .[6] IAXO collaboration, I. G. Irastorza,
The International Axion Observatory IAXO. Letter ofIntent to the CERN SPS committee , Tech. Rep. CERN-SPSC-2013-022. SPSC-I-242, CERN,Geneva, Aug, 2013.[7] E. Armengaud et al.,
Conceptual Design of the International Axion Observatory (IAXO) , JINST (2014) T05002, [ ].[8] H.-Y. Cheng, The Strong CP Problem Revisited , Phys. Rept. (1988) 1.[9] J. M. Pendlebury et al.,
Revised experimental upper limit on the electric dipole moment of theneutron , Phys. Rev.
D92 (2015) 092003, [ ].[10] M. Pospelov and A. Ritz,
Electric dipole moments as probes of new physics , Annals Phys. (2005) 119–169, [ hep-ph/0504231 ].[11] R. D. Peccei and H. R. Quinn,
CP Conservation in the Presence of Instantons , Phys. Rev.Lett. (1977) 1440–1443.[12] R. D. Peccei and H. R. Quinn, Constraints Imposed by CP Conservation in the Presence ofInstantons , Phys. Rev.
D16 (1977) 1791–1797.[13] F. Wilczek,
Problem of Strong P and T Invariance in the Presence of Instantons , Phys. Rev.Lett. (1978) 279–282.[14] G. Grilli di Cortona, E. Hardy, J. Pardo Vega and G. Villadoro, The QCD axion, precisely , JHEP (2016) 034, [ ].[15] D. B. Kaplan, Opening the Axion Window , Nucl. Phys.
B260 (1985) 215–226.[16] S. L. Cheng, C. Q. Geng and W. T. Ni,
Axion - photon couplings in invisible axion models , Phys. Rev.
D52 (1995) 3132–3135, [ hep-ph/9506295 ]. – 62 –
17] J. E. Kim,
Constraints on very light axions from cavity experiments , Phys. Rev.
D58 (1998)055006, [ hep-ph/9802220 ].[18] L. Di Luzio, F. Mescia and E. Nardi,
Redefining the Axion Window , Phys. Rev. Lett. (2017) 031801, [ ].[19] M. Farina, D. Pappadopulo, F. Rompineve and A. Tesi,
The photo-philic QCD axion , JHEP (2017) 095, [ ].[20] P. Agrawal, J. Fan, M. Reece and L.-T. Wang, Experimental Targets for Photon Couplings ofthe QCD Axion , JHEP (2018) 006, [ ].[21] E. Witten, Some Properties of O(32) Superstrings , Phys. Lett. (1984) 351–356.[22] J. P. Conlon,
The QCD axion and moduli stabilisation , JHEP (2006) 078,[ hep-th/0602233 ].[23] P. Svrcek and E. Witten, Axions In String Theory , JHEP (2006) 051, [ hep-th/0605206 ].[24] A. Arvanitaki, S. Dimopoulos, S. Dubovsky, N. Kaloper and J. March-Russell, StringAxiverse , Phys. Rev.
D81 (2010) 123530, [ ].[25] J. Redondo,
Solar axion flux from the axion-electron coupling , JCAP (2013) 008,[ ].[26] M. Dine, W. Fischler and M. Srednicki,
A Simple Solution to the Strong CP Problem with aHarmless Axion , Phys. Lett. (1981) 199–202.[27] A. R. Zhitnitsky,
On Possible Suppression of the Axion Hadron Interactions. (In Russian) , Sov. J. Nucl. Phys. (1980) 260.[28] A. Ernst, A. Ringwald and C. Tamarit, Axion Predictions in SO (10) × U (1) PQ Models , JHEP (2018) 103, [ ].[29] P. Sikivie, Experimental Tests of the Invisible Axion , Phys. Rev. Lett. (1983) 1415–1417.[30] K. Van Bibber, N. R. Dagdeviren, S. E. Koonin, A. Kerman and H. N. Nelson, Proposedexperiment to produce and detect light pseudoscalars , Phys. Rev. Lett. (1987) 759–762.[31] G. Raffelt and L. Stodolsky, Mixing of the Photon with Low Mass Particles , Phys. Rev.
D37 (1988) 1237.[32] J. Redondo and A. Ringwald,
Light shining through walls , Contemp. Phys. (2011) 211–236,[ ].[33] R. Bernabei et al., Search for solar axions by Primakoff effect in NaI crystals , Phys. Lett.
B515 (2001) 6–12.[34] D. Li, R. J. Creswick, F. T. Avignone and Y. Wang,
Theoretical Estimate of the Sensitivity ofthe CUORE Detector to Solar Axions , JCAP (2015) 065, [ ].[35] R. B¨ahre et al.,
Any light particle search II —Technical Design Report , JINST (2013)T09001, [ ].[36] G. G. Raffelt, J. Redondo and N. Viaux Maira, The meV mass frontier of axion physics , Phys. Rev.
D84 (2011) 103008, [ ].[37]
CAST collaboration, K. Zioutas et al.,
First results from the CERN Axion Solar Telescope(CAST) , Phys. Rev. Lett. (2005) 121301, [ hep-ex/0411033 ].[38] CAST collaboration, S. Andriamonje et al.,
An Improved limit on the axion-photon couplingfrom the CAST experiment , JCAP (2007) 010, [ hep-ex/0702006 ].[39]
CAST collaboration, E. Arik et al.,
Probing eV-scale axions with CAST , JCAP (2009)008, [ ]. – 63 – CAST collaboration, M. Arik et al.,
Search for Solar Axions by the CERN Axion SolarTelescope with He Buffer Gas: Closing the Hot Dark Matter Gap , Phys. Rev. Lett. (2014) 091302, [ ].[41] K. J. Bae, J.-H. Huh and J. E. Kim,
Update of axion CDM energy , JCAP (2008) 005,[ ].[42] L. Visinelli,
Axions in Cold Dark Matter and Inflation Models , Ph.D. thesis, Utah U., 2011. .[43] P. Arias, D. Cadamuro, M. Goodsell, J. Jaeckel, J. Redondo and A. Ringwald,
WISPy ColdDark Matter , JCAP (2012) 013, [ ].[44] G. G. Raffelt,
Particle physics from stars , Ann. Rev. Nucl. Part. Sci. (1999) 163–216,[ hep-ph/9903472 ].[45] K. Choi, K. S. Jeong, K.-I. Okumura and M. Yamaguchi, Mixed Mediation of SupersymmetryBreaking with Anomalous U(1) Gauge Symmetry , JHEP (2011) 049, [ ].[46] K.-S. Choi, I.-W. Kim and J. E. Kim, String compactification, QCD axion andaxion-photon-photon coupling , JHEP (2007) 116, [ hep-ph/0612107 ].[47] M. Cicoli, M. Goodsell and A. Ringwald, The type IIB string axiverse and its low-energyphenomenology , JHEP (2012) 146, [ ].[48] E. Pajer and M. Peloso, A review of Axion Inflation in the era of Planck , Class. Quant. Grav. (2013) 214002, [ ].[49] P. Sikivie, Axion Cosmology , Lect. Notes Phys. (2008) 19–50, [ astro-ph/0610440 ].[50] M. Cicoli, J. P. Conlon and F. Quevedo,
Dark radiation in LARGE volume models , Phys.Rev.
D87 (2013) 043520, [ ].[51] T. Higaki and F. Takahashi,
Dark Radiation and Dark Matter in Large VolumeCompactifications , JHEP (2012) 125, [ ].[52] D. M. Ghilencea, L. E. Ibanez, N. Irges and F. Quevedo, TeV scale Z-prime bosons fromD-branes , JHEP (2002) 016, [ hep-ph/0205083 ].[53] M. Goodsell, J. Jaeckel, J. Redondo and A. Ringwald, Naturally Light Hidden Photons inLARGE Volume String Compactifications , JHEP (2009) 027, [ ].[54] M. Cicoli, M. Goodsell, J. Jaeckel and A. Ringwald, Testing String Vacua in the Lab: From aHidden CMB to Dark Forces in Flux Compactifications , JHEP (2011) 114, [ ].[55] T. W. Grimm and J. Louis, The Effective action of N = 1 Calabi-Yau orientifolds , Nucl.Phys.
B699 (2004) 387–426, [ hep-th/0403067 ].[56] R. Allahverdi, M. Cicoli, B. Dutta and K. Sinha,
Correlation between Dark Matter and DarkRadiation in String Compactifications , JCAP (2014) 002, [ ].[57] G. D. Coughlan, W. Fischler, E. W. Kolb, S. Raby and G. G. Ross,
Cosmological Problemsfor the Polonyi Potential , Phys. Lett. (1983) 59–64.[58] T. Banks, D. B. Kaplan and A. E. Nelson,
Cosmological implications of dynamicalsupersymmetry breaking , Phys. Rev.
D49 (1994) 779–787, [ hep-ph/9308292 ].[59] B. de Carlos, J. A. Casas, F. Quevedo and E. Roulet,
Model independent properties andcosmological implications of the dilaton and moduli sectors of 4-d strings , Phys. Lett.
B318 (1993) 447–456, [ hep-ph/9308325 ].[60] R. Blumenhagen, M. Cvetic, S. Kachru and T. Weigand,
D-Brane Instantons in Type IIOrientifolds , Ann. Rev. Nucl. Part. Sci. (2009) 269–296, [ ].[61] M. B. Green and J. H. Schwarz, Anomaly Cancellation in Supersymmetric D=10 GaugeTheory and Superstring Theory , Phys. Lett. (1984) 117–122. – 64 –
62] R. Blumenhagen, J. P. Conlon, S. Krippendorf, S. Moster and F. Quevedo,
SUSY Breaking inLocal String/F-Theory Models , JHEP (2009) 007, [ ].[63] L. Aparicio, M. Cicoli, S. Krippendorf, A. Maharana, F. Muia and F. Quevedo, Sequestered deSitter String Scenarios: Soft-terms , JHEP (2014) 071, [ ].[64] M. Cicoli, V. A. Diaz, V. Guidetti and M. Rummel, The 3.5 keV Line from Stringy Axions , JHEP (2017) 192, [ ].[65] M. Srednicki, Axion Couplings to Matter. 1. CP Conserving Parts , Nucl. Phys.
B260 (1985)689–700.[66] J. E. Kim,
Weak Interaction Singlet and Strong CP Invariance , Phys. Rev. Lett. (1979)103.[67] M. A. Shifman, A. I. Vainshtein and V. I. Zakharov, Can Confinement Ensure Natural CPInvariance of Strong Interactions? , Nucl. Phys.
B166 (1980) 493–506.[68] L. Di Luzio, F. Mescia and E. Nardi,
Window for preferred axion models , Phys. Rev.
D96 (2017) 075003, [ ].[69] F. Wilczek,
Axions and Family Symmetry Breaking , Phys. Rev. Lett. (1982) 1549–1552.[70] L. Di Luzio, F. Mescia, E. Nardi, P. Panci and R. Ziegler, The Astrophobic Axion , .[71] F. Arias-Aragon and L. Merlo, The Minimal Flavour Violating Axion , JHEP (2017) 168,[ ].[72] L. Calibbi, F. Goertz, D. Redigolo, R. Ziegler and J. Zupan, Minimal axion model from flavor , Phys. Rev.
D95 (2017) 095009, [ ].[73] Y. Ema, K. Hamaguchi, T. Moroi and K. Nakayama,
Flaxion: a minimal extension to solvepuzzles in the standard model , JHEP (2017) 096, [ ].[74] F. Bj¨orkeroth, L. Di Luzio, F. Mescia and E. Nardi, U (1) flavour symmetries as Peccei-Quinnsymmetries , JHEP (2019) 133, [ ].[75] G. Ballesteros, J. Redondo, A. Ringwald and C. Tamarit, StandardModel—axion—seesaw—Higgs portal inflation. Five problems of particle physics andcosmology solved in one stroke , JCAP (2017) 001, [ ].[76] G. Ballesteros, J. Redondo, A. Ringwald and C. Tamarit,
Unifying inflation with the axion,dark matter, baryogenesis and the seesaw mechanism , Phys. Rev. Lett. (2017) 071802,[ ].[77] A. G. Dias, A. C. B. Machado, C. C. Nishi, A. Ringwald and P. Vaudrevange,
The Quest foran Intermediate-Scale Accidental Axion and Further ALPs , JHEP (2014) 037,[ ].[78] E. Ma, T. Ohata and K. Tsumura, Majoron as the QCD axion in a radiative seesaw model , Phys. Rev.
D96 (2017) 075039, [ ].[79] T. Dafni, C. A. J. O’Hare, B. Laki´c, J. Gal´an, F. J. Iguaz, I. G. Irastorza et al.,
Weighing thesolar axion , Phys. Rev.
D99 (2019) 035037, [ ].[80] J. Jaeckel and L. J. Thormaehlen,
Distinguishing Axion Models with IAXO , JCAP (2019) 039, [ ].[81] J. Preskill, M. B. Wise and F. Wilczek,
Cosmology of the Invisible Axion , Phys. Lett. (1983) 127–132.[82] L. F. Abbott and P. Sikivie,
A Cosmological Bound on the Invisible Axion , Phys. Lett. (1983) 133–136.[83] M. Dine and W. Fischler,
The Not So Harmless Axion , Phys. Lett. (1983) 137–141. – 65 –
Planck collaboration, N. Aghanim et al.,
Planck 2018 results. VI. Cosmological parameters , .[85] O. Wantz and E. P. S. Shellard, Axion Cosmology Revisited , Phys. Rev.
D82 (2010) 123508,[ ].[86] R. L. Davis,
Cosmic Axions from Cosmic Strings , Phys. Lett.
B180 (1986) 225–230.[87] D. H. Lyth,
Estimates of the cosmological axion density , Phys. Lett.
B275 (1992) 279–283.[88] S. Borsanyi et al.,
Calculation of the axion mass based on high-temperature lattice quantumchromodynamics , Nature (2016) 69–71, [ ].[89] R. D. Pisarski and L. G. Yaffe,
THE DENSITY OF INSTANTONS AT FINITETEMPERATURE , Phys. Lett. (1980) 110–112.[90] D. J. Gross, R. D. Pisarski and L. G. Yaffe,
QCD and Instantons at Finite Temperature , Rev.Mod. Phys. (1981) 43.[91] S. Borsanyi, M. Dierigl, Z. Fodor, S. D. Katz, S. W. Mages, D. Nogradi et al., Axioncosmology, lattice QCD and the dilute instanton gas , Phys. Lett.
B752 (2016) 175–181,[ ].[92] M. Gorghetto and G. Villadoro,
Topological Susceptibility and QCD Axion Mass: QED andNNLO corrections , JHEP (2019) 033, [ ].[93] C. Bonati, M. D’Elia, M. Mariti, G. Martinelli, M. Mesiti, F. Negro et al., Axionphenomenology and θ -dependence from N f = 2 + 1 lattice QCD , JHEP (2016) 155,[ ].[94] P. Petreczky, H.-P. Schadler and S. Sharma, The topological susceptibility in finite temperatureQCD and axion cosmology , Phys. Lett.
B762 (2016) 498–505, [ ].[95] A. D. Linde,
Axions in inflationary cosmology , Phys. Lett.
B259 (1991) 38–47.[96] S. Kasuya and M. Kawasaki,
Axion isocurvature fluctuations with extremely blue spectrum , Phys. Rev.
D80 (2009) 023516, [ ].[97] S. Folkerts, C. Germani and J. Redondo,
Axion Dark Matter and Planck favor non-minimalcouplings to gravity , Phys. Lett.
B728 (2014) 532–536, [ ].[98] K. S. Jeong and F. Takahashi,
Suppressing Isocurvature Perturbations of QCD Axion DarkMatter , Phys. Lett.
B727 (2013) 448–451, [ ].[99] A. D. Linde,
Generation of Isothermal Density Perturbations in the Inflationary Universe , Phys. Lett. (1985) 375–380.[100] D. Seckel and M. S. Turner,
Isothermal Density Perturbations in an Axion DominatedInflationary Universe , Phys. Rev.
D32 (1985) 3178.[101] T. Kobayashi, R. Kurematsu and F. Takahashi,
Isocurvature Constraints and AnharmonicEffects on QCD Axion Dark Matter , JCAP (2013) 032, [ ].[102] M. Kawasaki, K. Kohri and N. Sugiyama,
Cosmological constraints on late time entropyproduction , Phys. Rev. Lett. (1999) 4168, [ astro-ph/9811437 ].[103] M. Kawasaki, K. Kohri and N. Sugiyama, MeV scale reheating temperature andthermalization of neutrino background , Phys. Rev.
D62 (2000) 023506, [ astro-ph/0002127 ].[104] S. Hannestad,
What is the lowest possible reheating temperature? , Phys. Rev.
D70 (2004)043506, [ astro-ph/0403291 ].[105] K. Ichikawa, M. Kawasaki and F. Takahashi,
The Oscillation effects on thermalization of theneutrinos in the Universe with low reheating temperature , Phys. Rev.
D72 (2005) 043522,[ astro-ph/0505395 ]. – 66 – Constraint on the Effective Number of NeutrinoSpecies from the WMAP and SDSS LRG Power Spectra , JCAP (2007) 007,[ astro-ph/0611784 ].[107] P. F. de Salas, M. Lattanzi, G. Mangano, G. Miele, S. Pastor and O. Pisanti,
Bounds on verylow reheating scenarios after Planck , Phys. Rev.
D92 (2015) 123534, [ ].[108] P. W. Graham and A. Scherlis,
Stochastic axion scenario , Phys. Rev.
D98 (2018) 035017,[ ].[109] F. Takahashi, W. Yin and A. H. Guth,
QCD axion window and low-scale inflation , Phys. Rev.
D98 (2018) 015042, [ ].[110] I. I. Tkachev,
Phase transitions at preheating , Phys. Lett.
B376 (1996) 35–40,[ hep-th/9510146 ].[111] S. Kasuya and M. Kawasaki,
Topological defects formation after inflation on latticesimulation , Phys. Rev.
D58 (1998) 083516, [ hep-ph/9804429 ].[112] I. Tkachev, S. Khlebnikov, L. Kofman and A. D. Linde,
Cosmic strings from preheating , Phys.Lett.
B440 (1998) 262–268, [ hep-ph/9805209 ].[113] S. Kasuya and M. Kawasaki,
Remarks on cosmic string formation during preheating on latticesimulations , Phys. Rev.
D61 (2000) 083510, [ hep-ph/9903324 ].[114] M. Kawasaki, T. T. Yanagida and K. Yoshino,
Domain wall and isocurvature perturbationproblems in axion models , JCAP (2013) 030, [ ].[115] T. W. B. Kibble,
Topology of Cosmic Domains and Strings , J. Phys. A9 (1976) 1387–1398.[116] T. W. B. Kibble, Evolution of a system of cosmic strings , Nucl. Phys.
B252 (1985) 227.[117] D. P. Bennett,
The evolution of cosmic strings , Phys. Rev.
D33 (1986) 872.[118] M. Gorghetto, E. Hardy and G. Villadoro,
Axions from Strings: the Attractive Solution , JHEP (2018) 151, [ ].[119] P. Sikivie, Of Axions, Domain Walls and the Early Universe , Phys. Rev. Lett. (1982)1156–1159.[120] A. Vilenkin and A. E. Everett, Cosmic Strings and Domain Walls in Models with Goldstoneand PseudoGoldstone Bosons , Phys. Rev. Lett. (1982) 1867–1870.[121] Ya. B. Zeldovich, I. Yu. Kobzarev and L. B. Okun, Cosmological Consequences of theSpontaneous Breakdown of Discrete Symmetry , Zh. Eksp. Teor. Fiz. (1974) 3–11.[122] A. Vilenkin, Gravitational Field of Vacuum Domain Walls and Strings , Phys. Rev.
D23 (1981) 852–857.[123] G. B. Gelmini, M. Gleiser and E. W. Kolb,
Cosmology of Biased Discrete Symmetry Breaking , Phys. Rev.
D39 (1989) 1558.[124] K.-S. Choi, H. P. Nilles, S. Ramos-Sanchez and P. K. S. Vaudrevange,
Accions , Phys. Lett.
B675 (2009) 381–386, [ ].[125] A. Ringwald and K. Saikawa,
Axion dark matter in the post-inflationary Peccei-Quinnsymmetry breaking scenario , Phys. Rev.
D93 (2016) 085031, [ ].[126] D. Harari and P. Sikivie,
On the Evolution of Global Strings in the Early Universe , Phys. Lett.
B195 (1987) 361–365.[127] R. L. Davis and E. P. S. Shellard,
DO AXIONS NEED INFLATION? , Nucl. Phys.
B324 (1989) 167–186.[128] A. Dabholkar and J. M. Quashnock,
Pinning Down the Axion , Nucl. Phys.
B333 (1990)815–832. – 67 –
Computer simulations of the motion and decay of global strings , Nucl. Phys.
B363 (1991) 247–280.[130] R. A. Battye and E. P. S. Shellard,
Global string radiation , Nucl. Phys.
B423 (1994) 260–304,[ astro-ph/9311017 ].[131] R. A. Battye and E. P. S. Shellard,
Axion string constraints , Phys. Rev. Lett. (1994)2954–2957, [ astro-ph/9403018 ].[132] M. Yamaguchi, M. Kawasaki and J. Yokoyama, Evolution of axionic strings and spectrum ofaxions radiated from them , Phys. Rev. Lett. (1999) 4578–4581, [ hep-ph/9811311 ].[133] C. Hagmann, S. Chang and P. Sikivie, Axion radiation from strings , Phys. Rev.
D63 (2001)125018, [ hep-ph/0012361 ].[134] M. Nagasawa and M. Kawasaki,
Collapse of axionic domain wall and axion emission , Phys.Rev.
D50 (1994) 4821–4826, [ astro-ph/9402066 ].[135] S. Chang, C. Hagmann and P. Sikivie,
Studies of the motion and decay of axion walls boundedby strings , Phys. Rev.
D59 (1999) 023505, [ hep-ph/9807374 ].[136] T. Hiramatsu, M. Kawasaki, T. Sekiguchi, M. Yamaguchi and J. Yokoyama,
Improvedestimation of radiated axions from cosmological axionic strings , Phys. Rev.
D83 (2011)123531, [ ].[137] T. Hiramatsu, M. Kawasaki, K. Saikawa and T. Sekiguchi,
Production of dark matter axionsfrom collapse of string-wall systems , Phys. Rev.
D85 (2012) 105020, [ ].[138] T. Hiramatsu, M. Kawasaki, K. Saikawa and T. Sekiguchi,
Axion cosmology with long-liveddomain walls , JCAP (2013) 001, [ ].[139] M. Kawasaki, K. Saikawa and T. Sekiguchi,
Axion dark matter from topological defects , Phys.Rev.
D91 (2015) 065014, [ ].[140] L. Fleury and G. D. Moore,
Axion dark matter: strings and their cores , JCAP (2016)004, [ ].[141] L. M. Fleury and G. D. Moore,
Axion String Dynamics I: 2+1D , JCAP (2016) 005,[ ].[142] V. B. Klaer and G. D. Moore,
How to simulate global cosmic strings with large string tension , JCAP (2017) 043, [ ].[143] V. B. Klaer and G. D. Moore,
The dark-matter axion mass , .[144] A. Vaquero, J. Redondo and J. Stadler, Early seeds of axion miniclusters , .[145] M. Giannotti, I. G. Irastorza, J. Redondo, A. Ringwald and K. Saikawa, Stellar Recipes forAxion Hunters , JCAP (2017) 010, [ ].[146]
Planck collaboration, P. A. R. Ade et al.,
Planck 2015 results. XIII. Cosmologicalparameters , Astron. Astrophys. (2016) A13, [ ].[147] A. G. Riess et al.,
A 2.4% Determination of the Local Value of the Hubble Constant , Astrophys. J. (2016) 56, [ ].[148] R. Cooke, M. Pettini, R. A. Jorgenson, M. T. Murphy and C. C. Steidel,
Precision measuresof the primordial abundance of deuterium , Astrophys. J. (2014) 31, [ ].[149] P. Graf and F. D. Steffen,
Thermal axion production in the primordial quark-gluon plasma , Phys. Rev.
D83 (2011) 075011, [ ].[150] E. Masso, F. Rota and G. Zsembinszki,
On axion thermalization in the early universe , Phys.Rev.
D66 (2002) 023004, [ hep-ph/0203221 ]. – 68 – Thermal axion production , JCAP (2014) 011,[ ].[152] D. Baumann, D. Green and B. Wallisch,
New Target for Cosmic Axion Searches , Phys. Rev.Lett. (2016) 171301, [ ].[153] R. Z. Ferreira and A. Notari,
Observable windows for the QCD axion through N eff , .[154] F. D’Eramo, R. Z. Ferreira, A. Notari and J. L. Bernal, Hot Axions and the H tension , JCAP (2018) 014, [ ].[155] J. L. Bernal, L. Verde and A. G. Riess,
The trouble with H , .[156] A. G. Riess et al., New Parallaxes of Galactic Cepheids from Spatially Scanning the HubbleSpace Telescope: Implications for the Hubble Constant , Astrophys. J. (2018) 136,[ ].[157] A. G. Riess, S. Casertano, W. Yuan, L. M. Macri and D. Scolnic,
Large Magellanic CloudCepheid Standards Provide a 1% Foundation for the Determination of the Hubble Constantand Stronger Evidence for Physics Beyond LambdaCDM , .[158] D. Baumann, D. Green and B. Wallisch, Searching for light relics with large-scale structure , JCAP (2018) 029, [ ].[159] M. Laine and Y. Schroder,
Quark mass thresholds in QCD thermodynamics , Phys. Rev.
D73 (2006) 085009, [ hep-ph/0603048 ].[160]
CMB-S4 collaboration, K. N. Abazajian et al.,
CMB-S4 Science Book, First Edition , .[161] J. P. Conlon and M. C. D. Marsh, The Cosmophenomenology of Axionic Dark Radiation , JHEP (2013) 214, [ ].[162] K. Choi, A. Falkowski, H. P. Nilles and M. Olechowski, Soft supersymmetry breaking in KKLTflux compactification , Nucl. Phys.
B718 (2005) 113–133, [ hep-th/0503216 ].[163] B. S. Acharya, P. Kumar, K. Bobkov, G. Kane, J. Shao and S. Watson,
Non-thermal DarkMatter and the Moduli Problem in String Frameworks , JHEP (2008) 064, [ ].[164] S. Angus, Dark Radiation in Anisotropic LARGE Volume Compactifications , JHEP (2014)184, [ ].[165] A. Hebecker, P. Mangat, F. Rompineve and L. T. Witkowski, Dark Radiation predictions fromgeneral Large Volume Scenarios , JHEP (2014) 140, [ ].[166] M. Cicoli and F. Muia, General Analysis of Dark Radiation in Sequestered String Models , JHEP (2015) 152, [ ].[167] B. S. Acharya and C. Pongkitivanichkul, The Axiverse induced Dark Radiation Problem , JHEP (2016) 009, [ ].[168] D. Gorbunov and A. Tokareva, On the dark radiation problem in the axiverse , JCAP (2017) 016, [ ].[169] A. Mazumdar, S. Qutub and K. Saikawa,
Nonthermal axion dark radiation and constraints , Phys. Rev.
D94 (2016) 065030, [ ].[170] M. C. D. Marsh,
The Darkness of Spin-0 Dark Radiation , JCAP (2015) 017,[ ].[171] J. P. Conlon and M. C. D. Marsh,
Excess Astrophysical Photons from a 0.1–1 keV CosmicAxion Background , Phys. Rev. Lett. (2013) 151301, [ ]. – 69 – Theintergalactic magnetic field constrained by Fermi/LAT observations of the TeV blazar 1ES0229+200 , Mon. Not. Roy. Astron. Soc. (2010) L70–L74, [ ].[173] C. Evoli, M. Leo, A. Mirizzi and D. Montanino,
Reionization during the dark ages from acosmic axion background , JCAP (2016) 006, [ ].[174]
Planck collaboration, P. A. R. Ade et al.,
Planck 2015 results. XIX. Constraints onprimordial magnetic fields , Astron. Astrophys. (2016) A19, [ ].[175] H. B¨ohringer, G. Chon and P. P. Kronberg,
The Cosmic Large-Scale Structure in X-rays(CLASSIX) Cluster Survey I: Probing galaxy cluster magnetic fields with line of sight rotationmeasures , Astron. Astrophys. (2016) A22, [ ].[176] S. Angus, J. P. Conlon, M. C. D. Marsh, A. J. Powell and L. T. Witkowski,
Soft X-ray Excessin the Coma Cluster from a Cosmic Axion Background , JCAP (2014) 026, [ ].[177] M. Bonamente, R. Lieu, M. K. Joy and J. H. Nevalainen,
The soft x-ray emission in a largesample of galaxy clusters with rosat pspc , Astrophys. J. (2002) 688, [ astro-ph/0205473 ].[178] D. Kraljic, M. Rummel and J. P. Conlon,
ALP Conversion and the Soft X-ray Excess in theOutskirts of the Coma Cluster , JCAP (2015) 011, [ ].[179] A. J. Powell,
A Cosmic ALP Background and the Cluster Soft X-ray Excess in A665, A2199and A2255 , JCAP (2015) 017, [ ].[180]
Planck collaboration, P. A. R. Ade et al.,
Planck 2015 results. XX. Constraints on inflation , Astron. Astrophys. (2016) A20, [ ].[181] K. Freese, J. A. Frieman and A. V. Olinto,
Natural inflation with pseudo - Nambu-Goldstonebosons , Phys. Rev. Lett. (1990) 3233–3236.[182] F. C. Adams, J. R. Bond, K. Freese, J. A. Frieman and A. V. Olinto, Natural inflation:Particle physics models, power law spectra for large scale structure, and constraints fromCOBE , Phys. Rev.
D47 (1993) 426–455, [ hep-ph/9207245 ].[183] T. Rudelius,
On the Possibility of Large Axion Moduli Spaces , JCAP (2015) 049,[ ].[184] A. de la Fuente, P. Saraswat and R. Sundrum,
Natural Inflation and Quantum Gravity , Phys.Rev. Lett. (2015) 151303, [ ].[185] T. Rudelius,
Constraints on Axion Inflation from the Weak Gravity Conjecture , JCAP (2015) 020, [ ].[186] M. Montero, A. M. Uranga and I. Valenzuela,
Transplanckian axions!? , JHEP (2015) 032,[ ].[187] J. Brown, W. Cottrell, G. Shiu and P. Soler, Fencing in the Swampland: Quantum GravityConstraints on Large Field Inflation , JHEP (2015) 023, [ ].[188] M. Czerny and F. Takahashi, Multi-Natural Inflation , Phys. Lett.
B733 (2014) 241–246,[ ].[189] M. Czerny, T. Higaki and F. Takahashi,
Multi-Natural Inflation in Supergravity , JHEP (2014) 144, [ ].[190] M. Czerny, T. Higaki and F. Takahashi, Multi-Natural Inflation in Supergravity and BICEP2 , Phys. Lett.
B734 (2014) 167–172, [ ].[191] D. Croon and V. Sanz,
Saving Natural Inflation , JCAP (2015) 008, [ ].[192] R. Daido, F. Takahashi and W. Yin,
The ALP miracle: unified inflaton and dark matter , JCAP (2017) 044, [ ]. – 70 – UV-Protected Inflation , Phys. Rev. Lett. (2011) 161302,[ ].[194] F. Takahashi,
New inflation in supergravity after Planck and LHC , Phys. Lett.
B727 (2013)21–26, [ ].[195] R. Daido, F. Takahashi and W. Yin,
The ALP miracle revisited , .[196] T. Moroi, K. Mukaida, K. Nakayama and M. Takimoto, Axion Models with High ScaleInflation , JHEP (2014) 151, [ ].[197] G. N. Felder, J. Garcia-Bellido, P. B. Greene, L. Kofman, A. D. Linde and I. Tkachev, Dynamics of symmetry breaking and tachyonic preheating , Phys. Rev. Lett. (2001) 011601,[ hep-ph/0012142 ].[198] G. N. Felder, L. Kofman and A. D. Linde, Tachyonic instability and dynamics of spontaneoussymmetry breaking , Phys. Rev.
D64 (2001) 123517, [ hep-th/0106179 ].[199] R. Micha and I. I. Tkachev,
Turbulent thermalization , Phys. Rev.
D70 (2004) 043538,[ hep-ph/0403101 ].[200] K. D. Lozanov and M. A. Amin,
Self-resonance after inflation: oscillons, transients andradiation domination , .[201] A. Sarkar, S. Das and S. K. Sethi, How Late can the Dark Matter form in our universe? , JCAP (2015) 004, [ ].[202] A. Ringwald,
The hunt for axions , PoS
NEUTEL2015 (2015) 021, [ ].[203] M. Giannotti, I. Irastorza, J. Redondo and A. Ringwald,
Cool WISPs for stellar coolingexcesses , JCAP (2016) 057, [ ].[204] P. Di Vecchia, M. Giannotti, M. Lattanzi and A. Lindner,
Round Table on Axions andAxion-like Particles , in , 2019, .[205] A. H. Corsico, L. G. Althaus, M. M. M. Bertolami, A. D. Romero, E. Garcia-Berro, J. Isernet al.,
The rate of cooling of the pulsating white dwarf star G117 − B15A: a newasteroseismological inference of the axion mass , Mon. Not. Roy. Astron. Soc. (2012)2792, [ ].[206] A. H. Corsico, L. G. Althaus, A. D. Romero, A. S. Mukadam, E. Garcia-Berro, J. Isern et al.,
An independent limit on the axion mass from the variable white dwarf star R548 , JCAP (2012) 010, [ ].[207] A. H. C´orsico, L. G. Althaus, M. M. Miller Bertolami, S. O. Kepler and E. Garc´ıa-Berro,
Constraining the neutrino magnetic dipole moment from white dwarf pulsations , JCAP (2014) 054, [ ].[208] A. H. C´orsico, A. D. Romero, L. G. Althaus, E. Garc´ıa-Berro, J. Isern, S. O. Kepler et al.,
Anasteroseismic constraint on the mass of the axion from the period drift of the pulsating DAwhite dwarf star L19-2 , JCAP (2016) 036, [ ].[209] T. Battich, A. H. C´orsico, L. G. Althaus, M. M. Miller Bertolami and M. M. M. Bertolami,
First axion bounds from a pulsating helium-rich white dwarf star , JCAP (2016) 062,[ ].[210] M. M. Miller Bertolami,
Limits on the neutrino magnetic dipole moment from the luminosityfunction of hot white dwarfs , Astron. Astrophys. (2014) A123, [ ].[211] M. M. Miller Bertolami, B. E. Melendez, L. G. Althaus and J. Isern,
Revisiting the axionbounds from the Galactic white dwarf luminosity function , JCAP (2014) 069,[ ]. – 71 – Particle-physics constraints from the globular cluster M5: Neutrino Dipole Moments , Astron.Astrophys. (2013) A12, [ ].[213] N. Viaux, M. Catelan, P. B. Stetson, G. Raffelt, J. Redondo, A. A. R. Valcarce et al.,
Neutrino and axion bounds from the globular cluster M5 (NGC 5904) , Phys. Rev. Lett. (2013) 231301, [ ].[214] S. Arceo-D´ıaz, K. P. Schr¨oder, K. Zuber and D. Jack,
Constraint on the magnetic dipolemoment of neutrinos by the tip-RGB luminosity in ω -Centauri , Astropart. Phys. (2015)1–11.[215] A. Ayala, I. Dom´ınguez, M. Giannotti, A. Mirizzi and O. Straniero, Revisiting the bound onaxion-photon coupling from Globular Clusters , Phys. Rev. Lett. (2014) 191302,[ ].[216] O. Straniero, A. Ayala, M. Giannotti, A. Mirizzi and I. Dominguez,
Axion-Photon Coupling:Astrophysical Constraints , in
Proceedings, 11th Patras Workshop on Axions, WIMPs andWISPs (Axion-WIMP 2015): Zaragoza, Spain, June 22-26, 2015 , pp. 77–81, 2015, DOI.[217] R. C. D.-P. E. D. Skillman,
The ratio of blue to red supergiants in sextans a from hst imaging , astro-ph/0203284 .[218] K. B. W. McQuinn, E. D. Skillman, J. J. Dalcanton, A. E. Dolphin, J. Holtzman, D. R. Weiszet al., Observational Constraints on Red and Blue Helium Burning Sequences , Astrophys. J. (2011) 48, [ ].[219] A. Friedland, M. Giannotti and M. Wise,
Constraining the Axion-Photon Coupling withMassive Stars , Phys. Rev. Lett. (2013) 061101, [ ].[220] G. Carosi, A. Friedland, M. Giannotti, M. J. Pivovaroff, J. Ruz and J. K. Vogel,
Probing theaxion-photon coupling: phenomenological and experimental perspectives. A snowmass whitepaper , in
Proceedings, 2013 Community Summer Study on the Future of U.S. Particle Physics:Snowmass on the Mississippi (CSS2013): Minneapolis, MN, USA, July 29-August 6, 2013 ,2013,
Axion mass limit from observations of the neutron star in Cassiopeia A , JCAP (2014) 031, [ ].[222] M. Giannotti,
Hints of new physics from stars , PoS
ICHEP2016 (2016) 076, [ ].[223] S. Aoyama and T. K. Suzuki,
Effects of axions on Nucleosynthesis in massive stars , Phys.Rev.
D92 (2015) 063016, [ ].[224] I. Domınguez, M. Giannotti, A. Mirizzi and O. Straniero,
On the influence of axions on Mup , Mem. Soc. Ast. It. (2017) 270–273.[225] I. Dom´ınguez, M. Giannotti, A. Mirizzi and O. Straniero, Impact of Axions on the MinimumMass of Core Collapse Supernova Progenitors , in
Proceedings, 13th Patras Workshop onAxions, WIMPs and WISPs, (PATRAS 2017): Thessaloniki, Greece, 15 May 2017 - 19, 2017 ,pp. 151–154, 2018, , DOI.[226] O. Straniero, I. Dominguez, L. Piersanti, M. Giannotti and A. Mirizzi,
On the mass andluminosity of type II supernova progenitors. Hints for new physics? , in preparation.[227] S. O. Kepler, D. E. Winget, R. E. Nather, P. A. Bradley, A. D. Grauer, G. Fontaine et al.,
Adetection of the evolutionary time scale of the DA white dwarf G117 - B15A with the WholeEarth Telescope , Ap. J. (Sept., 1991) L45–L48.[228] J. Isern, M. Hernanz and E. Garcia-Berro,
Axion cooling of white dwarfs , Astrophys. J. (1992) L23. – 72 –
ConstrainingNeutrino Cooling using the Hot White Dwarf Luminosity Function in the Globular Cluster 47Tucanae , Astrophys. J. (2015) 141, [ ].[230]
LSST Dark Matter Group collaboration, A. Drlica-Wagner et al.,
Probing theFundamental Nature of Dark Matter with the Large Synoptic Survey Telescope , .[231] O. Straniero, I. Dominguez, M. Giannotti and A. Mirizzi, Axion-electron coupling from theRGB tip of Globular Clusters , in
Proceedings, 13th Patras Workshop on Axions, WIMPs andWISPs, (PATRAS 2017): Thessaloniki, Greece, 15 May 2017 - 19, 2017 , pp. 172–176, 2018, , DOI.[232] E. Pancino, M. Bellazzini, G. Giuffrida and S. Marinoni,
Globular clusters with Gaia , MonthlyNotices of the Royal Astronomical Society (May, 2017) 412–427, [ ].[233] D. Lauterborn, S. Refsdal and A. Weigert,
Stars with Central Helium Burning and theOccurrence of Loops in the H-R Diagram , Astron. Astrophys. (1971) 97–117.[234] M. Giannotti, Stellar Evolution Bounds on the ALP-Photon Coupling: new Results andPerspectives , in
Proceedings, 10th Patras Workshop on Axions, WIMPs and WISPs(AXION-WIMP 2014): Geneva, Switzerland, June 29-July 4, 2014 , pp. 71–74, 2014, , DOI.[235] W. C. G. Ho and C. O. Heinke,
A Neutron Star with a Carbon Atmosphere in the CassiopeiaA Supernova Remnant , Nature (2009) 71–73, [ ].[236] C. O. Heinke and W. C. G. Ho,
Direct Observation of the Cooling of the Cassiopeia ANeutron Star , Astrophys. J. (2010) L167–L171, [ ].[237] P. S. Shternin, D. G. Yakovlev, C. O. Heinke, W. C. G. Ho and D. J. Patnaude,
Coolingneutron star in the Cassiopeia A supernova remnant: Evidence for superfluidity in the core , Mon. Not. Roy. Astron. Soc. (2011) L108–L112, [ ].[238] J. Keller and A. Sedrakian,
Axions from cooling compact stars , Nucl. Phys.
A897 (2013)62–69, [ ].[239] A. Sedrakian,
Axion cooling of neutron stars , Phys. Rev.
D93 (2016) 065044, [ ].[240] K. Hamaguchi, N. Nagata, K. Yanagi and J. Zheng,
Limit on the Axion Decay Constant fromthe Cooling Neutron Star in Cassiopeia A , Phys. Rev.
D98 (2018) 103015, [ ].[241] M. V. Beznogov, E. Rrapaj, D. Page and S. Reddy,
Constraints on Axion-like Particles andNucleon Pairing in Dense Matter from the Hot Neutron Star in HESS J1731-347 , Phys. Rev.
C98 (2018) 035802, [ ].[242] L. B. Leinson,
Superfluid phases of triplet pairing and rapid cooling of the neutron star inCassiopeia A , Phys. Lett.
B741 (2015) 87–91, [ ].[243] G. G. Raffelt and D. S. P. Dearborn,
Bounds on Hadronic Axions From Stellar Evolution , Phys. Rev.
D36 (1987) 2211.[244] M. Giannotti,
ALP hints from cooling anomalies , in
Proceedings, 11th Patras Workshop onAxions, WIMPs and WISPs (Axion-WIMP 2015): Zaragoza, Spain, June 22-26, 2015 ,pp. 26–30, 2015, , DOI.[245] G. G. Raffelt,
Astrophysical axion bounds , Lect. Notes Phys. (2008) 51–71,[ hep-ph/0611350 ].[246] T. Fischer, S. Chakraborty, M. Giannotti, A. Mirizzi, A. Payez and A. Ringwald,
Probingaxions with the neutrino signal from the next galactic supernova , Phys. Rev.
D94 (2016)085012, [ ]. – 73 – Supernova 1987A Constraints on Sub-GeV DarkSectors, Millicharged Particles, the QCD Axion, and an Axion-like Particle , JHEP (2018)051, [ ].[248] M. Meyer, D. Horns and M. Raue, First lower limits on the photon-axion-like particle couplingfrom very high energy gamma-ray observations , Phys. Rev.
D87 (2013) 035027, [ ].[249] D. Hooper and P. D. Serpico,
Detecting Axionlike Particles with Gamma Ray Telescopes , Physical Review Letters (2007) 231102, [ ].[250] A. de Angelis, M. Roncadelli and O. Mansutti, Evidence for a new light spin-zero boson fromcosmological gamma-ray propagation? , Phys. Rev. D (2007) 121301, [ ].[251] A. De Angelis, O. Mansutti and M. Roncadelli, Axion-Like Particles, Cosmic Magnetic Fieldsand Gamma-Ray Astrophysics , Phys. Lett.
B659 (2008) 847–855, [ ].[252] N. Bassan and M. Roncadelli,
Photon-axion conversion in Active Galactic Nuclei? , ArXive-prints (2009) , [ ].[253] A. Kartavtsev, G. Raffelt and H. Vogel,
Extragalactic photon-ALP conversion at CTAenergies , .[254] L. M. Widrow, Origin of galactic and extragalactic magnetic fields , Reviews of ModernPhysics (2002) 775–823, [ astro-ph/0207240 ].[255] R. E. Pudritz, M. J. Hardcastle and D. C. Gabuzda, Magnetic Fields in Astrophysical Jets:From Launch to Termination , Space Sci. Rev. (2012) 27–72, [ ].[256] F. Tavecchio, M. Roncadelli and G. Galanti,
Photons to axion-like particles conversion inActive Galactic Nuclei , Phys. Lett.
B744 (2015) 375–379, [ ].[257] G. Galanti and M. Roncadelli,
Behavior of axionlike particles in smoothed out domainlikemagnetic fields , Phys. Rev.
D98 (2018) 043018, [ ].[258] M. Meyer, D. Montanino and J. Conrad,
On detecting oscillations of gamma rays intoaxion-like particles in turbulent and coherent magnetic fields , JCAP (2014) 003,[ ].[259] G. Galanti and M. Roncadelli,
Extragalactic photon-axion-like particle oscillations up to 1000TeV , Journal of High Energy Astrophysics (Nov., 2018) 1–17, [ ].[260] A. I. Nikishov, Absorption of high-energy photons in the Universe , Sov. Phys. JETP (1962) 393–394.[261] R. J. Gould and G. P. Schr´eder, Pair Production in Photon-Photon Collisions , PhysicalReview (1967) 1404–1407.[262] M. G. Hauser and E. Dwek,
The Cosmic Infrared Background: Measurements andImplications , ARA&A (2001) 249–307, [ astro-ph/0105539 ].[263] E. Dwek and F. Krennrich, The extragalactic background light and the gamma-ray opacity ofthe universe , Astroparticle Physics (2013) 112–133, [ ].[264] M. G. Hauser, R. G. Arendt et al., The COBE Diffuse Infrared Background ExperimentSearch for the Cosmic Infrared Background. I. Limits and Detections , ApJ (1998) 25–43,[ astro-ph/9806167 ].[265] J. D. Finke,
External Compton Scattering in Blazar Jets and the Location of the Gamma-RayEmitting Region , Astrophys. J. (2016) 94, [ ].[266] A. Mirizzi, G. G. Raffelt and P. D. Serpico,
Signatures of axion-like particles in the spectra ofTeV gamma-ray sources , Phys. Rev.
D76 (2007) 023001, [ ].[267] F. Tavecchio, M. Roncadelli, G. Galanti and G. Bonnoli,
Evidence for an axion-like particlefrom PKS 1222+216? , Phys. Rev. D (2012) 085036, [ ]. – 74 – ”Leading Blob” Model in a Stochastic AccelerationScenario: The Case of the 2009 Flare of Mkn 501 , ApJ (2011) L19, [ ].[269] M. Simet, D. Hooper and P. D. Serpico, Milky Way as a kiloparsec-scale axionscope ,Phys. Rev. D (2008) 063001, [ ].[270] A. De Angelis, O. Mansutti, M. Persic and M. Roncadelli, Photon propagation and the veryhigh energy γ -ray spectra of blazars: how transparent is the Universe? , MNRAS (2009)L21–L25, [ ].[271] M. A. S´anchez-Conde, D. Paneque, E. Bloom et al., Hints of the existence of axionlikeparticles from the gamma-ray spectra of cosmological sources , Phys. Rev. D (2009) 123511,[ ].[272] A. Dom´ınguez, M. A. S´anchez-Conde and F. Prada, Axion-like particle imprint incosmological very-high-energy sources , JCAP (2011) 20, [ ].[273] A. de Angelis, G. Galanti and M. Roncadelli, Relevance of axionlike particles forvery-high-energy astrophysics , Phys. Rev. D (2011) 105030, [ ].[274] W. Essey and A. Kusenko, On Weak Redshift Dependence of Gamma-Ray Spectra of DistantBlazars , ApJ (2012) L11, [ ].[275] G. Rubtsov and S. Troitsky,
Breaks in gamma-ray spectra of distant blazars and transparencyof the Universe , JETP Lett. (2014) 397–401, [ ].[276] G. Galanti, M. Roncadelli, A. De Angelis and G. F. Bignami,
Axion-like particles explain theunphysical redshift-dependence of AGN gamma-ray spectra , ArXiv e-prints (2015) ,[ ].[277] D. Horns and M. Meyer,
Indications for a pair-production anomaly from the propagation ofVHE gamma-rays , JCAP (2012) 33, [ ].[278] M. Meyer, D. Horns and M. Raue, Revisiting the Indication for a low opacity Universe forvery high energy gamma-rays , ArXiv e-prints (Nov., 2012) , [ ].[279] M. Meyer, D. Horns and M. Raue,
First lower limits on the photon-axion-like particle couplingfrom very high energy gamma-ray observations , Phys. Rev. D (2013) 035027, [ ].[280] K. Kohri and H. Kodama, Axion-Like Particles and Recent Observations of the CosmicInfrared Background Radiation , Phys. Rev.
D96 (2017) 051701, [ ].[281] W. Essey and A. Kusenko,
A new interpretation of the gamma-ray observations of distantactive galactic nuclei , Astroparticle Physics (2010) 81–85, [ ].[282] A. Furniss, P. M. Sutter, J. R. Primack and A. Dominguez, A Correlation Between HardGamma-ray Sources and Cosmic Voids Along the Line of Sight , ArXiv e-prints (2014) ,[ ].[283] J. Biteau and D. A. Williams,
The Extragalactic Background Light, the Hubble Constant, andAnomalies: Conclusions from 20 Years of TeV Gamma-Ray Observations , ApJ (2015)60, [ ].[284] D. A. Sanchez, S. Fegan and B. Giebels,
Evidence for a cosmological effect in gamma-rayspectra of BL Lacs , Astron. Astrophys. (2013) A75, [ ].[285] A. Dom´ınguez and M. Ajello,
Spectral Analysis of Fermi-LAT Blazars above 50 GeV , ApJ (2015) L34, [ ].[286]
MAGIC collaboration, E. Aliu et al.,
Very-High-Energy Gamma Rays from a DistantQuasar: How Transparent Is the Universe? , Science (2008) 1752, [ ].[287] J. Aleksi´c et al.,
MAGIC Discovery of Very High Energy Emission from the FSRQ PKS1222+21 , ApJ (Mar., 2011) L8, [ ]. – 75 – H.E.S.S. discovery of VHE γ -rays from the quasar PKS 1510-089 ,A&A (June, 2013) A107, [ ].[289] F. Tavecchio, J. Becerra-Gonzalez, G. Ghisellini et al., On the origin of the γ -ray emissionfrom the flaring blazar PKS 1222+216 , A&A (Oct., 2011) A86, [ ].[290] M. Petropoulou, K. Nalewajko, M. Hayashida and A. Mastichiadis, A hadronic minute-scaleGeV flare from quasar 3C 279? , MNRAS (May, 2017) L16–L20, [ ].[291] G. Galanti, F. Tavecchio, M. Roncadelli and C. Evoli,
Photon-ALP oscillations from a blazarto us up to 1000 TeV , .[292] C. Cs´aki, N. Kaloper and J. Terning, Dimming supernovae without cosmic acceleration , Phys.Rev. Lett. (Apr, 2002) 161302.[293] C. Burrage, A.-C. Davis and D. J. Shaw, Active galactic nuclei shed light on axionlikeparticles , Phys. Rev. Lett. (May, 2009) 201101.[294] G. W. Pettinari and R. Crittenden,
On the evidence for axionlike particles from activegalactic nuclei , Phys. Rev. D (Oct, 2010) 083502.[295] H.E.S.S. Collaboration collaboration, A. Abramowski, F. Acero, F. Aharonian et al.,
Constraints on axionlike particles with H.E.S.S. from the irregularity of the PKS 2155-304energy spectrum , Phys. Rev. D (2013) 102003, [ ].[296] Fermi-LAT collaboration, M. Ajello et al.,
Search for Spectral Irregularities due toPhoton–Axionlike-Particle Oscillations with the Fermi Large Area Telescope , Phys. Rev. Lett. (2016) 161101, [ ].[297] G. B. Taylor, N. E. Gugliucci, A. C. Fabian, J. S. Sanders, G. Gentile and S. W. Allen,
Magnetic fields in the centre of the Perseus cluster , MNRAS (June, 2006) 1500–1506,[ astro-ph/0602622 ].[298] J. W. Brockway, E. D. Carlson and G. G. Raffelt,
SN 1987A gamma-ray limits on theconversion of pseudoscalars , Physics Letters B (1996) 439–443,[ arXiv:astro-ph/9605197 ].[299] J. A. Grifols, E. Masso and R. Toldra,
Gamma-rays from SN1987A due to pseudoscalarconversion , Phys. Rev. Lett. (1996) 2372–2375, [ astro-ph/9606028 ].[300] A. Payez, C. Evoli, T. Fischer, M. Giannotti, A. Mirizzi and A. Ringwald, Revisiting theSN1987A gamma-ray limit on ultralight axion-like particles , JCAP (2015) 006,[ ].[301] I. Bombaci, G. Galanti and M. Roncadelli,
No axion-like particles from core-collapsesupernovae? , arXiv e-prints (Dec., 2017) , [ ].[302] F. Gat´e, R. Alves Batista, J. Biteau, J. Lefaucheur, S. Mangano, M. Meyer et al., Studyingcosmological γ -ray propagation with the Cherenkov Telescope Array , ArXiv e-prints (Sept.,2017) , [ ].[303] M. Meyer and J. Conrad,
Sensitivity of the Cherenkov Telescope Array to the detection ofaxion-like particles at high gamma-ray opacities , JCAP (2014) 16, [ ].[304] D. Wouters and P. Brun, Anisotropy test of the axion-like particle Universe opacity effect: acase for the Cherenkov Telescope Array , JCAP (2014) 16, [ ].[305] H. Vogel, R. Laha and M. Meyer, Diffuse axion-like particle searches , ArXiv e-prints (Dec.,2017) , [ ].[306] G. Galanti, M. Roncadelli, A. De Angelis and G. F. Bignami,
Advantages of axion-likeparticles for the description of very-high-energy blazar spectra , .[307] M. Meyer, M. Giannotti, A. Mirizzi, J. Conrad and M. Sanchez-Conde, The Fermi Large AreaTelescope as a Galactic Supernovae Axionscope , . – 76 – Constraints on Axion-like Particles from X-Ray Observations of theHydra Galaxy Cluster , Astrophys. J. (2013) 44, [ ].[309] M. Berg, J. P. Conlon, F. Day, N. Jennings, S. Krippendorf, A. J. Powell et al.,
Searches forAxion-Like Particles with NGC1275: Observation of Spectral Modulations , .[310] M. C. D. Marsh, H. R. Russell, A. C. Fabian, B. P. McNamara, P. Nulsen and C. S. Reynolds, A New Bound on Axion-Like Particles , .[311] J. P. Conlon, F. Day, N. Jennings, S. Krippendorf and M. Rummel, Constraints onAxion-Like Particles from Non-Observation of Spectral Modulations for X-ray Point Sources , JCAP (2017) 005, [ ].[312] J. P. Conlon, M. C. D. Marsh and A. J. Powell,
Galaxy cluster thermal x-ray spectra constrainaxionlike particles , Phys. Rev.
D93 (2016) 123526, [ ].[313] E. Bulbul, M. Markevitch, A. Foster, R. K. Smith, M. Loewenstein and S. W. Randall,
Detection of An Unidentified Emission Line in the Stacked X-ray spectrum of GalaxyClusters , Astrophys. J. (2014) 13, [ ].[314] A. Boyarsky, O. Ruchayskiy, D. Iakubovskyi and J. Franse,
Unidentified Line in X-RaySpectra of the Andromeda Galaxy and Perseus Galaxy Cluster , Phys. Rev. Lett. (2014)251301, [ ].[315] H. Ishida, K. S. Jeong and F. Takahashi, , Phys. Lett.
B732 (2014) 196–200, [ ].[316] D. P. Finkbeiner and N. Weiner,
X-ray line from exciting dark matter , Phys. Rev.
D94 (2016)083002, [ ].[317] T. Higaki, K. S. Jeong and F. Takahashi,
The 7 keV axion dark matter and the X-ray linesignal , Phys. Lett.
B733 (2014) 25–31, [ ].[318] J. Jaeckel, J. Redondo and A. Ringwald, , Phys. Rev.
D89 (2014) 103511, [ ].[319] H. M. Lee, S. C. Park and W.-I. Park,
Cluster X-ray line at . from axion-like darkmatter , Eur. Phys. J.
C74 (2014) 3062, [ ].[320] J.-C. Park, S. C. Park and K. Kong,
X-ray line signal from 7 keV axino dark matter decay , Phys. Lett.
B733 (2014) 217–220, [ ].[321] M. T. Frandsen, F. Sannino, I. M. Shoemaker and O. Svendsen,
X-ray Lines from DarkMatter: The Good, The Bad, and The Unlikely , JCAP (2014) 033, [ ].[322] K. Nakayama, F. Takahashi and T. T. Yanagida,
The 3.5 keV X-ray line signal from decayingmoduli with low cutoff scale , Phys. Lett.
B735 (2014) 338–339, [ ].[323] K.-Y. Choi and O. Seto,
X-ray line signal from decaying axino warm dark matter , Phys. Lett.
B735 (2014) 92–94, [ ].[324] M. Cicoli, J. P. Conlon, M. C. D. Marsh and M. Rummel, , Phys. Rev.
D90 (2014) 023540,[ ].[325] I. G. Irastorza et al.,
Towards a new generation axion helioscope , JCAP (2011) 013,[ ].[326] I. Shilon, A. Dudarev, H. Silva and H. H. J. ten Kate,
Conceptual Design of a New LargeSuperconducting Toroid for IAXO, the New International AXion Observatory , IEEE Trans.Appl. Supercond. (2013) 4500604, [ ].[327] H. H. J. ten Kate, The ATLAS superconducting magnet system at the Large Hadron Collider , Physica C (2008) 2137–2142. – 77 –
ATLAS Magnet System Nearing Completion , IEEE Trans. Appl.Supercond. (2008) 352–355.[329] F. A. Harrison, W. W. Craig, F. E. Christensent, C. J. Hailey and m. more, The nuclearspectroscopic telescope array (nustar) high-energy x-ray mission , Astrophysical Journal (2013) 103.[330] A. C. Jakobsen, M. J. Pivovaroff and F. E. Christensen,
X-ray optics for axion helioscopes , Proc. SPIE (2013) 886113–886113–7.[331] Y. Giomataris, P. Rebourgeard, J. P. Robert and G. Charpak,
MICROMEGAS: A Highgranularity position sensitive gaseous detector for high particle flux environments , Nucl.Instrum. Meth.
A376 (1996) 29–35.[332] S. Andriamonje, D. Attie, E. Berthoumieux, M. Calviani, P. Colas et al.,
Development andperformance of Microbulk Micromegas detectors , JINST (2010) P02001.[333] P. Abbon et al., The Micromegas detector of the CAST experiment , New J. Phys. (2007)170, [ physics/0702190 ].[334] S. Aune et al., X-ray detection with Micromegas with background levels below 10 − keV − cm − s − , JINST (2013) C12042, [ ].[335] F. Aznar et al., A Micromegas-based low-background x-ray detector coupled to a slumped-glasstelescope for axion research , JCAP (2015) 008, [ ].[336] C. Krieger, K. Desch, J. Kaminski, M. Lupberger and T. Vafeiadis,
An InGrid based LowEnergy X-ray Detector for the CAST Experiment , PoS
TIPP2014 (2014) 060.[337] S. Mertens, T. Lasserre, S. Groh, G. Drexlin, F. Glueck, A. Huber et al.,
Sensitivity ofNext-Generation Tritium Beta-Decay Experiments for keV-Scale Sterile Neutrinos , JCAP (2015) 020, [ ].[338] K. Dolde, S. Mertens, D. Radford, T. Bode, A. Huber, M. Korzeczek et al.,
Impact of ADCnon-linearities on the sensitivity to sterile keV neutrinos with a KATRIN-like experiment , Nucl. Instrum. Meth.
A848 (2017) 127–136, [ ].[339] P. C.-O. Ranitzsch, J.-P. Porst, S. Kempf, C. Pies, S. Sch¨afer, D. Hengstler et al.,
Developmentof metallic magnetic calorimeters for high precision measurements of calorimetric 187re and163ho spectra , Journal of Low Temperature Physics (2012) 1004–1014.[340] L. Gottardi et al.,
Development of TES-based detectors array for the X-ray Integral Field Unit(X-IFU) on the future x-ray observatory ATHENA , Proc. SPIE Int. Soc. Opt. Eng. (2014) 91442M, [ ].[341] F. Jansen et al.,
XMM-Newton observatory. I. The spacecraft and operations. , Astron.Astrophys. (2001) L1–L6.[342] J. Redondo, talk at Patras Workshop on Axions, WIMPs and WISPs, CERN, June 2014. http://axion-wimp2014.desy.de/ .[343] P. Sikivie, N. Sullivan and D. Tanner,
Axion Dark Matter Detection using an LC Circuit , .[344] Y. Kahn, B. R. Safdi and J. Thaler, Broadband and Resonant Approaches to Axion DarkMatter Detection , Phys. Rev. Lett. (2016) 141801, [ ].[345] M. Silva-Feaver et al.,
Design Overview of the DM Radio Pathfinder Experiment , .[346] A. l. Melc´on et al., Axion Searches with Microwave Filters: the RADES project , JCAP (2018) 040, [ ].[347]
The CAST Collaboration collaboration, H. Fischer and C. CAST,
CAST - Status Reportto the SPSC for the 127th Meeting and Planning for 2018 , Tech. Rep. CERN-SPSC-2017-041.SPSC-SR-223, CERN, Geneva, Oct, 2017. – 78 –
Prospects forsearching axionlike particle dark matter with dipole, toroidal, and wiggler magnets , Phys. Rev.D (Feb, 2012) 035018.[349] H. Baer, A. Lessa and W. Sreethawong, Coupled Boltzmann calculation of mixedaxion/neutralino cold dark matter production in the early universe , JCAP (2012) 036,[ ].].