Heavy Ion Physics at the LHC: What's new ? What's next ?
aa r X i v : . [ h e p - e x ] N ov Heavy Ion Physics at the LHC: What’s new ? What’snext ?
J. Schukraft ∗ CERN, DIV. PH, 1211 Geneva 23, Switzerland
Abstract
Towards the end of 2010, some 25 years after the very first collisions of ultra-relativistic heavy ions at fixed target energies, and some 10 years after the start ofoperation of the Relativistic Heavy Ion Collider (RHIC), the LHC opened a new erain heavy ion physics with lead on lead collisions at √ s NN = 2 .
76 TeV. After a shortreminder of the main results from lower energies, this review highlights a few selectedareas where significant progress has been made during the first three years of ion oper-ation at the LHC.
The subject of ultra-relativistic heavy ion physics is the study of strongly interacting matterunder extreme conditions of high temperature and/or high matter density. QCD predictsthat at a sufficiently high energy density there will be a transition from ordinary nuclearor hadronic matter to a plasma of free (’deconfined’) quarks and gluons, the ’Quark-GluonPlasma’ (QGP). The discovery and characterisation of this plasma phase is thought to requirea large volume of hot/dense matter and is therefore pursued in collisions of heavy nuclei atthe highest energies.This article presents a subjective selection of results and interpretations from the firstthree years of the LHC heavy ion program; it is based in parts on recent reviews [1, 2]in which most of the relevant LHC data and references to the primary literature can befound. Theoretical aspects and implications are covered and referenced in more detail in anaccompanying article [3].
After pioneering experiments at relativistic energies (GeV/nucleon) in the 1970’s in the USA(LBNL) and Russia (JINR), the quest for the Quark-Gluon Plasma took off in 1986 at the ∗ Talk given at the ’Nobel Symposium on LHC results’, Krusenberg, Sweden, 13 - 17 May 2013, to bepublished in
Physica Scripta ≈
30) and in the early 1990’s with actual heavy ions (mass ≈ ’compelling evidence has been found for a new state of matter,featuring many of the characteristics expected for a Quark-Gluon Plasma’ [8]. This conclusionwas based primarily on three experimental observations: the copious production of hadronscontaining a strange quark (’strangeness enhancement’), the yields of low mass lepton pairs(’rho melting’), and the reduced production of J/ Ψ mesons (’anomalous J/ Ψ suppression’).The very high abundance of strange particles, in particular of hyperons (the omega-to-pion ratio increases by up to a factor 20 from pp to PbPb!), was predicted as a consequenceof QGP formation. Today it is interpreted more generally as a manifestation of statisticalhadronisation from a thermalized medium, where most hadrons, not only those containingstrange quarks, are created in thermal equilibrium ratios [9–12]. These ratios are governedessentially by a single scale parameter T, interpreted as a chemical freeze-out temperature.The invariant mass distribution of prompt low mass lepton pairs showed an enhancementin the continuum yield just below the rho/omega resonance [14,15]. This was qualitatively inagreement with signals expected from chiral symmetry restoration, where the mass and/orwidth of hadrons are modified in the vicinity of the QGP phase boundary (in this case therho, observed inside the medium via its two lepton decay).Finally, J/ Ψ production was found significantly suppressed in central PbPb collisionsrelative to expectation (eg relative to a normalisation process like Drell Yan) beyond whatcould be attributed to confounding cold nuclear matter effects like changes in the nuclearparton distribution functions or hadronic final state interactions [17–19]. Dubbed ’anomalous’ J/ Ψ suppression to distinguish it from the ’normal’ one due to cold nuclear matter, the effectwas within experimental and theoretical uncertainties fully consistent with the ’smoking gun’signal predicted for deconfinement in the QGP.These experimental results have all stood the test of time, having been confirmed andrefined subsequently at the SPS as well as at RHIC. The essence of the assessment, that thereis a new state of matter in the SPS energy range, featuring some of the hallmarks of a QGP(thermalisation, deconfinement, chiral symmetry restoration) seems however, in hindsight,today more compelling than in 2000, given for example much improved low mass lepton pairresults from the SPS NA60 experiment [16] and new insights in some of the processes relevantfor thermal particle production and quarkonia suppression from RHIC and LHC (see below).The initial results from RHIC were summarized and assessed in 2005, based on a compre-hensive (re)analysis of the first few years of RHIC running [20]. The experiments concludedthat at RHIC ’a new state of hot, dense matter’ was created ’out of the quarks and gluons.. but it is a state quite different and even more remarkable than had been predicted’ [21].Unlike the expectation, with hindsight overly naive, that the QGP would resemble an almostideal gas of weakly coupled quarks and gluons, the hot matter was found to behave like anextremely strongly interacting, almost perfect liquid, sometimes called the sQGP (where the’s’ stand for ’strongly interacting’). It is almost opaque and absorbs much of the energy ofany fast parton which travels through – a process referred to as ’jet quenching – and it reactsto pressure gradients by flowing almost unimpeded and with very little internal friction (i.e.has very small shear viscosity) [22, 23]. 2lso at RHIC, the crucial experimental results as well as the inferred characteristics ofthe QGP — a ’hot, strongly interacting, nearly perfect liquid’ – stood the test of time [1].The temperature was inferred by measuring direct ’thermal’ photons with an inverse slopeof order 200 MeV, which leads to a (model dependent) estimate of an initial temperatureof at least 300 MeV. The characterisation of the QGP as almost opaque was based on theobservation of a very significant suppression, up to a factor of five, of high p T particlesfound in central, head-on nuclear collisions. This suppression of high p T particles, whichare typically leading jet fragments, is indicative of very strong final state interactions of thescattered partons with the medium, leading to significant energy loss via elastic scatteringor enhanced gluon radiation. As essential control measurements, the suppression was neitherseen in d-Au reactions, therefore excluding cold nuclear matter effects as the cause, nor withcolour neutral probes like direct photons, which clearly establishes the effect as due to thestrong (=QCD) interaction in the final state.The ’ideal liquid’ aspect of the QGP was based on the measurement of collective par-ticle motions, the so called ’elliptic flow’, which develops in response to initial geometricalconditions and internal pressure gradients in the nuclear overlap zone. The elliptic flowmagnitude at RHIC was found to essentially exhaust the maximal possible one predicted byhydrodynamics for the given initial deformation, equivalent to the response of an ideal liquidwith vanishing shear viscosity. The shear viscosity over entropy ratio, η/s was found to becompatible with a conjectured lower bound of η/s ≥ / π ( ~ = k B = 1), a value reached ina very strongly interacting system when the mean free path approaches the quantum limit,the Compton wavelength. Prior to LHC, 25 years of heavy ion experimentation had already revealed a ’QGP-like’ stateat the SPS and not ’ the
QGP’, but ’ a sQGP’ at RHIC. One could consider the discoveryphase for the QGP as essentially over, the qualitative characterisation as well under way, andquantitative precision measurements of QGP properties as just having started [24]. A maingoal for the heavy ion program at LHC was therefore to measure with increased precision theparameters which characterise this new state of matter, making use of the particular strengthof the LHC: a powerful new generation of large acceptance state-of-the-art experiments AT-LAS, CMS, ALICE, and LHCb , and a huge increase in beam energy with the associatedlarger cross sections for hard probes and higher particle density, which makes for a QGPwhich will be ’hotter, larger, and longer living’. And indeed, LHC made significant progresstowards increasing the precision on shear viscosity (see section 4) and plasma opacity (section5) already during the first two years of ion running. However, when dealing with QCD in thenonpertubative regime, surprises should not come as a surprise, and a number of unexpectedfindings at LHC have helped shed new light on some old problems or issues. Two of thosewill be mentioned below, relating to particle production (i.e. thermalisation, section 2) andquarkonia suppression (i.e. deconfinement, section 3). And finally the very first discoverymade at LHC is discussed in section 5: the appearance of a mysterious long range ’ridge’correlation in high multiplicity pp reactions. It reappeared later – and much stronger – in LHCb participates in the p-nucleus part of the heavy ion program
The production of different hadronic particle species in high energy collisions is a non-pertubative process and in general parameterized in phenomenological models and eventgenerators, requiring usually a large number of parameters. A rather more economical ap-proach is the thermal/statistical model of hadronisation [9–13], which assumes that particlesare created in thermal (or phase space) equilibrium; particles of mass m are essentially sup-pressed by a Boltzmann factor e − m/T . In its simplest implementation, the scale parameter T,identified as the chemical freeze-out temperature, is the only free parameter of the model anddefines the production of all hadronic species, together with some conservation laws (baryons,strangeness,..) and an overall normalisation constant proportional to the total particle mul-tiplicity. For small systems (small number of final state hadrons), an additional parameter γ s (or, equivalently, a reduced correlation volume) has to be introduced to describe the factthat strange hadrons are suppressed compared to the grand canonical thermal expectation.The temperature parameter T is found in all high energy collisions ( pp , e + e − , AA ) to beabout 160 MeV, while γ s increases from 0 . − . pp to 0 . − AA . The fact thatmost hadrons containing light and strange quarks are described by the thermal model withtypically better than 10-20% precision was considered an essential and well established factin heavy ion collisions.The origin of the success of the thermal model over a very large range of collision systemsand beam energies is however not obvious. One of two qualitative mechanisms are usuallyinvoked which can be summarized as i) born into (phase space) equilibrium or ii) evolving into(thermal) equilibrium [12, 25, 26]. The former i) could arise for example from Fermi’s goldenrule, the fact that reaction rates are proportional in this case to the product of a QCD matrixelement and a hadronic final state phase space factor. As many different channels (matrixelements) contribute to particle yields, phase space dominates in inclusive measurements andthe most conspicuous QCD remnant is the observed suppression ( γ s <
1) of strange particlesin elementary reactions ( pp , e + e − ), reflecting the higher mass of the strange quark and localstrangeness conservation. This strangeness suppression is then ’somehow’ lifted in largesystems ( AA ), effectively replacing local strangeness conservation by average global (grandcanonical) conservation. The alternative explanation ii) postulates that inelastic reactionseither in the partonic phase or in the final state hadronic phase drive initial abundancesquickly towards thermal equilibrium via detailed balance. Reaction rates, in particular thoseinvolving more than two initial state hadrons, decrease drastically with temperature andtherefore inelastic reactions seize abruptly when the system expands and cools, preservingchemical equilibrium ratios and a freeze-out temperature still very close to the hadronisationtransition.Both explanations for the uncanny success of statistical models are conceivable, but dif-ficult to underpin by quantitative dynamical calculations. And there is no fully satisfactoryexplanation for the strikingly similar,yet distinct, pattern of particle production in small4 pp , e + e − ) and large ( AA ) systems: In i) the mechanism of strangeness enhancement (i.e.the change in γ s or the correlation volume) remains qualitative and hand waving, whereasin ii) particle production in small systems, which would not be expected to thermalize, isusually not even addressed. Some experimental observations are also counterintuitive: Ifhadronisation is sudden, from an equilibrated partonic phase (the sQGP) with little inelastichadronic final state interactions, the particle ratios should reflect parton equilibrium, whichin general would result in particle fractions very different from hadron equilibrium. If, on thecontrary, particle yields are established via hadronic reactions, a common chemical freeze-outof all particle species seems a priori unlikely. One should expect to see at least some indica-tions for sequential freeze-out, where hadrons with large inelastic cross section stay longer inequilibrium and freeze out later (at lower temperature) than those with significantly smallerreaction cross sections.Measuring identified particles at LHC was nevertheless considered a somewhat boringexercise, as finding thermal particle ratios essentially identical to the ones measured at RHICwas thought to be one of the safest predictions [27]. It therefore came as quite a surprise whensome particle fractions, in particular for the mundane proton, one of the most frequentlyproduced hadrons, were found to differ considerably from expectations (and, to a lesserextent, from the ones measured at RHIC), while others, including those for multi-strangehyperons, were well in line with thermal predictions. Also the particle ratios measured inproton collisions seem less well described by thermal fits than data at lower energy, evenwhen allowing for strangeness under-saturation ( γ s <
1) [28].The increasing deviation of hadron ratios with energy in pp collisions, despite the overallincrease in multiplicity which should bring results closer to statistical predictions, could fitnaturally with the ’QCD plus phase space’ interpretation. At the LHC, cross sections forhard processes are large and hard scattering is important even for the average minimumbias collision, so maybe semi-hard QCD processes are making their presence felt as strongerdeviations from phase space dominance?Converging on a potential reason for the measured nuclear particle ratios at LHC provedharder, and a lively discussion is taking place centring on three explanations: i) Reducedthermal freeze-out temperature, ii) sequential hadron freeze-out, and iii) non-equilibriumparton freeze-out. Standard one parameter thermal fits (i) give a somewhat poor descriptionof measured particle ratios and return a chemical freeze-out temperature T significantly belowthe one extracted previously from RHIC (by 6 to 10 MeV) [29]. Even if one accepts the LHCfits as tolerable and within the range of accuracy of the thermal model, the question why andhow the chemical freeze-out temperature would come down with increasing energy remainsopen. Sequential models (ii) try to estimate abundance changes from inelastic hadronic finalstate interactions after initial chemical freeze-out [30]. In any thermally evolving system,sequential (time and temperature ordered) freeze-out of different degrees of freedom withdifferent cross sections and mean free paths must exist at some level; the question is one ofmagnitude rather than of principle. Some non-equilibrium calculations, usually done withthe help of event generators or kinetic transport models, give results remarkably close to theLHC data and also improve thermal model fits at lower energies, making this mechanisma plausible explanation. However, given the multitude of unknown hadronic reaction crosssections which are needed in the final state rescattering calculations, as well as the needto consider multiprong initial states to satisfy detailed balance requirements, make these5alculations more of an art than an exact science and the arguments qualitative rather thanquantitative ones. Finally (iii), extensions to the standard model [31], where non-equilibriumthermodynamics (including supercooling below the phase transition) is applied to the partonrather than to the hadron phase, can well fit all currently available data, albeit at the expenseof two additional free parameters.The final resolution of the ’proton puzzle’ is still outstanding, and will require a morecomplete set of particle ratio measurements at LHC as well as revisiting the RHIC results toconfirm with better significance if particle ratios in central nuclear collisions indeed evolvewith energy. Whichever explanation will finally prevail, the unexpected LHC results are awelcome fresh input likely to advance our understanding of the ’unreasonable success’ of thestatistical/thermal model of particle production. Heavy flavour quarks (charm, bottom) have always been an important tool to probe thequark gluon plasma [17–19]. They are created early in the collision, by (semi)hard processesamenable to QCD calculations, and their further dynamical evolution is then modified bythe surrounding medium. In particular the J/ ψ and Υ families should be suppressed inheavy ion collisions in comparison with pp , primarily as a consequence of deconfinement(’melting’) in the QGP. The magnitude of the suppression for different quarkonium statesshould depend on their binding energy, with strongly bound states such as the Υ showingless or no modification.While the ’anomalous’ J/ Ψ suppression discovered at the SPS was considered one of thestrongest indications for the QGP, the RHIC results showed essentially the same suppressionat a much higher energy, contrary to most expectations and predictions from both QGP andnon-QGP models. These initially very confusing results kept the interpretation of the smokinggun signal for deconfinement ambiguous for the last 10 years. A possible explanation was thatthe directly produced J/ ψ is not supressed at all, neither at SPS nor at RHIC, and only thehigh mass charmonium states ψ ’ and χ c , which populate about 40% of the observed J/ ψ yield,are affected by the medium. These weakly bound high mass states should dissociate very closeto or even below the critical transition temperature, but they also may be easily destroyedby hadronic final state interaction, without the need for invoking a QGP. Alternatively, ithas been suggested that J/ ψ suppression actually increases with energy but is more or lessbalanced by a new production mechanism, i.e. recombination at the phase boundary of twoindependently produced charm quarks [32].LHC data seems to have resolved the J/ ψ puzzle in favour of the coalescence picture [33].As predicted by recombination, the large charm cross section at LHC leads to less J/ ψ sup-pression at LHC, albeit not to J/ ψ enhancement relative to pp, which was within the range ofcoalescence prediction and would have made the case clear cut. The suppression is also lessstrong at low p T , where phase space favours recombination, in clear contrast to the opposite p T dependence found at SPS and RHIC.LHC results from the Υ family [34] are fully consistent with the expectation from adeconfining hot medium that quarkonia survival decreases with binding energy, i.e. in termsof suppression factors: Υ(3S) > Υ(2S) > Υ(1S). The Υ(1S) is suppressed by about a factor6f two in central collisions, the Υ(2S) by almost an order of magnitude, and only upper limitshave been measured for the Υ(3S). As only about 50% of the observed Υ(1S) are directlyproduced, these results are actually compatible with almost complete melting of all high massbottonium states and survival of a lone, strongly bound Υ(1S), which according to latticeQCD may melt only at temperatures far above the critical temperature.While at first sight charm quark coalescence may appear as yet another process compli-cating and masking quarkonium deconfinement, it is actually a respectable and importantdeconfinement signal in itself: only in a colour conducting, deconfining medium can quarksroam freely over large distances ( >> ψ production over a wide range of energies.Together with, and complementary to, the original advocated process of dissociation, it mayeven eventually deliver on the promise of quarkonia production as an unambiguous signal ofdeconfined partonic matter. The observation of robust collective flow phenomena in heavy ion reactions [35] is arguablythe most direct evidence for the creation of a strongly interacting, macroscopic (i.e. largecompared to the mean free path) and dense matter system in nuclear collisions. Matterproperties like the equation of state, sound velocity or shear viscosity, can be extractedby comparing measurements and hydrodynamic model calculations of elliptic (i.e. azimuthdependent) and radial (azimuthally averaged) flow. Flow depends not only on matter proper-ties but also on initial conditions, in particular the geometrical distribution of energy densitywithin the nuclear overlap zone and the resulting pressure gradients. In general the nuclearimpact parameter, and therefore the reaction zone geometry relevant for initial conditions,can be measured rather well event-by-event for each individual collision, using global eventobservables like particle multiplicity or forward ’zero degree’ energy. However, the remainingmodel dependence of the energy density profile is sufficiently large to dominate the systematicerrors and limit the accuracy with which the matter parameters can be extracted from thedata. Before LHC turn-on, the defining property of the ’perfect liquid’, the shear viscosity-to-entropy ratio η/s , was only known to be within a factor of about five to the conjecturedquantum limit 1 / π .Azimuthally dependent collective motions are usually analysed in terms of a Fourier ex-pansion with respect to the reaction plane, with the first order component, v ∝ cos( ϕ ),called directed and the second order component, v ∝ cos(2 ϕ ), called elliptic flow. Higherorder components were thought to be small or vanish for symmetry reasons. Between 2005and 2010, based on observations at RHIC, suggestions were made that the geometrical over-lap shape could fluctuate event-by-event, even at fixed impact parameter, because of thestochastic nature of nucleon-nucleon collisions. These fluctuations would generate ’lumpy’7nitial conditions which could give rise to higher harmonic Fourier components. However,these suggestions remained controversial.When first azimuthal flow data from the LHC became available in early 2011, the evidencefrom all three experiments, as well as new results shown by the two RHIC collaborations,was overwhelming [36]: Fluctuations, event-by-event, of the energy density in the initial statedo give rise to complex collective flow patterns, which, when analysed in terms of Fouriercoefficients, are measurable and significant up to at least 6th order ( v , v , ..v )! The signalstrength of different harmonics, their particle mass, centrality and momentum dependencewere all in excellent agreement with expectations from hydrodynamics.The correlation patterns induced by flow fluctuations had actually been strong and clearlyvisible since many years also in the RHIC data; however, before 2011, they were in general notrecognized as hydrodynamic in origin but discussed in terms of fancy names (’near side ridge,away side cone’) and fancy explanations (’gluon Cerenkov radiation, Mach cone, ..’) [37]. AtLHC, the large acceptance of the experiments, together with the high particle density (as acollective effect, the flow signal increases strongly with multiplicity) made the observationand interpretation straightforward and unambiguous.The fact that energy density fluctuations on the scale of a fraction of the nuclear radiusin the initial state are faithfully converted into measurable velocity fluctuations in the finalstate was a most amazing, and also most useful, discovery: One could not only identify theaverage, almond shaped ’face of the collision zone’, but recognize much finer structures, the’warts and wrinkles’, of nuclear collisions. The analysis of flow has been invigorated and isadvancing rapidly ever since, with direct measurements of the fluctuation spectrum [38], usingevent-by-event measurement and selection of flow as an analysis tool [39], and even findingnon-linear mode mixing between different harmonics [40]. Like temperature fluctuations inthe cosmic microwave background radiation, which can be mapped to initial state densityfluctuations in the early Universe, collective flow fluctuations strongly constrain the initialconditions (initial density distributions) and therefore allows a better measurement of fluidproperties. Since 2011, the limit for the shear viscosity has come down by a factor of two ( η/s smaller than two to three times 1 / π ). It is now precise enough to see a hint of a temperaturedependence (slightly increasing from RHIC to LHC) [3], and future improvements in dataaccuracy and hydro modelling should either further improve the limit, or give a finite value for η/s . In either case, improved precision is relevant as the shear viscosity is directly related tothe in-medium cross section and therefore contains information about the degrees of freedomrelevant in the sQGP via the strength and temperature dependence of their interactions. High energy partons interact with the medium and loose energy, primarily through inducedgluon radiation and, to a smaller extent, elastic scattering [41]. The amount of energy lost,∆ E , is expected to depend on medium properties, in particular the opacity (density, interac-tion strength) and the path length L inside the medium, with different models predicting alinear (elastic ∆ E ), quadratic (radiative ∆ E ), and even cubic (AdS/CFT) dependence on L.In addition, ∆ E also depends on the parton type via the colour charge (quark versus gluon),the parton mass via formation time and interference effects (light versus heavy quarks), and8nally somewhat on the jet energy. The total jet energy is of course conserved and the en-ergy lost by the leading parton appears mostly in the radiated gluons, leading in effect to amedium modified softer fragmentation function. Jet quenching (i.e. measuring the mediummodified fragmentation functions) is therefore a very rich observable which probes not onlythe properties of the medium but also properties of the strong interaction.Jet quenching was discovered at RHIC not with jets, which are difficult to measure in thehigh multiplicity heavy ion background environment, but as a suppression of high p T ’leading’jet-fragments. While the effect was experimentally very clean and significant with suppressionfactors up to five, the information was also very limited, impeding a quantitative and modelindependent determination of matter properties or energy loss mechanisms.The high energy of LHC and the correspondingly large cross sections for hard processesmake high energy jets easily stand out from the background even in central nuclear collisions.Jet quenching is therefore readily recognized and measured, with many unbalanced dijets oreven monojets apparent in the data [42]. While the amount of energy lost in the mediumcan be of order tens of GeV and therefore even on average corresponds to a sizeable fractionof the total jet energy, it is nevertheless close to the one expected when extrapolating RHICresults to the higher density matter at LHC. The two jets remain essentially back-to-back(little or no angular broadening relative to pp) and the radiated energy (∆ E ) is found invery low p T particles ( < c ) and at large angles to the jet direction [43]. The lattertwo findings were initially a surprise, but are now incorporated naturally into models wherethe energy is lost in multiple, soft scatterings, and the radiated gluons are emitted at largeangles. The parton then leaves the matter and undergoes normal vacuum fragmentation, i.e.looking like a normal pp jet but with a reduced energy.Additional insight into the energy loss process has come from heavy flavours [44, 45].Like at RHIC, the suppression of charm mesons is virtually identical to the one of inclusivecharged particles, stubbornly refusing to show the difference expected from the strongercoupling (colour charge) of gluons, which are the source of the majority of charged particles,compared to quarks. The mass effect however seems to be as predicted: At intermediate p T ,beauty shows less suppression than charm, whereas at very high p T ( E/m >>
1) b-jets andinclusive jets show similar modifications.
The first discovery made at LHC was announced in Sept. 2010 [46] on a subject which was asunlikely as it was unfamiliar to most in the packed audience: The CMS experiment had founda mysterious ’long range rapidity correlation’ in a tiny subset of extremely high multiplicity pp collisions at 7 TeV [47]. While in the meantime far eclipsed by the discovery of ’a Higgs-likeparticle’, this ’near side ridge’ is arguably still the most unexpected LHC discovery to dateand spawned a large variety of different explanations, from mildly speculative to outrightweird [48]. The most serious contenders are saturation physics, as formulated in the ColorGlass Condensate model (CGC), and collective hydrodynamic flow. Hydrodynamics is ofcourse a very successful framework to describe long range correlations in the macroscopic hotmatter created in heavy ion reactions, but was not supposed to be applicable in small systemslike pp collisions, where typically only a few, or at most a few ten, particles are produced per9nit of rapidity. The CGC is a ’first principles’ classical field theory approximation to QCDwhich is applicable to very dense (high occupation number) parton systems like those foundat small-x and small Q in the initial state wave function of hadrons. It has been successfullyused to describe some regularities seen eg in ep collisions at HERA (’geometric scaling’) andto model the initial conditions in heavy ion physics.Lacking further experimental input, no real progress was made to unravel the origin ofthese long range pp correlations until the ridge made a robust come-back with the first LHCproton-nucleus run some two years later (p-Pb at √ s NN = 5 TeV [49]). The correlationstrength was actually significantly stronger than in pp at the same multiplicity, and in quicksuccession it was discovered that: the ridge was actually double-sided, showing correlationsbetween particles both close by in azimuth as well as back-to-back; a Fourier analysis revealedboth even ( v ) as well as odd ( v ) component; the correlation strength measured with fourparticles was almost identical to the one measured with two, indicating strongly a collectiveor at least multi-particle origin; and finally the dependence of the correlation strength onparticle mass was virtually identical to the one expected from hydrodynamic flow.All characteristics of the p-Pb ridge are very natural for and in good agreement with ahydrodynamic collective flow origin of the correlation. Even the strength of the signal andits multiplicity dependence are of the correct order of magnitude (within a factor of two) ifone uses some reasonable geometrical initial conditions and a standard hydro model and just postulates that the system, some 1 fm in size and lifetime, behaves like a macroscopic idealfluid. Just like the matter created in central Pb-Pb collisions, which is of order 5000 f m and therefore larger by orders of magnitude!The question how such a tiny (few f m ) system could thermalize in essentially no time,maybe even become a small serving of sQGP, has kept the case open, despite what looks likeconsiderable evidence. However, the idea of having small sQGP fireballs created even in pp collisions at the LHC may be less radical than it seems on first sight:At high energy, a proton is qualitatively similar to a small nucleus in some respect;an extended, finite size collection of many (sea)partons which can undergo simultaneousand independent scatterings (called multi-parton interactions in pp , and N coll in AA). Alsothe final state energy- or particle density (particles per unit volume) in high multiplicity pp is comparable to or even larger than in central Pb-Pb. The thermalisation time scalecommonly assumed for the sQGP created in nuclear collisions is significantly less than 1fm/ c [3]; therefore the local volume required for thermalisation is presumably of comparablesize, i.e. at most a few fm ! Once equilibrated, the limit on η/s implies that the meanfree path in the sQGP is compatible with zero or the Compton wavelength. We do findhigher harmonic flow components which originate from density fluctuations with scales oforder 1 fm or less. There is therefore actually quite some direct and indirect evidence that inmatter with initial energy densities like those produced in high multiplicity pp and pA (and,of course, AA ), the 1 fm scale is big enough (and long lived enough) to approach thermalequilibrium and exhibit collective phenomena. The applicability of hydro does not dependon absolute scales (fm or km), but on dimensionless numbers, and e.g. the ratio of systemsize over mean free path is a big number even in a small (but extremely dense) system likehigh multiplicity pp !In any case, the ridge discovery in pp and pA at LHC is definitely more than a curiosityand likely to have profound implications for heavy ion physics, one way or another. If a sQGP10like) state can be created and studied in much smaller systems than anticipated, we can addan ’extra dimension’, namely size, to our toolbox and compare pp, pA , and AA to look forfinite size effects, which may reveal information on correlation lengths and relaxation timescales not otherwise easily available . If, on the contrary, initial state effects and saturationphysics are the answer, we would have discovered at LHC yet another new state of matter,the Colour Glass Condensate, opening a rich new field of activity for both experiment andtheory. As shown in selected examples above, results have come fast and on a wide range of topics forheavy ion physics during the first three years of LHC Run-1: From subtle, as yet to be fullydigested hints (particle ratios) to rather suggestive and clear messages (J/ ψ recombination),the LHC has been shining a new and very instructive light on old problems. Some propertiesof the sQGP have been measured with significantly better precision ( η/s , opacity), improvingalong the way substantially our understanding of the underlying mechanism (jet quenching)or even leading to a paradigm shift (higher harmonic flow) which opened a vast new rangeof observables to precision experiment and precision theory. And if the strong suspicion thatthe surprising long range structures in pp and pA collisions are of collective hydro originturns out to be correct, the ’new state of matter’ will have once more shown that ’ .. it is astate quite different and even more remarkable than had been predicted.’ Some of the main experimental issues which can be addressed with nuclear beams inthe upcoming runs at full LHC energy on the short to medium term are fairly clear basedon the current results: One could hope to largely complete the measurements needed forunderstanding quarkonia production as a deconfinement signal and QGP thermometer. Thisincludes quantifying ’other effects’ from the past (and probably a future, high luminosity) pA run, and reducing the statistical error in particular for the higher mass members of theJ/ ψ and Υ families as well as midrapidity low p T J/ ψ . Precision and sophistication in theflow analysis should further improve over the coming years, on both experimental and theoryfronts, hopefully reaching a precision in e.g. η/s of order 30% or better, at which pointthe result would be precise enough for quantum corrections to the AdS/CFT lower boundto become relevant. While this is definitely a long shot, its worth aiming at because thereare not many alternatives on the horizon for experimental tests of quantum string theory.And last, not least, the ’ridge puzzle’ may be solved quite soon – eg using multi-particlemethods to rigorously (dis)prove collectivity – to decide between initial state (CGC) or finalstate (hydro) origin. If the correlation signal turns out to be the smoking gun for saturationphysics, yet another new state of matter will have been discovered at the LHC; a dense, cold,quasi-classical state of gluon matter. And even if hydro is the answer, the search should go onin pA for any other sign of saturation physics. In particular in the forward direction at verylow Feynman x (right in the LHCb acceptance), the conditions for saturation phenomenashould be just right and proton-nucleus collisions will be the best place to look for themfor some time, until an electron-ion collider comes into operation. If in fact a sensitiveexperimental search for the saturation physics will be carried out at the LHC, the results we also may have to consider changing the name ’heavy ion physics’ γ -jet channel and precision heavy quark mea-surements down to zero momentum. On this time scale one hopefully can also address signalssensitive to chiral symmetry restoration (e.g. low mass lepton pairs); a defining property ofthe QGP which is experimentally extremely challenging and has therefore received compar-atively less attention.The exploration of the phases of strongly interacting matter is one of the four main pillarsof contemporary nuclear physics, and one should see the LHC ion program in this broadercontext: The RHIC program is very active and competitive and continues to map the phasediagram at lower temperatures, in addition looking via an energy scan for the transitionbetween normal matter and the sQGP and a ’tri-critical’ point somewhere in the region ator below SPS fixed target energy. A continuation and strengthening of the SPS fixed targetprogram is under discussion and two new low energy facilities (FAIR at GSI and NICA atJINR) are being built to study compressed matter, i.e. matter at high baryon density and(comparatively) low temperature where the phase structure may be quite different (1 st orderphase transition) and the matter is closer related to neutron stars than to the early universe.The LHC however is and will be the energy frontier facility not only of high energy physics butalso of nuclear physics for the foreseeable future, with a well-defined and extensive programand wish list of measurements. And if the first three years are any guide, strong interactionphysics, while firmly rooted in the Standard Model, shows no end to surprises and discoveriesand promises to keep physics with heavy ions at the LHC interesting (and fun) for quite sometime to come. References [1] B. Muller, J. Schukraft and B. Wyslouch, “First Results from Pb+Pb collisions at theLHC,” Ann. Rev. Nucl. Part. Sci. (2012) 361 [arXiv:1202.3233 [hep-ex]]. [2] J. Schukraft, “Results from the first heavy ion run at the LHC,” J. Phys. Conf. Ser. (2012) 012011 [arXiv:1112.0550 [hep-ex]].[3] B. Muller, “ Investigation of Hot QCD Matter: Theoretical Aspects,” To be publishedin the proceedings of the ’Nobel Symposium on LHC results’,Krusenberg, Sweden, 13 -17 May 2013, in Physica Scripta .[4] J. Schukraft, AIP Conf. Proc. (2000) 3.[5] F. Antinori, A. Billmeier and J. Zaranek, “Introduction to ultrarelativistic heavy-ionphysics at the CERN-SPS,” AIP Conf. Proc. (2003) 294. As far as available, reviews or overview articles – rather than primary literature – are cited where thereader can find further details and references. (2004) 33[hep-ph/0405051].[7] U. Heinz and M. Jacob, preprint nucl-th/0002042.[8] http://press.web.cern.ch/press-releases/2000/02/new-state-matter-created-cern.[9] P. Braun-Munzinger, K. Redlich and J. Stachel, “Particle production in heavy ion col-lisions,” In *Hwa, R.C. (ed.) et al.: Quark gluon plasma* 491-599 [nucl-th/0304013].[10] F. Becattini, “An Introduction to the Statistical Hadronization Model,” arXiv:0901.3643[hep-ph].[11] A. Andronic, P. Braun-Munzinger and J. Stachel, “Thermal hadron production in rela-tivistic nuclear collisions,” Acta Phys. Polon. B (2009) 1005 [arXiv:0901.2909 [nucl-th]].[12] F. Becattini, P. Castorina, A. Milov and H. Satz, “A Comparative analysis of statisticalhadron production,” Eur. Phys. J. C (2010) 377 [arXiv:0911.3026 [hep-ph]].[13] J. Letessier and J. Rafelski, Eur. Phys. J. A (2008) 221 [nucl-th/0504028].[14] H. J. Specht, “Lepton-pair production in nuclear collisions: Past, present, future,” Nucl.Phys. A (2008) 338 [arXiv:0710.5433 [nucl-ex]].[15] R. Rapp, “Dilepton Production in Heavy-Ion Collisions,” arXiv:1306.6394 [nucl-th].[16] H. J. Specht [NA60 Collaboration], AIP Conf. Proc. (2010) 1 [arXiv:1011.0615[nucl-ex]].[17] L. Kluberg and H. Satz, “Color Deconfinement and Charmonium Production in NuclearCollisions,” arXiv:0901.3831 [hep-ph].[18] N. Brambilla, S. Eidelman, B. K. Heltsley, R. Vogt, G. T. Bodwin, E. Eichten,A. D. Frawley and A. B. Meyer et al. , “Heavy quarkonium: progress, puzzles, andopportunities,” Eur. Phys. J. C (2011) 1534 [arXiv:1010.5827 [hep-ph]].[19] R. Rapp, D. Blaschke and P. Crochet, “Charmonium and bottomonium production inheavy-ion collisions,” Prog. Part. Nucl. Phys. (2010) 209 [arXiv:0807.2470 [hep-ph]].[20] The Brahms, Phenix, Phobos, and Star Collaborations, Nucl. Phys. A (2005) 443 [hep-ph/0405125].[23] B. Muller and J. L. Nagle, “Results from the relativistic heavy ion collider,” Ann. Rev.Nucl. Part. Sci. (2006) 93 [nucl-th/0602029].1324] J. Schukraft, Phil. Trans. Roy. Soc. Lond. A (2012) 917 [arXiv:1109.4291 [hep-ex]].[25] R. Stock, “Hadron formation in high energy elementary and nuclear collisions,” Int. J.Mod. Phys. E (2007) 687.[26] A. Andronic, P. Braun-Munzinger and J. Stachel, “Hadron production in centralnucleus-nucleus collisions at chemical freeze-out,” Nucl. Phys. A (2006) 167[nucl-th/0511071].[27] N. Armesto, N. Borghini, S. Jeon, U. A. Wiedemann, S. Abreu, V. Akkelin, J. Alamand J. L. Albacete et al. , “Heavy Ion Collisions at the LHC - Last Call for Predictions,”J. Phys. G (2008) 054001 [arXiv:0711.0974 [hep-ph]].[28] H. Oeschler [ALICE Collaboration], arXiv:1102.2745 [hep-ex].[29] A. Andronic, P. Braun-Munzinger, K. Redlich and J. Stachel, Nucl. Phys. A904-905 (2013) 535c [arXiv:1210.7724 [nucl-th]].[30] F. Becattini, M. Bleicher, T. Kollegger, T. Schuster, J. Steinheimer and R. Stock, Phys.Rev. Lett. (2013) 082302 [arXiv:1212.2431 [nucl-th]].[31] M. Petran, J. Letessier, V. Petracek and J. Rafelski, arXiv:1303.2098 [hep-ph].[32] P. Braun-Munzinger and J. Stachel, arXiv:0901.2500 [nucl-th].[33] B. Abelev et al. [ALICE Collaboration], Phys. Rev. Lett. (2012) 072301[arXiv:1202.1383 [hep-ex]].[34] S. Chatrchyan et al. [CMS Collaboration], Phys. Rev. Lett. (2011) 052302[arXiv:1105.4894 [nucl-ex]].[35] Voloshin SA, Poskanzer AM, Snellings R, “Collective phenomena in non-central nu-clear collisions,” arXiv:0809.2949 [nucl-ex] Landolt-Boernstein: Relativistic Heavy IonPhysics volume 1/23 (Springer-Verlag, 2010), pp. 5–54, 0809.2949.[36] Y. Schutz and U. A. Wiedemann, “Quark matter. Proceedings, 22nd International Con-ference on Ultra-Relativistic Nucleus-Nucleus Collisions, Quark Matter 2011, Annecy,France, May 23-28, 2011,” J. Phys. G (2011) 120301.[37] J. L. Nagle, “Ridge, Bulk, and Medium Response: How to Kill Models and LearnSomething in the Process,” Nucl. Phys. A (2009) 147C [arXiv:0907.2707 [nucl-ex]].[38] G. Aad et al. [ATLAS Collaboration], arXiv:1305.2942 [hep-ex].[39] J. Schukraft, A. Timmins and S. A. Voloshin, Phys. Lett. B (2013) 394[arXiv:1208.4563 [nucl-ex]].[40] Z. Qiu and U. Heinz, Phys. Lett. B (2012) 261 [arXiv:1208.1200 [nucl-th]].1441] A. Majumder and M. Van Leeuwen, “The Theory and Phenomenology of PerturbativeQCD Based Jet Quenching,” Prog. Part. Nucl. Phys. A (2011) 41 [arXiv:1002.2206[hep-ph]].[42] G. Aad et al. [ATLAS Collaboration], Phys. Rev. Lett. (2010) 252303[arXiv:1011.6182 [hep-ex]].[43] S. Chatrchyan et al. [CMS Collaboration], Phys. Rev. C (2011) 024906[arXiv:1102.1957 [nucl-ex]].[44] T. Renk, “Jet quenching and heavy quarks,” arXiv:1309.3059 [hep-ph].[45] A. Dainese, “Heavy-quark production in heavy-ion collisions,” J. Phys. Conf. Ser. (2013) 012034.[46] http://indico.cern.ch/conferenceDisplay.py?confId=107440.[47] V. Khachatryan et al. [CMS Collaboration], JHEP (2010) 091 [arXiv:1009.4122[hep-ex]].[48] W. Li, “Observation of a ’Ridge’ correlation structure in high multiplicity proton-protoncollisions: A brief review,” Mod. Phys. Lett. A27