Measurement of the $^{70}$Ge($n,\gamma$) cross section up to 300~keV at the CERN n_TOF facility

Neutron capture data on intermediate mass nuclei are of key importance to nucleosynthesis in the weak component of the slow neutron capture processes, which occurs in massive stars. The ($n,\gamma$) cross section on $^{70}$Ge, which is mainly produced in the $s$~process, was measured at the neutron time-of-flight facility n_TOF at CERN. Resonance capture kernels were determined up to 40~keV neutron energy, and average cross sections up to 300~keV. Stellar cross sections were calculated from $kT=5$~keV to $kT=100$ keV and are in very good agreement with a previous measurement by Walter and Beer (1985), and recent evaluations. Average cross sections are in agreement with Walter and Beer (1985) over most of the neutron energy range covered, while being systematically smaller for neutron energies above 150~keV. We have calculated isotopic abundances produced in $s$-process environments in a 25 solar mass star for two initial metallicities (below solar, and close to solar). While the low metallicity model reproduces best the solar system germanium isotopic abundances the close to solar model shows a good global match to solar system abundances between mass numbers A=60-80.

Neutron capture data on intermediate mass nuclei are of key importance to nucleosynthesis in the weak component of the slow neutron capture processes, which occurs in massive stars. The (n, γ) cross section on 70 Ge, which is mainly produced in the s process, was measured at the neutron time-of-flight facility n TOF at CERN. Resonance capture kernels were determined up to 40 keV neutron energy, and average cross sections up to 300 keV. Stellar cross sections were calculated from kT = 5 keV to kT = 100 keV and are in very good agreement with a previous measurement by Walter and Beer (1985), and recent evaluations. Average cross sections are in agreement with Walter and Beer (1985) over most of the neutron energy range covered, while being systematically smaller for neutron energies above 150 keV. We have calculated isotopic abundances produced in s-process environments in a 25 solar mass star for two initial metallicities (below solar, and close to solar). While the low metallicity model reproduces best the solar system germanium isotopic abundances the close to solar model shows a good global match to solar system abundances between mass numbers A=60-80.

I. MOTIVATION
The elemental abundances above Fe are mainly produced by two distinct neutron capture processes in stars and stellar explosions, the slow neutron capture process (s process) [1] and the rapid neutron capture process (r process) [2], while only about 1% of heavy element abundances is produced by charged particle and photoinduced reactions (p process) [3]. The s process occurs at moderate neutron densities of typically 10 7 − 10 12 cm −3 , and is characterised by a sequence of neutron captures followed by β-decays [1]. Since radioactive decays are usually faster than neutron capture times, the reaction path proceeds along the line of stable nuclei. In massive stars (> 8 solar masses M ⊙ ), the s process occurs in two different burning stages, towards the end of He core burning at temperatures of about 0.3 GK (1 GK=10 9 K), and later during carbon shell burning, when temperatures reach 1 GK [4][5][6][7][8]. This so called weak component of the s process, produces dominantly elements between Fe and Zr, owed to the relatively low neutron exposures reached. The main component of the s process contributes dominantly abundances between Zr and Bi, and occurs in low and intermediate mass stars (about 1-5 M ⊙ ) during their Asymptotic Giant Branch (AGB) phase [1,9]. In our cosmos, germanium is produced by more than one nucleosynthesis process, however, the bulk of it (about 80%) is thought to be produced by the weak s-process in massive stars [10]. Around 12% are estimated to originate from the main s-process in AGB stars, while only a few percent are produced by explosive nucleosynthesis * corresponding author: claudia.lederer-woods@ed.ac.uk processes, such as the r process, operating at very high neutron densities, and the alpha-rich freeze out forming the Fe group nuclei at the end of a massive star's life. To disentangle the different contributions to observed germanium abundances in our solar system and stellar atmospheres, it is crucial to accurately determine the s process contribution to germanium abundances. Figure 1 shows the s-process reaction path in the germanium mass region. 70 Ge occupies a special position: It is shielded from rapid neutron capture nucleosynthesis by the stable isobar 70 Zn, suggesting that 70 Ge is dominantly produced in the s process (so-called s-only isotope), and its abundance thus can be used as an anchor point when comparing to observed isotopic abundances. Neutron capture cross sections are a key nuclear physics input to calculate abundances produced in s-process nucleosynthesis. Since neutrons are rapidly thermalised in the stellar environment, the effective stellar cross section is a convolution of the energy dependent neutron capture cross section with a Maxwellian velocity distribution, thus, stellar neutron capture cross sections are called Maxwellian Averaged Cross Sections (MACS). For s-process environments, MACS values need to be known up to kT values of about 90 keV, corresponding to the highest temperatures reached in the s process. Investigations using Monte Carlo variation of reaction rates to assess the impact of specific reactions on the final abundances predicted by stellar models have been performed recently. The reaction 70 Ge(n, γ) has been identified as a key reaction determining the uncertainty of 70 Ge in the main s-process, both in the thermal pulse and in interpulse burning of AGB stars [11]. The reaction 70 Ge(n, γ) and the recently measured 73 Ge(n, γ) [12] were found to be key reactions for the abundances of 70 Ge and 73 Ge, respectively, in the enhanced weak s-process occurring in rotating, metal-poor massive stars [13]. 73 Ge(n, γ) also appeared as key rate in the regular weak s-process in the same study. Existing experimental data on stellar neutron capture on germanium isotopes are scarce, which motivated a campaign at n TOF to measure cross sections on all stable germanium isotopes [14]. Results on 73 Ge(n, γ) have recently been published and show that the germanium isotopic abundance pattern produced in the s process in a low metallicity massive star is consistent with the isotopic abundance pattern of our solar system [12]. This manuscript describes the study of the 70 Ge(n, γ) reaction. There are few experimental data available on this reaction in the keV neutron energy range; Walter and Beer [15] measured cross sections from 3 to 240 keV and calculated MACSs between kT = 20 − 50 keV. In addition, Maletski et al. [16] have published neutron resonance data on 70 Ge+n reactions for neutron energies up to 28.6 keV, however, partial radiative widths Γ γ have only been determined for three low energy resonances, while for the remaining resonances only neutron widths Γ n and resonance energies E R are available. Harvey and Hockaday [17] have determined total cross sections for natural germanium, identifying several 70 Ge+n resonances up to 43 keV neutron energy. Evaluated 70 Ge(n, γ) cross sections recommended in major nuclear data libraries such as ENDF/B-VIII [18] and JENDL-4.0 [19] use experimental data by Maletski at al. and transmission data on natural germanium by Harvey and Hockaday for neutron energies below 14 keV, while for higher neutron energies evaluated cross sections are based on statistical parameters extracted from data at lower neutron energies. This work provides neutron capture resonance data up to energies of 40 keV. In combination with average neutron capture cross sections determined for neutron energies up to 300 keV, we also calculated Maxwellian averaged cross sections for the entire range of astrophysical interest, from kT = 5 keV, to kT = 100 keV.

II. MEASUREMENT
The measurement was performed at the neutron timeof-flight facility n TOF at CERN [20]. At n TOF, neutrons are produced by spallation reactions of a 20 GeV/c proton beam by the CERN-PS on a cylindrical 1.3 tonne lead target (40 cm length, 60 cm diameter) [21]. The initially highly energetic neutrons are moderated using a combination of water and borated water layers surrounding the target (the former also cools the target). The resulting moderated neutron spectrum exhibits a nearly isolethargic energy dependence and ranges from thermal energies (25 meV) to several GeV. The proton beam is pulsed, with a time width of 7 ns r.m.s., and a repetition rate of about 0.8 Hz. At n TOF, there are two experimental areas at flight path lengths of 185 m (EAR-1) and 20 m (EAR-2). The 70 Ge(n, γ) measurement was performed at EAR-1, taking advantage of the excellent neutron energy resolution ranging from 3 × 10 −4 at 1 eV to 3 × 10 −3 at 100 keV [21]. The prompt emission of γ rays following neutron capture was detected with a set of four liquid scintillation detectors, filled with deuterated benzene (C 6 D 6 ). These detectors are optimal for measurement of radiative neutron capture since they are almost insensitive to neutrons which are scattered from the capture sample. The detectors were arranged at angles of 125 degrees with respect to the neutron beam to minimise effects of anisotropic γ-ray emission for ℓ > 0 states in the compound nucleus. The 70 Ge(n, γ) reaction was studied using a highly enriched 70 GeO 2 sample in cylindrical form. In addition, measurements were taken with a metallic germanium sample of natural composition to identify any contributions from other Ge isotopes present in the enriched sample, and with a Au sample which was used to normalise the neutron capture data (see Sec. III). The samples were glued on a thin mylar foil fixed to an Al ring. The background induced by the sample holder was measured by placing an empty sample holder in the beam. To keep any systematic effects due to sample positioning to a minimum, the samples were accurately centred on the holder using a jig and a hollow metallic cylinder aligned with the annular frame of the sample holder. Table I lists all samples and their characteristics used in the measurement.
Detector signals were recorded using 14-bit fast digitizers operated at a sampling rate of 1 GHz. Data acquisition was triggered by the pickup signal of the PSaccelerator and for each neutron pulse data were recorded for a duration of 100 ms corresponding to neutron energies of about 0.02 eV. Detector signal times and amplitudes were extracted for each neutron burst using an off-line pulse shape algorithm [22].

III. DATA ANALYSIS AND RESULTS
A. Time-of-flight to neutron energy conversion The neutron time-of-flight spectra were converted to neutron energy using the relativistic relation where m n is the mass of the neutron, L is the flight path length, and t f is the neutron time-of-flight. The determination of t f relies on an accurate knowledge of the neutron production time. When the proton pulse hits the spallation target, a burst of γ-rays and ultra-relativistic particles is produced alongside neutrons. This burst is called γ-flash, and produces a short, high amplitude signal in the C 6 D 6 detectors, allowing determination of the time of neutron production for each pulse with high accuracy. The flight path length L was determined using a Au sample in the beam. Au has several low energy resonances for which resonance energies have been accurately determined at the Joint Research Centre in Geel [23]. These are now included in the latest evaluated resonance data for 197 Au(n, γ) by ENDF/B-VIII [18]. The flight path has been fitted to reproduce the resonance energies for these well known resonances and determined as 183.94 ± 0.04 m.

B. Capture Yield
The neutron capture yield as a function of neutron energy is obtained as where C is the count spectrum of the enriched 70 Ge sample, B is the background, ǫ c is the efficiency to detect a capture event, and Φ is the neutron flux. The factor f N is a normalisation factor taking into account that the capture sample does not cover the entire size of the neutron beam. In the following sub-sections, the determination of each of those terms will be described.

Detection efficiency
The efficiency to detect the γ-ray emission after a neutron capture event depends on the de-excitation path of the compound nucleus, which in general varies from event to event. The total energy detection principle was used by combining the above-mentioned detection system with the Pulse Height Weighting Technique (PHWT) [24,25]. The total energy detection principle is applicable to detection systems of low efficiency, where at most one γ ray per capture event is detected. If the γ-ray detection efficiency is proportional to the γ-ray energy, i.e. ǫ γ ∝ E γ , it can be shown that the efficiency to detect a capture event is proportional to the excitation energy of the compound system, i.e. ǫ c ∝ S n +E cm , where S n is the neutron separation energy, and E cm is the center-of-mass energy. To achieve ǫ γ ∝ E γ , the PHWT is used, where weighting factors are applied to each detected event, depending on their amplitude. These were determined in GEANT4 Monte Carlo simulations [26], simulating the detector response to mono-energetic γ-rays from 0.2 to 10.0 MeV. Simulations took into account the detailed geometry of the experimental setup and samples used, including the spatial distribution of the neutron beam, and neutron and γ-ray self absorption in the samples. The data further need to be corrected for a loss of γ-rays below the detector threshold (200 keV), and for transitions without γ-ray emission (electron conversion). The correction for both Au and Ge was calculated using the code dicebox [27], that simulates individual capture cascades based on experimental information on low-lying levels and uses level density and photon strength function models at higher energies. The systematic uncertainty of the neutron capture yield due to the PHWT is 2% [28].

Background Subtraction
The background B(E n ) consists of several components that need to be corrected for. Background unrelated to the neutron beam, due to ambient radioactivity and cosmic rays is determined in beam off runs and subtracted from the Au and Ge sample spectra. Background produced by neutron reactions with material in and around the beam line is measured by taking data with an empty sample holder. Figure 2 shows the 70 Ge spectrum, compared to backgrounds due to the empty sample holder, and due to am-bient activity. Weighting functions have been applied to all these spectra, hence the Figure shows weighted counts as a function of neutron energy. The empty sample holder contributes to background over the entire neutron energy range, while the ambient activity, which is constant in time, only contributes at low neutron energies. The third component is background related to the sample, specifically by neutrons scattered off the GeO 2 sample. These scattered neutrons may be captured in surrounding material and produce γ-ray cascades detected by the C 6 D 6 system. The C 6 D 6 detectors themselves are extremely insensitive to capturing neutrons [29,30], and therefore any background due to direct interaction of scattered neutrons with the detectors themselves can be neglected for the present measurement. While care has been taken to minimize material around the beam pipe, and using material with low neutron cross sections, such as Al, there may be residual background due to neutrons scattered from the sample and being captured at a later time in structural material, such as the walls [31]. For the resonance region, individual neutron resonances were fitted including a flat, constant background to account for these backgrounds. For the unresolved resonance region, the background is determined using neutron filters. Neutron filters are materials in the beam line, which absorb all neutrons at certain resonance energies. In the present case, an Al filter was used, with negligible transmission of neutrons at around 35, 86 and 150 keV neutron energy. Consequently, any counts in these filter dips must come from neutrons scattered off the sample and captured somewhere in the experimental area at a later time. This background was determined by measuring a Ge spectrum with filters, and an empty sample holder spectrum with filters. After subtraction of the empty spectrum, the Ge+ Al filter spectrum was scaled to the Ge spectrum to account for the overall loss of neutrons. A smooth function was fitted to the flat minima of the filter dips and the resulting background was subtracted from the Ge spectrum. Figure 3 shows the 70 Ge+filter spectrum and the fitted background. The background was found to be at a level of 10-15% of the Ge spectrum. Due to uncertainties in the scaling factor applied to the Ge+filter spectrum, and low statistics in the filter dips, the resulting uncertainty of the capture yield in the unresolved resonance region is 4%.

Neutron Flux and Normalisation
The neutron flux was measured using several different detection systems and reference reactions in a dedicated campaign. Measurements were performed with a set of silicon detectors, detecting tritons and alpha particles emitted in the 6 Li(n, t) reaction from a LiF sample. A Micromegas detector was used to determine the neutron flux using the reference reactions 6 Li(n, t) and 235 U(n, f ), and an ionisation chamber by Physikalisch Technische Bundesanstalt Braunschweig measured (n, f ) reactions on 235 U. The data from all three detection systems were combined to determine an accurate neutron flux, which has a systematic uncertainty below 1% for neutron energies < 3 keV, and of 3.5% between 3 keV and 1 MeV [32]. More details on the neutron flux evaluation at n TOF can be found in Ref. [33]. A normalisation factor f N needs to be applied since the neutron beam is larger than the capture sample. In addition, such a factor corrects for any inaccuracies in determination of the solid angle covered by the C 6 D 6 detectors. The saturated resonance technique [34] was applied to determine the normalisation factor with high accuracy. In the present case, data were normalised to the well known 4.9-eV resonance in the 197 Au(n, γ) reaction. For this resonance, the capture cross section is much larger than neutron scattering (Γ γ /Γ n ≈ 8 [18]). If the Au sample is chosen to be sufficiently thick, all neutrons at the resonance energy will be absorbed by the sample. In addition, due to the high ratio of capture/scattering almost all neutrons will be radiatively captured within the sample, even if their first interaction is elastic scattering. Thus, almost 100% of neutrons passing the capture sample produce a Au neutron capture cascade, providing an absolute normalisation point for the neutron capture yield at 4.9 eV neutron energy. The size of the neutron beam has a slight dependence on neutron energy. This dependence has been determined in Monte Carlo simulations of the neutron transport from the spallation target to the experimental area, and has been verified in beam profile measurements [21]. In the region of interest, the change in neutron beam size is at most 1.5%. This small correction was applied to the data. As the saturated resonance technique is insensitive to the precise individual resonance parameters for the 4.9-eV resonance, the systematic uncertainty, including possible small errors in sample positioning, is 1%.

C. Resonance Analysis
Neutron resonances in the capture yield were fitted with the multi-level, multi-channel R-matrix code SAMMY [35]. SAMMY takes into account all experimental effects, such as multiple interaction events (multiple scattering) and self shielding, and the broadening of resonances due to thermal motion (Doppler Broadening) and the resolution of the experimental setup. In addition, the full sample composition, including impurities was taken into account. Capture data themselves do not usually allow a reliable determination of individual resonance parameters, such as resonance spin J and partial neutron and radiative widths, Γ n and Γ γ . In general, only energy and capture kernel k, defined as can be obtained reliably. The statistical factor g is given by g = (2J + 1) (2s + 1)(2I + 1) where J is the resonance spin, the neutron spin s = 1/2, and the ground state spin of the target nucleus I( 70 Ge) = 0 + , hence in our case g = (2J + 1)/2. Resonance structures could be resolved up to neutron energies of 40 keV. However, from about 25 keV the analysis of individual resonance parameters became increasingly difficult, due to the worsening of the experimental resolution combined with lower statistics and increasing natural resonance widths. For this reason, an averaged cross section was determined from 25 keV onwards (see Sec. III D). Resonance energies and capture kernels k are shown in Tables II and III for neutron resonance energies below, and above 25 keV, respectively. Examples of resonance fits in various neutron energy ranges are shown in Fig.  4. Uncertainties in Tables II and III are fit uncertainties only.
Systematic uncertainties in the capture kernels are due to the PHWT (2%), the normalisation (1%), the neutron flux (1% for E n < 3 keV, 3.5% for E n > 3 keV), and the sample enrichment (1%). This amounts to total systematic uncertainties of 2.7% below, and 4.3% above 3 keV neutron energy, respectively. In total, we fitted 110 resonances up to energies of 40 keV, of which 90 were not listed in previous evaluations.
Average resonance parameters, namely the average radiative width Γ γ and the average resonance spacing D 0 , were determined using the resonances below 25 keV assuming there are no unresolved doublets. There are 35 strong resonances for which the SAMMY fit yielded Γ n > 10 × Γ γ . As k ≈ gΓ γ for these resonances, we used the kernels determined from n TOF data and the spins from a transmission measurement [17] to estimate the distribution of individual Γ γ values in terms of the average radiative width Γ γ and the width of the distribution σ Γγ . Using the same method as in Ref. [12], namely the maximum likelihood fit assuming a gaussian distribution of Γ γ values, we obtained Γ γ = 205(12) meV and σ Γγ = 70 (10). Our value of the average radiative width is compatible with the literature values 185(30) meV [36] and 185(50) meV [37].
To determine D 0 we have adopted the neutron strength functions for s-, p-and d-wave neutrons S 0 = 2.1 × 10 −4 , S 1 = 1.5 × 10 −4 and S 2 = 3 × 10 −4 , and channel radius of 7 fm from Ref. [36]. The numbers of observed resonances above certain thresholds in the resonance kernel were compared to the simulated resonance sequences assuming the validity of the statistical model as in Ref. [38]. Our data are perfectly consistent with D 0 = 830(100) eV, this value is compatible within 2σ with the literature value of 1170(230) eV [36,37].
Present values of D 0 and Γ γ in combination with adopted S 0 and S 1 imply that the neutron width of the strongest p-wave resonances can reach the values of the strongest s-wave resonances already at E n of a few keV. The determination of S 0 and S 1 without firm ℓ assignment is thus impossible, we can merely state that the literature values are reasonable. For example, simulations with a half value of S 1 during the procedure of obtaining the D 0 yielded inconsistent D 0 values for different adopted thresholds.

D. Unresolved Resonance Region
The threshold for observing individual resonances increases with energy, and from about 25 keV a nonnegligible fraction of smaller resonances is missed, which would result in an underestimation of the cross section. Therefore, we determined an averaged neutron capture cross section between 25 keV and 300 keV from the neutron capture yield. The capture yield was corrected for multiple scattering and self shielding effects using a Monte Carlo code which takes into account the sample composition and geometry, as well as neutron scattering and capture cross sections. These corrections to the yield amounted to 6-7%. A conservative estimate of 20% uncertainty in these corrections results in a 1.4% uncertainty of the corrected neutron capture yield.
The unresolved cross section was obtained after subtraction of backgrounds due to neutron scattering (see Sec. III B 2), and contributions due to impurities. In the present case, the main impurity in the sample is 72 Ge with a content of 2.23%. Other germanium isotopes, as well as oxygen in the sample have negligible contributions to the measured capture yield, either because of their low abundance in the sample, or because of their small cross sections. We used ENDF/B-VIII [18] cross sections to estimate the contribution of the 72 Ge impurity to the measured cross section yielding a background at a level of 1-2%. At present there is no experimental information on the 72 Ge(n, γ) cross section in the relevant energy range and the cross section recommended by the ENDF/B-VIII evaluation is based on average resonance parameters obtained at lower neutron energies by Maletski et al. [16]. Assuming a 20% uncertainty in the Ge cross section, the contribution to the uncertainty in the 70 Ge(n, γ) cross section is at most 0.4%. Total systematic uncertainties of the unresolved cross section are 6%, consisting of 4.3% systematic uncertainty as outlined in Sec. III C, the background subtraction with neutron filters (4%), correction for sample impurities (0.4%), and multiple scattering and self-shielding corrections (1.4%). Figure 5 shows a plot of the unresolved cross section obtained in this work, compared to experimental results by Walter and Beer [15], and evaluated cross sections published in the ENDF/B-VIII library. The cross sections determined in this work are in fair agreement with Walter and Beer, as well as ENDF/B-VIII over most of the neutron energy range covered, while being systematically smaller than ENDF/B-VIII and Walter and Beer above 150 keV.

IV. STELLAR CROSS SECTIONS AND ASTROPHYSICAL IMPLICATIONS
Maxwellian Averaged Cross Sections were determined by combining the cross section obtained from resonance parameters up to 25 keV neutron energy (resolved resonance region RRR), with the unresolved averaged cross section for neutron energies between 25 and 300 keV (unresolved resonance region URR). From 300 keV, we used cross sections recommended by the ENDF/B-VIII evaluation, but scaled by a factor 0.8 to take into account that the new experimental data indicate an overestimation of the cross section at higher neutron energies of 10-35%.  [15] and the ENDF/B-VIII evaluation [18].
MACS values were calculated for kT values between 5 and 100 keV using the formula Uncertainties taking into account all three components were determined in the following way: The common (absolute) uncertainty for RRR + URR MACSs σ RRR+URR is due to the weighting procedure (2%), the sample enrichment (1%), the neutron flux (3.5%), normalisation (1%), and counting statistics including uncertainties of the resonance fits (1%). An additional uncertainty just applicable to the URR, σ URR , comes from the background subtraction using filters (4%), self shielding and multiple scattering (1.4%), and subtraction of the 72 Ge background (0.4%). We have assigned a 20% uncertainty to the scaled ENDF/B-VIII cross sections (σ ENDF ) that were used for E n > 300 keV. The contribution of cross sections above 300 keV to the MACS is negligible for kT values below 50 keV, and between 2% and 11% for kT = 60 − 100 keV. The total uncertainty of the MACS was then determined as Hence, total uncertainties for MACSs values vary from 4.4% at kT = 5 keV to 5.6% at kT = 100 keV. Table IV lists MACS values determined in this work from kT = 5 to 100 keV. MACSs in this work are compared to recommended values in the Karlsruhe Astrophysical Database of Nucleosynthesis in Stars (KADoNiS) version 1.0 [39], which is widely used as reference for reaction rates in astrophysical calculations, and experimental MACSs determined by Walter and Beer [15]. Agreement with Walter and Beer values is very good, and there is also good agreement with Kadonis-1.0, considering uncertainties. The trend of the MACS values with kT is flatter for Walter and Beer, which can be explained by their higher averaged cross section at high neutron energies. Because 70 Ge is an even-even nucleus, the contribution of reactions on excited target states in the stellar plasma is negligible up to plasma temperatures of about 1.5 GK, which is well above the s-process temperature range [40,41]. Therefore the experimentally determined MACS can directly be used as stellar MACS in astrophysical calculations for the s process.
We have calculated s process nucleosynthesis abundances for a massive 25 M ⊙ star for two initial metallicities, below solar (z = 0.006), and close to solar metallicity (z = 0.01) 1 , using the multi-zone post-processing code mppnp [42]. The neutron capture network, which is largely based on rates recommended in the latest released version of Kadonis, Kadonis-v0.3 [39], included the new 70 Ge(n, γ) MACSs obtained in this work, and recent results from 73 Ge(n, γ) [12]. Due to the very good agreement with previous results, there are only small changes in the resultant 70 Ge abundances produced in massive stars (about 3%) using the new 70 Ge MACS. However, our data provide an independent confirmation of the only previous measurement at stellar energies, and in addition, include also MACS for kT > 50 keV, thus improving the accuracy of stellar model predictions. Figure 6 shows the weak s process abundances produced before supernova explosion, scaled to solar system abundances and normalised to 70 Ge. Isotopes of the same element have been connected by solid lines. Contributions to the solar system abundances due to the main s process and the p process have been subtracted using results by Arlandini et al. [43] and Travaglio et al. [44], respectively. The absolute contributions of these two nucleosynthesis processes have been determined by normalising to the main s-only isotope 150 Sm, and p-only isotope 84 Sr, respectively. Hence, the solar system abundances shown in Fig. 6 only contain contributions from the weak s-process, and explosive nucleosynthesis processes (r process, alpha-rich freeze out). The results show that the z = 0.006 model reproduces best the solar isotopic abundance pattern of germanium isotopes, while the model close to solar metallicity (z = 0.01) provides a better global fit to the other s-only isotopes 76 Se, 80 Kr and 82 Kr. Furthermore, the z = 0.01 model indicates that the weak s process contributes a large fraction to solar 65 Cu, 67,68 Zn, 69,71 Ga, 72,73,74 Ge, 75 As and 77,78 Se abundances. For a firm conclusion about the absolute contribution of the weak s-process however, stellar nucleosynthesis calculations using the new cross sections need to be implemented into calculations of Galactic Chemical Evolution, taking into account stars of different initial metallicities and masses, as well as contributions from explosive nucleosynthesis processes during the core collapse supernova explosion.

V. SUMMARY
We have measured 70 Ge(n, γ) cross sections up to 300 keV neutron energy at the neutron time-of-flight facility n TOF at CERN. Resonance capture kernels of 110 resonances were determined up to 40 keV, while averaged cross sections have been determined up to 300 keV. Maxwellian Averaged Cross Sections have been calculated, which are in very good agreement with the only other previous measurement of this reaction at stellar energies [15], providing an important independent confirmation of stellar cross sections used in astrophysical calculations. The new MACSs combined with recent results for MACSs on 73 Ge(n, γ) [12] have been used to calculate abundances produced in two massive star models with sub-solar, and close to solar metallicity. Abundances for s-only isotopes match reasonably well with solar system values, in particular for the model close to solar metallicity, while the = 0.006 model reproduces the germanium isotopic abundance pattern in our solar system.  6. Abundances produced in a 25 M⊙ star for two initial metallicities z, normalised to solar system abundances for which contributions from the main s process and p process have already been subtracted. The distribution has been normalised to 70 Ge. s-only isotopes are shown as full circles and isotopes of the same element are connected by solid lines. The z = 0.006 metallicity model produces germanium isotopic abundances close to solar proportions, while the z = 0.01 model provides a better global fit for A=60-80.