Stellar astrophysics in the near UV with VLT-CUBES
H. Ernandes, C. J. Evans, B. Barbuy, B. Castilho, G. Cescutti, N. Christlieb, S. Cristiani, G. Cupani, P. Di Marcantonio, M. Franchini, C. Hansen, A. Quirrenbach, R. Smiljanic
SStellar astrophysics in the near UV with VLT-CUBES
H. Ernandes , , , C. J. Evans , B. Barbuy , B. Castilho , G. Cescutti , N. Christlieb , S. Cristiani , G.Cupani , P. Di Marcantonio , M. Franchini ,C. Hansen , A. Quirrenbach , R. Smiljanic Universidade de S˜ao Paulo, IAG, Rua do Mat˜ao 1226, Cidade Universit´aria, S˜ao Paulo, 05508-900, Brazil UK Astronomy Technology Centre, Royal Observatory, Blackford Hill, Edinburgh, EH9 3HJ, UK IfA, University of Edinburgh, Royal Observatory, Blackford Hill, Edinburgh, EH9 3HJ, UK Laborat´orio Nacional de Astrof´ısica/MCTIC, Rua Estados Unidos, 154 - 37504-364, Itajub´a, MG, Brazil INAF - Osservatorio Astronomico di Trieste, via G. B. Tiepolo 11, 34131 Trieste, Italy Landessternwarte, Zentrum f¨ur Astronomie der Universit¨at Heidelberg, K¨oniggstuhl 12, 69117, Heidelberg, Germany Max Planck Institute for Astronomy,Koenigstuhl 17, 69117 Heidelberg Germany Nicolaus Copernicus Astronomical Center, Polish Academy of Sciences, Bartycka 18, 00-716, Warsaw, Poland
ABSTRACT
Alongside future observations with the new European Extremely Large Telescope (ELT), optimised instrumentson the 8-10m generation of telescopes will still be competitive at ‘ground UV’ wavelengths (3000-4000 ˚A). Thenear UV provides a wealth of unique information on the nucleosynthesis of iron-peak elements, molecules, andneutron-capture elements. In the context of development of the near-UV CUBES spectrograph for ESO’s VeryLarge Telescope (VLT), we are investigating the impact of spectral resolution on the ability to estimate chemicalabundances for beryllium and more than 30 iron-peak and heavy elements. From work ahead of the Phase Aconceptual design of CUBES, here we present a comparison of the elements observable at the notional resolvingpower of CUBES ( R ∼ R ∼ Keywords: near-ultraviolet, spectrograph, stellar abundances, VLT
1. INTRODUCTION
The 2020s will see the first operations of the European Extremely Large Telescope (ELT), starting a significantnew era for observational astronomy. With a primary aperture of 39 m and adaptive optics to correct foratmospheric turbulence, the ELT will provide an unprecedented combination of sensitivity and exquisite angularresolution. However, to deliver good performance across a wide wavelength range, four of the five ELT mirrors willbe coated with protected silver, resulting in diminished performance at shorter wavelengths ( < ∼ R ) is a key parameter.One of the leading instruments for stellar abundances at such wavelengths has been the Ultraviolet and VisualEchelle Spectrograph (UVES) on the VLT (Dekker et al. 2000), typically providing R = 40 000 with a 1 (cid:48)(cid:48) slit.Abundances for many elements can potentially be estimated at lower resolution, but we need to investigate theirfeasibility on a line-by-line basis. Here we present a first study of the impact of spectral resolution on a broadrange of elemental diagnostic lines in the near UV (over the range 3020-4000 ˚A). Specifically, we investigate whichare accessible with R = 20 000 from CUBES compared to R = 40 000 with UVES. a r X i v : . [ a s t r o - ph . I M ] F e b . STELLAR NUCLEOSYNTHESIS Understanding the origins of the elements in the periodic table has a prominent role in both astronomy andnuclear physics. Indeed, production of the chemical elements that we are made of and use every day in ourlives is one of the most profound questions we can ask. Each isotope has a complex production channel, whichincludes numerous nuclear reactions.To simplify the problem somewhat, we can group elements together that have common formation mechanisms.For instance, the lightest elements (H, He, Li) were formed in the first minutes after the Big Bang (withsome debate remaining as to whether trace amounts of Be are also primordial). He is also produced via theproton-proton chain in main-sequence stars, whereas Li and Be are produced and destroyed by this process,which depletes them even further than their otherwise low relative abundances (see Fig. 1). The light α -elements typically have spectral lines in the visible and infrared, but elements around the iron-peak (21 ≤ Z ≤ >
30) are endothermic andwould also have to overcome the Coulomb barrier, so these elements are generally not formed by proton capture.Production of such elements therefore occurs via neutron-capture nucleosynthesis, which is described by twomajor mechanisms, the rapid and slow capture of neutrons (r-process and s-process, respectively). The s-processoccurs when the neutron-capture timescale is much lower than that for β -decay ( τ n (cid:29) τ β ), hence this processflows in the the valley of beta stability. The three peaks of the s-process (highlighted by the blue panels inFig. 1) appear due to the bottleneck effect of the magic numbers 50, 82 and 126. The r-process is defined bythe converse, τ n (cid:28) τ β , where neutron capture occurs before nuclei have time to undergo β -decay. Given thesetimescales, the two processes are associated with very different astrophysical environments.To address fundamental questions such as the origins of the heavy elements and their complex nucleosynthesiswe need access to the wealth of information that near-UV spectra contain. For example, the chemical abundancesof metal-poor stars provide us with valuable probes of the nucleosynthesis processes of the first stars and theearly evolution of the Milky Way, but near-UV observations are limited to only small samples (tens of stars)with current instrumentation. Ahead of investigating the required spectral resolution for near-UV spectroscopyand to identify key diagnostic lines, we briefly outline some of the cases motivating new observations. Z A bundan c e Be ScTiVCrMnCoNiCuZn Ge YZrNb MoRuRh PdAg Sn LaCeNd EuGd TbDy ErTm Hf OsIr PbBi Th U
SpeciesElementsCUBES
Figure 1: Abundances of elements in the Solar System vs. atomic number (Z) normalised by the abundance of Si to 10 (Lodders,2003). Iron-peak elements and the three peaks of the s-process are highlighted by the pink and blue panels, respectively. Elementswith near-UV (3020-4000 ˚A) spectral lines and observable with CUBES (at R = 20 000) are indicated in red. .1 Beryllium Although one of the lightest, simplest elements, there remain profound questions regarding the production ofBeryllium in the early Universe. For instance, the recent upper limit for the Be abundance in an extremelymetal-poor star ([Fe/H] = − II resonance lines at 3130.42,3131.06 ˚A, which require good S/N ( (cid:38)
50) and sufficient resolution to clearly discern them from nearby, relativelystrong V II ( λ II ( λ ∼
200 stars with estimated Be abundances(from Keck-HIRES and VLT-UVES), which span near-solar metallicities down to [Fe/H] < − V ∼
12 mag; observations down to at least three magnitudes deeper with CUBES will provide the large homoge-neous samples required, particularly at the metal-poor end given discoveries from ongoing wide-field, multi-bandphotometric surveys (e.g. Pristine, Starkenburg et al. 2017; SkyMapper, Da Costa et al. 2019; Wolf et al. 2018).
The iron-peak group is divided in two: the lower group (21 ≤ Z ≤ ≤ Z ≤
32) which aresynthesized in two processes: α -rich freeze-out and the weak s-process (Woosley, Heger & Weaver, 2002; Limongi& Chieffi, 2003); see also the review by Barbuy et al. (2018) and recent results for Sc, V. Mn, Cu, and Zn inglobular clusters in the Galactic bulge (Ernandes et al. 2018). The near UV is less critical for these elements,but can still provide useful information for species such as Zn I . As previously mentioned, heavy elements (i.e. Z >
30) are produced by two major mechanisms, the r-process ands-process. The s-process is typically associated with stars on the Asymptotic Giant Branch (AGB, e.g. Busso,Gallino & Wasserburg, 1999) and includes the elements highlighted by the blue panels in Fig. 1. The s-processalso occurs in massive stars during their He-burning phase. They mainly produce first-peak s-process elements(Sr, Y and Zr), and have an impact at low metallicity thanks to rapid rotation (Frischknecht et al. 2012, Limongi& Chieffi 2018).The formation channels for the r-process are particularly topical given the detection of the GW170817 kilonovafrom a binary neutron-star merger (Pian et al. 2017; Smartt et al. 2017; Watson et al 2019). The r-processis thought to occur both during the merging and in the milliseconds afterwards (e.g. Bovard et al. 2017), andis thought to play an important role in the chemical evolution of the Galaxy (Matteucci et al. 2014; Cescuttiet al. 2015). Other predicted sites of r-process nucleosynthesis include magnetohydrodynamically-driven jetsfrom core-collapse SNe, resulting from rapidly-rotating massive stars with a strong magnetic field (Winteler etal. 2012; Nishimura, Takiwaki & Thielemann, 2015) and accretion discs in the supernova-triggering collapse ofrapidly-rotating massive stars (or collapsars, Siegel et al. 2019).
Abundances of CNO bring a wealth of information on stellar evolution and the chemical evolution of the Galaxy.In contrast to the atomic transitions of the elements discussed above, CNO features in the near UV are dominatedby a series of molecular bands (see Fig. 2), which include the A-X OH transitions at the shortest wavelengthscan be used to estimate oxygen abundances.Many potential targets in this context will be drawn from the so-called Carbon-enhanced metal-poor (CEMP)stars, which have [C/Fe] > +1.0 (see Beers & Christlieb, 2005). Although rare, they demonstrate a diverse rangeof abundances of neutron-capture elements, commonly grouped as: ‘CEMP-no’ (no over-abundance of r-processelements), ‘CEMP-r’ and ‘CEMP-s’ (stars with over-abundances of r- and s-processed elements, respectively)and ‘CEMP-r/s’ (with apparent contributions from both processes enriching their photospheres). range of scenarios have been explored to investigate these patterns, including rotational mixing in rapidlyrotating, low-metallicity stars (e.g. Chiappini, 2013; Choplin et al. 2016) and supernova models which includeboth mixing and fallback of material to yield the observed abundance ratios (e.g. Umeda & Nomoto, 2002, 2005;Tominaga, Iwamoto & Nomoto, 2014). In short, the CEMP stars are perfect probes to investigate nucleosynthesisfrom the first stars (including production of neutron-capture elements) as well as mass transfer in binary systems(e.g. Abate et al. 2015). However, comprehensive near-UV spectroscopy of CEMP stars to date has been limitedby the sensitivity of current facilities to a few relatively bright targets (e.g. Placco et al. 2015; T. Hansen et al.2015; Hansen et al. 2019). OH NH CNNH
Figure 2:
Red:
Synthetic spectrum of a metal-poor star generated using the turbospectrum radiative transfer code (Plez, 2012),adopting physical parameters as for CS 31082-001 (Cayrel et al. 2001; Hill et al. 2002) including [Fe/H] = − Grey:
Syntheticspectrum of the same star, but now N-rich (∆[N/H] = +2.0 dex).
3. SPECTRAL ANALYSIS3.1 Simulated spectra
To investigate the feasibility of abundance estimates for different elements as a function of spectral resolvingpower, we created small grids of synthetic spectra with the pfant code, using model atmospheres interpolatedfrom the grid of 1D, hydrostatic, LTE marcs models from Gustafsson et al. (2008). Atomic data for thecalculations were taken from the VALD database (Ryabchikova et al. 2015). We calculated two sets of spectrathat will be illustrative of CUBES observations, with effective temperatures (T eff ) and surface gravities (log g )appropriate for a G-type dwarf and a K-type giant, with two metallicities (as traced by the iron abundance,[Fe/H]), as summarised in Table 1. A microturbulence ( v turb ) of 2.0 km s − was adopted in all calculations.Relative to the solar-scaled abundances (defined by [Fe/H]), the abundances for a broad range of elementswith near-UV absorption lines were varied to investigate the feasibility of observations (and responsiveness of thelines to abundance changes). We calculated models for both the dwarf and giant templates, at both metallicitiesand spectral resolving powers, varying the abundances of 39 elements simultaneously by − λ = 0.156 ˚A at R = 20 000, ∆ λ = 0.078 ˚A at R = 40 000) and then binned to mimic the sampling bythe detector, assuming 2.6 pixels per resolution element. To mimic real observations we introduced random noisein each of our models, to give simulated spectra with signal-to-noise (S/N) ratios of 50, 100, and 200 (per pixel).To illustrate the spectral richness of the near UV at the shortest wavelengths, an example section of one of themodel spectra at R = 20 000 (prior to adding noise) is shown in Fig. 3. Examples of specific lines (Ge I , Co I , Ni I ,Y II ) in the simulated spectra of the giant star, varying R at fixed S/N = 100, are shown in Fig. 4. Be, Sc, Ti, V, Cr, Mn, Co, Ni. Cu, Zn, Ge, Y, Zr, Nb, Mo, Ru, Rh, Pd, Ag, Sn, Ba, La, Nd, Sm, Eu, Gd, Tb, Dy, Ho, Er, Tm,Yb, Hf, Os, Ir, Pb, Bi, Th, and U.able 1: Summary of models used to investigate the diagnostic lines.
Parameter Dwarf GiantT eff [K] 5 500 4 500log g [dex] 4.0 2.0 v turb [km s − ] 2.0[Fe/H] − − R
20 000, 40 000
Wavelength ( Å ) A r b r i t r a r y F l u x ZnIGeI ZrII ZrIINbII OsIBiI
Giant [Fe/H]= -3.0
Figure 3: Example section of metal-poor ([Fe/H] = − Each line in Table 2 was visually inspected for all four relative abundances (i.e. − I λ I λ R = 20 000. In terms of diagnostic linescommonly used in studies of neutron-capture elements, the only species that is not feasible is Ba II , but thereare stronger lines available at longer wavelengths that are more commonly used to estimate abundances (Ba II R = 20 000 (cf. 40 000), one might sensiblyask if even lower resolution observations are feasible. We therefore also investigated simulated spectra with R = 10 000 to assess the impact of another factor of two in resolution. As demonstrated by the spectra in Fig. 5,this results in significant loss of information. Many close lines become strongly blended (e.g. the Co I doublet inthe right-hand panel of the figure), and no constraint is possible on Be (left-hand panel). Indeed, at this lowerresolution approximately 80% of the lines in Table 2 are lost as useful abundance diagnostics, with only the mostisolated and strongest lines remaining available. We therefore did not pursue such a low resolution any further. As indicated by the results in Table 3, the majority of the diagnostic lines are accessible at R = 20 000 (providedthere is sufficient S/N). Aside from optimisation of the instrument design, what the above comparisons neglectis the inherent loss in sensitivity of working at higher dispersion, i.e. a fairer comparison for the simulationsshown in Fig. 4 would be to show the resulting spectra for the same integration time. A r b r i t r a r y F l u x GeI
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x GeI
R=40,000 S/N=100 (a) Ge I λ A r b r i t r a r y F l u x CoI
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x CoI
R=40,000 S/N=100 (b) Co I λ A r b r i t r a r y F l u x NiI
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x NiI
R=40,000 S/N=100 (c) Ni I λ A r b r i t r a r y F l u x YII YIITbII
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x YII YIITbII
R=40,000 S/N=100 (d) Y II λ − R = 20 000 and 40 000 (with S/N = 100). Abundancevariations of − A r b r i t r a r y F l u x BeII BeIIZrII
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x BeII BeIIZrII
R=10,000 S/N=100 (a) Be II A r b r i t r a r y F l u x CoI CoI
R=20,000 S/N=100
Wavelength ( Å ) A r b r i t r a r y F l u x CoI CoI
R=10,000 S/N=100 (b) Co I λ R = 10 000 and 20 000 (at S/N = 100, for the [Fe/H] = − − The design philosophy for CUBES is to maximise the end-to-end efficiency of the instrument. It is unlikelythat a similar efficiency could be obtained at the present time for a design at the higher resolving power. In short, R = 20 000 provides a combination of excellent sensitivity with sufficient resolution to undertake quantitativeanalysis of the large majority of the lines considered here.The faintest stars observed with VLT-UVES and Keck-HIRES to date for quantitative analysis in the near UVhave V ∼
12 mag. For example, 2MASS J18082002 − V = 11.93, Schlaufman, Thompson & Casey, 2018)was observed with ten 1 h UVES exposures by Spite et al. (2019), giving S/N ∼
70 near the Be lines. It is clearthat going to fainter magnitudes (and to obtain better S/N) with UVES quickly starts to demand prohibitivelylong exposures/programmes of tens of hours per star. For comparison, using a developmental version of theUBES Exposure Time Calculator (ETC) , observations of a metal-poor dwarf with V = 16 mag should providea S/N = 100 at 3130 ˚A in approximately 3 ×
4. SUMMARY
Near UV spectroscopy enables the study of a diverse range of elements for stellar astrophysics and of the chemicalevolution of the Galaxy. Many of these are uniquely observable in the near UV, such as Be, Bi, and Os. Ourstudy of the elements accessible with observations at R = 20 000 compared to R = 40 000 reveals that nearly allspecies are feasible. In most instances the more dominant factor is S/N rather than resolving power, but theaccuracy owing to blends may be reduced for some lines and requires more detailed simulations (e.g. C. Hansenet al. 2015). Reducing the resolution by a factor of two (to R = 10 000) would render most of the diagnostic linesunusable for abundance analysis.Informed by these results, the conceptual design of CUBES adopted R ∼
20 000 as its baseline. Quantifyingthe tolerances on this specification, including its variation with wavelength and performance for the light-elementmolecular features, is now underway as part of the Phase A study. From initial performance estimates, quantita-tive spectroscopy in the near-UV will be possible to at least three magnitudes deeper than current programmes(in the same exposure time). This will enable exciting new insights in our understanding of nucleosynthesis andthe old stellar populations of the Milky Way.
Acknowledgements:
We acknowledge support from the Global Challenges Research Fund (GCRF) fromUK Research and Innovation (ST/R002630/1). R.S. acknowledges support by the Polish National Science Centrethrough project 2018/31/B/ST9/01469.
Abate et al. 2015, A&A, 581, A62 Lodders, 2003, ApJ, 591, 1220Barbuy et al. 2014, Ap&SS, 354, 191 Matteucci et al. 2014, MNRAS, 438, 2177Barbuy, Trevisan & de Almeida, 2018, PASA, 35, 46 Nishimura, Takiwaki & Thielemann, 2015, ApJ, 810, 109Beers & Christlieb, 2005, ARA&A, 43, 531 Nomoto, Kobayashi & Tominaga, 2013, ARA&A, 51, 457Boesgaard et al. 1999, AJ, 117, 1549 Pian et al. 2017, Nature, 551, 67Boesgaard et al. 2011, ApJ, 743, 140 Placco et at al. 2015, ApJ, 812, 109Bovard et al. 2017, PhysRevD, 96, 124005 Plez, 2012, Astrophysics Source Code Library [ascl:1205.004]Busso, Gallino & Wasserburg, 1999, ARA&A, 37, 239 Primas et al. 2000a, A&A, 362, 666Cayrel et al. 2001, Nature, 409, 691 Primas et al. 2000b, A&A, 364, L4Cescutti et al. 2015, A&A, 577, A139 Ryabchikova et al. 2015, Physica Scripta, 90, 054005Chiappini, 2013, AN, 334, 595 Schlaufman, Thompson & Casey, 2018, ApJ, 867, 98Choplin et al. 2016, A&A, 593, A36 Siegel et al. 2019, Nature, 569, 241Da Costa et al. 2019, MNRAS 489, 5900 Smartt et al. 2017, Nature, 551, 75Dekker et al. 2000, Proc. SPIE, 4008, 534 Smiljanic et al. 2009, A&A, 499, 103Ernandes et al. 2018, A&A, 616, A18 Smiljanic et al. 2014, Ap&SS, 354, 55Evans et al. 2016, Proc. SPIE, 9908, 9J Sneden et al. 2003, ApJ, 591, 936Evans et al. 2018, Proc. SPIE, 10702, 2E Spite et al. 2019, A&A, 624, A44Frischknecht et al. 2012, A&A, 538, 2 Starkenburg et al. 2017, MNRAS, 471, 2587Gustafsson et al. 2008, A&A, 486, 951 Tominaga, Iwamoto & Nomoto, 2014, ApJ, 785, 98Hansen, C. et al. 2015, AN, 336, 665 Umeda & Nomoto, 2002, ApJ, 565, 385Hansen, T. et al. 2015, ApJ, 807, 173 Umeda & Nomoto, 2005, ApJ, 619, 427Hansen et al. 2019, A&A, 623, 128 Watson et al. 2019, Nature, 574, 497Hill et al. 2002, A&A, 387, 560 Winteler et al. 2012, ApJ, 750, L22Limongi & Chieffi, 2003, ApJ, 592, 404 Wolf et al. 2018, PASA, 35, 10Limongi & Chieffi, 2018, ApJ, 237, 13 Woosley, Heger & Weaver, 2002, RvMP, 74, 1015 http://archives.ia2.inaf.it/cubes/ PPENDIX A. SUPPORTING MATERIAL -3.0 20k -3.0 40k -1.0 20k -1.0 40k [Fe/H]
BeII 3130.42BeII 3131.07ScII 3576.34ScII 3590.47TiI 3998.64TiII 3321.7TiII 3343.76TiII 3491.05VII 3951.96CrI 3578.68MnII 3441.99MnII 3460.32MnII 3482.9MnII 3488.68MnII 3495.83MnII 3497.53CoI 3412.34CoI 3412.63CoI 3449.16CoI 3529.03CoI 3842.05CoI 3845.47NiI 3437.28NiI 3483.77NiI 3500.85NiI 3597.71NiI 3807.14CuI 3247.53CuI 3273.95ZnI 3075.9ZnI 3302.58ZnI 3345.01GeI 3039.07YII 3549.01YII 3584.52YII 3600.74YII 3601.91YII 3611.04YII 3774.33ZrII 3054.84ZrII 3095.07ZrII 3125.92ZrII 3129.76ZrII 3273.05ZrII 3279.26ZrII 3284.71ZrII 3305.15ZrII 3334.62ZrII 3344.79ZrII 3356.09ZrII 3357.26ZrII 3404.83ZrII 3408.08ZrII 3410.24ZrII 3430.53ZrII 3457.56ZrII 3458.93ZrII 3479.02ZrII 3479.39ZrII 3481.15ZrII 3496.2ZrII 3505.67ZrII 3506.05ZrII 3525.81ZrII 3549.51ZrII 3551.95ZrII 3556.59ZrII 3588.31ZrII 3607.37ZrII 3751.59ZrII 3766.82ZrII 3836.76ZrII 3998.96 L i ne s Giant BG (a) Lines from Be II to Zr II . -3.0 20k -3.0 40k -1.0 20k -1.0 40k [Fe/H] NbII 3028.44NbII 3215.59NbII 3225.47MoI 3864.1RuI 3436.74RuI 3498.94RuI 3742.28RuI 3798.9RuI 3799.35RhI 3396.82RhI 3434.89RhI 3692.36RhI 3700.91PdI 3242.7PdI 3404.58PdI 3516.94AgI 3280.68AgI 3382.9SnI 3801.01BaII 3891.78LaII 3794.77LaII 3949.1LaII 3988.51LaII 3995.74CeII 3999.24NdII 3784.24NdII 3810.48NdII 3826.41NdII 3838.98SmII 3568.27SmII 3796.75SmII 3896.97EuII 3724.93EuII 3819.67EuII 3907.11GdII 3549.36GdII 3557.06GdII 3712.7GdII 3768.4TbII 3600.41TbII 3702.85DyII 3531.71DyII 3536.02DyII 3550.22DyII 3563.15DyII 3694.81DyII 3757.37DyII 3944.68DyII 3996.69HoII 3466.01HoII 3890.65ErII 3692.65ErII 3729.52ErII 3786.84ErII 3830.48ErII 3896.23ErII 3906.31TmII 3701.36TmII 3795.76TmII 3848.02YbII 3694.2HfII 3276.85HfII 3399.79HfII 3719.28OsI 3058.66IrI 3220.78IrI 3800.12PbI 3683.46BiI 3024.64ThII 3351.23ThII 3433.99ThII 3435.98ThII 3469.92ThII 3539.59ThII 3675.57UII 3859.57 L i ne s Giant BG (b) Lines from Nb II to U II .Figure 6: Detectable lines in the simulated K-type giant spectra with [Fe/H] = − − R = 20,000 and 40,000. Darkershaded boxes indicate where it is not possible to discern differences between abundance variations of − [Fe/H] BeII 3130.42BeII 3131.07ScII 3576.34ScII 3590.47TiI 3998.64TiII 3321.7TiII 3343.76TiII 3491.05VII 3951.96CrI 3578.68MnII 3441.99MnII 3460.32MnII 3482.9MnII 3488.68MnII 3495.83MnII 3497.53CoI 3412.34CoI 3412.63CoI 3449.16CoI 3529.03CoI 3842.05CoI 3845.47NiI 3437.28NiI 3483.77NiI 3500.85NiI 3597.71NiI 3807.14CuI 3247.53CuI 3273.95ZnI 3075.9ZnI 3302.58ZnI 3345.01GeI 3039.07YII 3549.01YII 3584.52YII 3600.74YII 3601.91YII 3611.04YII 3774.33ZrII 3054.84ZrII 3095.07ZrII 3125.92ZrII 3129.76ZrII 3273.05ZrII 3279.26ZrII 3284.71ZrII 3305.15ZrII 3334.62ZrII 3344.79ZrII 3356.09ZrII 3357.26ZrII 3404.83ZrII 3408.08ZrII 3410.24ZrII 3430.53ZrII 3457.56ZrII 3458.93ZrII 3479.02ZrII 3479.39ZrII 3481.15ZrII 3496.2ZrII 3505.67ZrII 3506.05ZrII 3525.81ZrII 3549.51ZrII 3551.95ZrII 3556.59ZrII 3588.31ZrII 3607.37ZrII 3751.59ZrII 3766.82ZrII 3836.76 L i ne s Dwarf BG (a) Lines from Be II to Zr II . -3.0 20k -3.0 40k -1.0 20k -1.0 40k [Fe/H] NbII 3028.44NbII 3215.59NbII 3225.47MoI 3864.1RuI 3436.74RuI 3498.94RuI 3742.28RuI 3798.9RuI 3799.35RhI 3396.82RhI 3434.89RhI 3692.36RhI 3700.91PdI 3242.7PdI 3404.58PdI 3516.94AgI 3280.68AgI 3382.9SnI 3801.01BaII 3891.78LaII 3794.77LaII 3949.1LaII 3988.51LaII 3995.74CeII 3999.24NdII 3784.24NdII 3810.48NdII 3826.41NdII 3838.98SmII 3568.27SmII 3796.75SmII 3896.97EuII 3724.93EuII 3819.67EuII 3907.11GdII 3549.36GdII 3557.06GdII 3712.7GdII 3768.4TbII 3600.41TbII 3702.85DyII 3531.71DyII 3536.02DyII 3550.22DyII 3563.15DyII 3694.81DyII 3757.37DyII 3944.68DyII 3996.69HoII 3466.01HoII 3890.65ErII 3692.65ErII 3729.52ErII 3786.84ErII 3830.48ErII 3896.23ErII 3906.31TmII 3701.36TmII 3795.76TmII 3848.02YbII 3694.2HfII 3276.85HfII 3399.79HfII 3719.28OsI 3058.66IrI 3220.78IrI 3800.12PbI 3683.46BiI 3024.64ThII 3351.23ThII 3433.99ThII 3435.98ThII 3469.92ThII 3539.59ThII 3675.57UII 3859.57 L i ne s Dwarf BG (b) Lines from Nb II to U II .Figure 7: Detectable lines in the simulated G-type dwarf spectra with [Fe/H] = − − R = 20,000 and 40,000. Darkershaded boxes indicate where it is not possible to discern differences between abundance variations of − Ion Wavelength (˚A) Ion Wavelength (˚A) Ion Wavelength (˚A)Be II II II II II II II II II II II II I II II II II II II II II II II II II II II I II II II II II II II II II II II II II II II II II II II II I II II I II II I II II I II II I II II I II II I II II I II II I II II I II II I II II I II II I II II I I II I I II I I II I I II II I II II I II II I II II I II II I II II I II II I I II I I II I I II I I II I I II I II II II II II II II II II II II II II II II II II II II II II II II able 3: Number of our selected lines that are detectable for each ion in the CUBES wavelength range for both a G-type dwarf anda K-type giant with [Fe/H] = − − R ∼