Modification of Cu Oxide and Cu Nitride Films by Energetic Ion Impact

: We have investigated lattice disordering of cupper oxide (Cu 2 O) and copper nitride (Cu 3 N) films induced by high-and low-energy ion impact, knowing that the effects of electronic excitation and elastic collision play roles by these ions, respectively. For high-energy ion impact, degradation of X-ray diffraction (XRD) intensity per ion fluence or lattice disordering cross-section (Y XD ) fits to the power-law: Y XD = (B XD S e ) NXD , with S e and B XD being the electronic stopping power and a constant. For Cu 2 O and Cu 3 N, N XD is obtained to be 2.42 and 1.75, and B XD is 0.223 and 0.54 (kev/nm) − 1 . It appears that for low-energy ion impact, Y XD is nearly proportional to the nuclear stopping power (S n ). The efficiency of energy deposition, Y XD /S e , as well as Y sp /S e , is compared with Y XD /S n , as well as Y sp /S n . The efficiency ratio R XD = (Y XD /S e )/(Y XD /S n ) is evaluated to be ~0.1 and ~0.2 at S e = 15 keV/nm for Cu 2 O and Cu 3 N, meaning that the efficiency of electronic energy deposition is smaller than that of nuclear energy deposition. R sp = (Y sp /S e )/(Y sp /S n ) is evaluated to be 0.46 for Cu 2 O and 0.7 for Cu 3 N at S e = 15 keV/nm.


Introduction
It is known that electronic excitation under energetic-ion impact leads to modifications of solids such as electronic sputtering (erosion of solid materials induced by electronic energy deposition) and track formation.For example, electronic sputtering by monoenergetic ions (>0.1 MeV/u) has been observed in many non-metallic solids: H 2 O ice by Brown et al. [1], H 2 O ice, and frozen gas of Xe, CO 2 , and SF 6 by Bottiger et al. [2], ice of CO, Ar, and N 2 , by Brown et al. [3], CO 2 ice by Mejia et al. [4], H 2 O ice by Galli et al. [5], SiO 2 , LiNbO 3 , Al 2 O 3 , and Si 3 N 4 by Qui et al. [6], UF 4 by Meins et al. [7], UO 2 by Bouffard et al. [8] and Schlutig [9], LiF and SiO 2 by Toulemonde et al. [10,11], SiO 2 , SrTiO 3 , SrCeO 3 , CeO 2 , MgO, TiO 2 ,and ZnO by Matsunami et al. [12], CaF 2 , LaF 3 , and UF 4 [11], NaCl, and RbI by Toulemonde et al. [13].Also, it has been observed for Si 3 N 4 , AlN, Y 2 O 3 and ZrO 2 [14], KBr and SiC [15], and Fe 2 O 3 and TiN [16].It is noticed that sputtering of frozen Xe films has been detected for low-energy electron impact, contrary to the anticipation of no atomic-displacement due to elastic collisions [2], confirming that electronic excitation plays an important role in sputtering.This paper concerns equilibrium charge incidence, which is achieved by the insertion of thin carbon and metal films.Sputtered atoms are collected in the catcher foils, and neutron activation methods for U detection [7][8][9] and ion beam analysis are utilized to obtain the sputtering yields.
The electronic energy deposition or electronic stopping power (S e ) of ions with equilibrium charge and the nuclear stopping power (S n ) can be calculated using TRIM or SRIM codes by Ziegler et al. [17,18].Briefly, TRIM/SRIM codes utilize semi-empirical formulae derived from experimental data and dielectric response theory to calculate S e for charged particles and the interaction potential (the so-called ZBL potential) fitted to experimental data for S n calculations.Characteristic features of electronic sputtering by high-energy ions [16] are reproduced: (a) Electronic sputtering yields per ion (Y SP ) superlinearly depend on S e and are approximated by the power-law fit: Y SP = (B SP S e ) Nsp with 1 ≦ N SP ≦ 4 for most cases, B SP being a material-dependent constant.(b) Y SP is larger by 10-10 3 than nuclear sputtering yields due to elastic collision cascades, which can be estimated assuming a linear dependence on S n .The representative sputtering yields at S e = 10 keV/nm by monatomic ions at normal incidence vary from 0.42 (MgO) [12] to 844 (WO 3 ) [16], and these yields are plotted as a function of the bandgap (E g ) and elastic constant.It appears that the sputtering yields tend to increase with the bandgap for E g > 3 eV, and the yields decrease with the elastic constant for oxides.Stoichiometric sputtering has been observed for many materials, although only heavy elements such as U have been detected [7][8][9].Non-stoichiometric sputtering has been argued for some fluorides (CaF 2 , LaF 3 and UF 4 ) [11].It should be mentioned that Bragg's additive rule is applied to obtain stopping powers for compound solids in the present study.The deviation from Bragg's rule is not serious for oxides and nitrides, being roughly 10% or less; around 1 MeV/u [16].In most cases, the variation in S e calculated using different versions of TRIM/SRIM is within 10% [16].
Furthermore, tracks (amorphous or disordered region along the ion trajectory for crystalline solids and density modification for amorphous solids) have been observed using transmission electron microscopy (TEM), Rutherford backscattering spectroscopychanneling (RBS-C), small-angle X-ray scattering (SAXS), and atomic force microscopy (AFM).Here, some remarks for the results by monatomic ions are pointed out.Firstly, the S e dependence of the track radius is much weaker than that of electronic sputtering.Secondly, tracks have been observed for S e larger than the threshold (S eth ) in oxides (quartz or single-crystal SiO 2 ), mica, V 2 O 5 glass, ZrSiO 4 (zircon), polyester resin, etc., (TEM) by Fleisher et al. [19] and summarized by Itoh et al. [20].The threshold values scatter [20], and it seems that S eth depends on the analysis methods.
Amorphization of SiO 2 quartz has been observed (RBS-C, TEM) by Meftah et al. [21], Toulemonde et al. [22], and (SAXS, TEM) Afra et al. [23].At S e = 20 keV/nm, the track radius reads 5.3 nm [21,22] and 3.8 nm (S e averaged over the track length, S e at the surface being 26 keV/nm (larger by 30% than the average)) [23].Density modification, i.e., lower density in the track core surrounded by the shell with higher density, has been observed (SAXS) for amorphous (or vitreous) SiO 2 , and the track radius is 5.4 nm (sum of core and shell radii) at S e = 16 keV/nm (SAXS) by Kluth et al. [24].The radius is somewhat larger than 4 nm in SiO 2 quartz [21,22].Track radius (optical infrared spectroscopy (IR)) reads 6 nm at S e = 20 keV/nm by Benyagoub et al. [25], a value comparable with the track radius in SiO 2 quartz.Toulemonde et al. have investigated track radius in amorphous SiO 2 for a wide range of S e , and they have found no apparent threshold S e of track formation due to synergy of the electronic and nuclear stopping powers [26].For Al 2 O 3 single crystal (sapphire), track radius (RBS-C) at Se = 30 keV/nm reads 1.6 nm by Canut et al. [27], 1.7 nm by Khalfaoui et al. [28], and 2 nm of disordered region at S e = 25 keV/nm (TEM analysis of strain) by O'Connell et al. [29].Hillock radius reads 10 nm at Se = 30 keV/nm (AFM), which is much larger than the track radius (the relationship between the hillock and track radii is not clear) [28].
Moreover, amorphization in a track for polycrystalline Si 3 N 4 has been observed (TEM) by Zinkle et al. [30], and amorphization and density modification (TEM) by Vuuren et al. [31].The track radius is 1.8 nm at S e ≃ 34 keV/nm [30] and 1.9 nm at S e = 23 keV/nm [31].The track radius in the amorphized region is 3.1 nm (S e = 23 keV/nm) [31].Density modification similar to a-SiO 2 has been observed in a-SiN 1.13 film (TEM) by Kitayama [32] and in a-SiN 0.95 :H (SAXS) by Santiago [33].The track radius is 5.5 nm (S e = 24 keV/nm by 0.2 GeV Au ions) [32] and 4.1 nm (S e = 24 keV/nm) [33].The track radius in amorphous Si 3 N 4 is larger than that in polycrystalline Si 3 N 4 .Disordered tracks have been observed in epilayer GaN by Kucheyev et al. (TEM) [34] and Mansouri et al. (AFM) [35] and in epilayer GaN and InN by Sall et al. (TEM) [36].The track radius is 5 nm (S e = 34 keV/nm) [34], 1.5 nm (S e = 23-28 keV/nm) [35], and 2.4 and 1.5 nm (S e = 46 and 24 keV/nm, respectively) [36].It is noticed that no track formation has been observed in AlN by monatomic ions up to S e = 33 keV/nm, while it has been observed by irradiation of C cluster (fullerene) ions with S e > 20 keV/nm.Cluster ion effects have not been understood yet.Disordered or amorphized regions have been observed in single-crystalline InP (RBS-C) by Kamarou et al. [37], and the track radius is 1.2-2.3 and 3 nm at S e = 19 and 23 keV/nm, which are comparable with those in [36].
Comparison of electronic sputtering with lattice disordering (degradation of X-ray diffraction (XRD) intensity (Y XD )) has been reported for KBr and SiC [15] and SiO 2 , ZnO, Fe 2 O 3 , TiN, and WO 3 [16].As in the case of sputtering, it appears that Y XD superlinearly depends on S e , and the S e dependence of Y XD fits well to the power-law [15,16].It would be worthwhile to extend investigations of lattice disordering to materials other than those mentioned above for a better understanding of electronic excitation effects.In this article, we have measured the lattice disordering of Cu 2 O and Cu 3 N films, and the results are compared with the sputtering results [38][39][40].The results of high-energy ions (electronic excitation effects) will be compared with those of low-energy ions (elastic collision effects).The efficiency of the energy deposition on the material modification will be discussed.

Materials and Methods
XRD was measured using Cu-k α radiation (typically, under X-ray operation conditions of 30 kV and 30 mA, with a scan step of 0.02 • and a scan speed of several degrees per min).The XRD intensity was multiplied by the scan speed to obtain the ion fluence dependence.The accuracy of the XRD intensity was estimated to be approximately 10%, taking the variation of repeated measurements into account.RBS was performed with 1.8 MeV He ions at a scattering angle of 160 • for evaluation of film thickness and composition.Similarly, the accuracy of the RBS was estimated considering the variation of repeated measurements.Ion irradiation was performed at room temperature and normal incidence.High-energy ion irradiation with lower incident charge than the equilibrium charge without a carbon foil is often employed for XRD measurement; however, the effect of non-equilibrium charge incidence does not come into play because the length for attaining the equilibrium charge is much smaller than the film thickness, as described later (after the description of sample preparation).
Cu 2 O films were prepared on MgO substrate (thickness of 0.5 mm) at 700 • C using a reactive RF (radio frequency) magnetron sputtering (RFMS) method with a Cu disk target (purity of 99.99%) in Ar and O 2 gas [38,39].According to XRD, the films are polycrystalline with diffraction peaks around 36 • , with a very weak peak around 61 • for unirradiated films [38,39].These are assigned as (111) and (220) diffraction peaks of a cubic structure [41].The peak of (111) diffraction is of concern in this study.The lattice parameter derived from the XRD result of unirradiated films is obtained to be 0.429 nm, which agrees well with the value of 0.427 nm in the literatures [41].Film thickness is obtained to be ~100 and 76-99 nm for high-and low-energy ion irradiation, respectively.The composition is stoichiometric (the composition ratio of oxygen and copper, O/Cu = 0.5 ± 0.05).For the calculation of stopping powers, the film density is taken to be 6.0 gcm −3 (i.e., 5.0 × 10 22 Cu cm −3 ) from the XRD results, which agrees with that in [42].The attenuation length (L XA ) of Cu-k α (8.0 keV) is obtained to be 36 µm [43], and the attenuation depth (L XA sin(36 • /2)) = 11 µm.The film thickness is much smaller than the attenuation depth, and thus, no correction is necessary for the XRD intensity.
Cu 3 N films were prepared on R-cut-Al 2 O 3 substrate (thickness of 0.3 mm) at 250 • C using a reactive RFMS deposition with a Cu disk of 99.99% in pure N 2 gas [40].Diffraction peaks are observed at ~23 • and 48 • and assigned as (100) and (200) diffraction planes of a cubic structure [41].The lattice parameter is obtained to be 0.3831 nm with a sample variation of 0.4% (20 samples) and is close to 0.3815 nm [41].From the lattice parameter, the film density is estimated to be 5.4 × 10 22 Cu cm −3 or 6.12 gcm −3 (close to 5.84 gcm −3 [42]), and this is employed for stopping power calculation.According to RBS of 1.8 MeV He, the film thickness used in this study is 100-200 and 88 nm for high-and low-energy ion irradiation, respectively.The composition is evaluated to be roughly stoichiometric, considering poor accuracy (~30%) because of poor statistics of nitrogen yields.Similarly to Cu 2 O, L XA of Cu-k α is obtained to be 34 µm [43], and the attenuation depth (L XA sin(23 • /2)) = 6.7 µm.The film thickness (90-200 nm) is much smaller than the attenuation depth, and thus, no correction is required for the XRD intensity.
The beam current density of 100 keV Ne + and N + ions has been 0.5-2 µAcm −2 (flux of 0.3-1.3× 10 13 cm −2 ).The macroscopic temperature rise is estimated to be 40 • C [44].The beam current of high-energy ions with the incident charge of +q has been ~1 particle nA (beam current divided by q or 6 × 10 9 ions per cm −2 ), and an estimated temperature rise is 40 • C.This temperature rise is not significant for Cu 2 O and will be discussed for Cu 3 N in Section 3.3.
Following [45], the characteristic length (L EQ ) for attaining the equilibrium charge is estimated from the empirical formula of single-electron loss cross-section [46,47], with the first ionization potential (Ip) [42,48]; L EQ = 1/(single-electron loss cross-section (σ 1L ) times N), N being the atomic density.L EQ and σ 1L are listed in Table 1.L EQ is much smaller than the film thickness, and hence, the charge-state effect is insignificant in this study.
Table 1.Ion, energy (MeV), first ionization potential of ion (I P , eV), effective number of removable electrons (N eff ), single-electron loss cross-section (σ 1L , 10 −16 cm 2 ), characteristic length (L EQ , nm) attaining the equilibrium charge for Cu 2 O and Cu 3 N.I P is taken from [42,48], and σ 1L is taken from [46,47].(the most intensive (111) diffraction of cubic cuprite), normalized to that of unirradiated films (normalized intensity), is shown in Figure 3 as a function of ion fluence for 90 MeV Ni +10 , 100 MeV Xe +14 , and 200 MeV Xe +14 ions.It appears that the normalized XRD intensity is proportional to the ion fluence.From the slope, the XRD degradation yield per unit ion fluence (Y XD , representing lattice disordering cross-section) is obtained, and the results are provided in Table 2.The linear relationship (Figure 3) can be understood by the idea that the normalized XRD intensity corresponds to 1 − χ min .Here, χ min is the minimum yield of RBS-C due to disordering or amorphization.A saturation formula [49] is often applied to the ion fluence dependence of χ min :χ min = 1 − exp(−AF), A and F being the disordering cross-section and ion fluence, and χ mi is proportional to F for low F. Lattice compaction has been observed, and the appreciable change begins at low fluence, as already reported [38].The decrease in the lattice parameter per unit fluence (Y LC ) is obtained to be 2.5, 4.3, and 6.8 × 10 −15 cm −2 for 90 MeV Ni, 100 MeV Xe, and 200 MeV Xe ions, respectively, with an estimated accuracy of 20%.S e dependence of the lattice parameter change is fitted by a power-law and obtained to be (0.123•S e ) 1.54 .It appears that the S e dependence of Y LC is weaker than that of Y XD , as described later.

Ion
The electronic stopping power (S e *) appropriate for XRD intensity degradation is calculated using TRIM 1997 [17], with the approximation that the appropriate energy (E*) is yielded by the energy at the middle of the film thickness (~50 nm), i.e., S e * = S e (E*) with E* = E(incidence)-S e (E)•50 nm (Table 2).This correction is very close to that in C-foil, employed for sputtering measurements, and thus, the same S e * is adopted for both the XRD intensity degradation and sputtering (Table 2).Correction of S e appears to be less than 1 percent.The areal density of Cu in the C-foil collector vs. ion fluence after revision [15] is shown in Figure 4.It is seen that Cu areal density is proportional to the ion fluence, and from the slope (Cu areal density/ion fluence), the sputtering yields of Cu (Y Cu ) are obtained.Here, the collection efficiency of Cu in the C-foil collector is taken to be 0.3 [38].The total sputtering yields Y sp of Cu 2 O equal to 1.5 times Y Cu , assuming stoichiometric sputtering.Y sp of Cu 2 O is summarized in Table 2. Comparison of S e dependence of the disordering Y XD with that of Y sp is shown in Figure 5.Both Y XD and Y sp fit to the power-law.It is seen that Y XD has a much stronger dependence than that of Y sp and that of Y LC , as mentioned above.
section and ion fluence, and χmi is proportional to F for low F. Lattice compaction has been observed, and the appreciable change begins at low fluence, as already reported [38].The decrease in the lattice parameter per unit fluence (YLC) is obtained to be 2.5, 4.3, and 6.8 × 10 -15 cm -2 for 90 MeV Ni, 100 MeV Xe, and 200 MeV Xe ions, respectively, with an estimated accuracy of 20%.Se dependence of the lattice parameter change is fitted by a powerlaw and obtained to be (0.123•Se) 1 . 54.It appears that the Se dependence of YLC is weaker than that of YXD, as described later.
The electronic stopping power (Se*) appropriate for XRD intensity degradation is calculated using TRIM 1997 [17], with the approximation that the appropriate energy (E*) is yielded by the energy at the middle of the film thickness (~50 nm), i.e., Se* = Se(E*) with E* = E(incidence)-Se(E)•50 nm (Table 2).This correction is very close to that in C-foil, employed for sputtering measurements, and thus, the same Se* is adopted for both the XRD intensity degradation and sputtering (Table 2).Correction of Se appears to be less than 1 percent.The areal density of Cu in the C-foil collector vs. ion fluence after revision [15] is shown in Figure 4.It is seen that Cu areal density is proportional to the ion fluence, and from the slope (Cu areal density/ion fluence), the sputtering yields of Cu (YCu) are obtained.Here, the collection efficiency of Cu in the C-foil collector is taken to be 0.3 [38].The total sputtering yields Ysp of Cu2O equal to 1.5 times YCu, assuming stoichiometric sputtering.Ysp of Cu2O is summarized in Table 2. Comparison of Se dependence of the disordering YXD with that of Ysp is shown in Figure 5.Both YXD and Ysp fit to the powerlaw.It is seen that YXD has a much stronger dependence than that of Ysp and that of YLC, as mentioned above.The optical absorption spectra of unirradiated Cu 2 O and irradiated by 200 MeV Xe at 7.15 × 10 12 cm −12 are shown in Figure 6.No appreciable change in optical absorption spectra by the ion irradiation is observed, as reported for 100 MeV Xe ions at 3 × 10 −12 cm −2 [38], except for a small reduction of a hump around the optical wavelength of 0.344 µm.The bandgap (E g ) is obtained to be 2.54 and 2.52 eV for Cu 2 O films of unirradiated and irradiated by 200 MeV Xe at 7.15 × 10 12 cm −2 (Figure 7).Here, a relationship is employed: (absorbance × photon energy) 2 is proportional to (photon energy-E g ).The present bandgap agrees well with 2.52 eV [50] and 2.51 eV after annealing [51], though a large scatter of the bandgap has been seen, (2-2.38 eV) pointed out in [39] and (2-2.6 eV) [50], and more com-plications are discussed by Meyer et al. [52].Aside from the determination of the bandgap, no appreciable change in the bandgap as well as optical absorption is observed for 200 MeV Xe at 7.15 × 10 12 cm −2 , 90 MeV Ni at 24 × 10 12 cm −2 , and 100 MeV Xe at 6 × 10 12 cm −2 , as reported for 100 MeV Xe at 3 × 10 12 cm −2 [38].The result of high-energy ion impact contrasts with that of low-energy ion impact, where E g deceases with the ion fluence [39], and the difference of the high-and low-energy ion impact will be discussed in the next section.It would be worthwhile to mention that the optical absorbance (OA = Log 10 (I o /I), I o and I being the incident and transmitted photon intensity) at 344 nm (3.6 eV) appears to be proportional to the film thickness L for unirradiated films.The absorbance coefficient (OA/L) is obtained to be 0.027 nm −1 after subtracting the OA of the MgO substrate (0.065), and the absorption coefficient α (defined by I = I o exp(-αL)) is 6.2 × 10 5 cm −1 .The present absorption coefficient agrees with 6.77 × 10 5 cm −1 [53].The OA coefficient is larger by a factor of ~4 than that of doped ZnO [54].being the incident and transmitted photon intensity) at 344 nm (3.6 eV) appears to be proportional to the film thickness L for unirradiated films.The absorbance coefficient (OA/L) is obtained to be 0.027 nm -1 after subtracting the OA of the MgO substrate (0.065), and the absorption coefficient α (defined by I = Io exp(-αL)) is 6.2 × 10 5 cm -1 .The present absorption coefficient agrees with 6.77 × 10 5 cm -1 [53].The OA coefficient is larger by a factor of ~4 than that of doped ZnO [54].

Cu2O: Low-Energy Ion Impact
XRD intensity at a diffraction angle of ~36°, normalized to that of unirradiated films (normalized intensity), is shown in Figure 8 as a function of ion fluence for 100 keV N and 100 keV Ne.As in the case of high-energy ion impact (Figure 3), it is found that the normalized XRD intensity decreases proportionally to the ion fluence up to a certain fluence.The slope of the normalized intensity vs. ion fluence is reduced for high fluence, and this could be due to the overlapping effect.From the slope for low fluence, the XRD degradation yield per unit ion fluence (YXD), i.e., lattice disordering cross-section, is obtained, and the results are provided in Table 3. Lattice compaction is ~2% at 10 × 10 16 cm -2 for both 100 keV N and Ne, nearly independent of the ion species.Ion, energy, film thickness, projected range (Rp), nuclear and electronic stopping powers averaged over the film thickness (L), and fraction retained of ions in the film (FR) are summarized in Table 3, noting that for L < Rp, FR may play roles in the modifications.Sputtering yields as well as the stopping powers at the surface are also provided in Table 3.The maximum fluence is 10 and 5.2 × 10 16 cm −2 for sputtering measurements by 100 keV N and Ne ions, respectively.Averaged stopping powers differ from those in [39], since the electronic energy loss carried by recoils is not taken into account in this study.This choice does not affect the following discussion.
From Table 3, it is found that the ratio of YXD (100 keV Ne) over YXD (100 keV N) is

Cu 2 O: Low-Energy Ion Impact
XRD intensity at a diffraction angle of ~36 • , normalized to that of unirradiated films (normalized intensity), is shown in Figure 8 as a function of ion fluence for 100 keV N and 100 keV Ne.As in the case of high-energy ion impact (Figure 3), it is found that the normalized XRD intensity decreases proportionally to the ion fluence up to a certain fluence.The slope of the normalized intensity vs. ion fluence is reduced for high fluence, and this could be due to the overlapping effect.From the slope for low fluence, the XRD degradation yield per unit ion fluence (Y XD ), i.e., lattice disordering cross-section, is obtained, and the results are provided in Table 3. Lattice compaction is ~2% at 10 × 10 16 cm −2 for both 100 keV N and Ne, nearly independent of the ion species.Ion, energy, film thickness, projected range (R p ), nuclear and electronic stopping powers averaged over the film thickness (L), and fraction retained of ions in the film (F R ) are summarized in Table 3, noting that for L < R p , F R may play roles in the modifications.Sputtering yields as well as the stopping powers at the surface are also provided in Table 3.The maximum fluence is 10 and 5.2 × 10 16 cm −2 for sputtering measurements by 100 keV N and Ne ions, respectively.Averaged stopping powers differ from those in [39], since the electronic energy loss carried by recoils is not taken into account in this study.This choice does not affect the following discussion.Table 3. XRD and sputtering data of Cu2O films for low-energy ions.Ion, incident energy (E, keV), film thickness L (nm), projected range (Rp nm), nuclear (<Sn>) and electronic (<Se>) stopping powers averaged over film thickness relevant to XRD intensity degradation (YXD), YXD(10 −16 cm 2 ), retained fraction of ions in film (FR %), nuclear (Sn) and electronic (Se) stopping powers at surface relevant to sputtering, sputtering yield (YSP).Projected range, stopping powers, and retained fraction are calculated using TRIM1997.Sputtering data from [39].No appreciable change in Eg has been observed for high-energy ions.Thus, the observation that the ion fluence dependence of the decrease in the bandgap (Eg) is the same for 100 keV N and Ne ions cannot be explained by nuclear energy deposition or electronic energy deposition.This could be partly due to the implantation effect of low-energy ions, considering the appreciable retained fraction of 11% (100 keV N) and 34% (100 keV Ne), since it has been observed that XRD intensity variation in WNOx films is identical for lowenergy (~1 keV) H and D implantation [56].However, it is not well understood yet at present why there is no ion species dependence of the bandgap decrease in Cu2O with the ion fluence, and more investigations are desired.From Table 3, it is found that the ratio of Y XD (100 keV Ne) over Y XD (100 keV N) is obtained to be 2.68, and this ratio reasonably agrees with the ratio of the average nuclear stopping powers of 2.3.It is also found, as expected, that Y sp (100 keV Ne)/Y sp (100 keV N) is obtained to be 2.08, in good agreement with the nuclear stopping power ratio of 2.3.Here, the so-called α-factor (order of unity, representing the efficiency of the nuclear stopping power contributing to nuclear sputtering [55]) is disregarded.Hence, lattice disordering and sputtering by low-energy ions are nearly proportional to the nuclear stopping power.

XRD patterns of
No appreciable change in E g has been observed for high-energy ions.Thus, the observation that the ion fluence dependence of the decrease in the bandgap (E g ) is the same for 100 keV N and Ne ions cannot be explained by nuclear energy deposition or electronic energy deposition.This could be partly due to the implantation effect of low-energy ions, considering the appreciable retained fraction of 11% (100 keV N) and 34% (100 keV Ne), since it has been observed that XRD intensity variation in WNO x films is identical for low-energy (~1 keV) H and D implantation [56].However, it is not well understood yet at present why there is no ion species dependence of the bandgap decrease in Cu 2 O with the ion fluence, and more investigations are desired.[40].In this study, only low fluence is of concern, and it is assumed that decomposition does not play a significant role, and this point will be discussed later.It appears that the normalized XRD intensity is proportional to the ion fluence up to certain fluence.From the slope for low fluence, the XRD degradation yield per unit ion fluence (Y XD , lattice disordering cross-section) is obtained, and the results are provided in Table 4.A small lattice compaction (−0.2% for 100 MeV Xe at 10 13 cm 2 or Y LC ~2 × 10 −16 cm 2 , which is much smaller than that for Cu 2 O (Section 3.1)) or no appreciable change in the lattice parameter has been observed for the other high-energy ion impacts, including the results shown in Figures 9 and 10.Sputtering yields (Y sp ) by high-energy ions are also provided in Table 4.The maximum fluence for the sputtering measurement is 0.8, 2, 1.4, and 1.5 × 10 12 cm −2 for 200 MeV Xe, 100 MeV Xe, 90 MeV Ni, and 60 MeV Ar ions.S e dependence of Y XD and Y sp is shown in Figure 12. and 10.Sputtering yields (Ysp) by high-energy ions are also provided in Table 4.The maximum fluence for the sputtering measurement is 0.8, 2, 1.4, and 1.5 × 10 12 cm −2 for 200 MeV Xe, 100 MeV Xe, 90 MeV Ni, and 60 MeV Ar ions.Se dependence of YXD and Ysp is shown in Figure 12.
Normalized XRD intensity as a function of ion fluence for 100 KeV N and 100 keV Ne ions is shown in Figure 13.The XRD degradation yield per unit ion fluence (YXD) is obtained, and the results are provided in Table 5.From Table 5, it is found that the ratio of YXD (100 keV Ne) over YXD (100 keV N) is obtained to be 2.45, and this ratio reasonably agrees with the average nuclear stopping power ratio of 2.2.Hence, the lattice disordering is nearly proportional to the nuclear stopping power.The Cu peak in the XRD appears at 0.82 × 10 15 cm −2 (~1 × 10 15 cm −2 [40]) for 100 keV Ne ion impact and 4.5 × 10 15 cm −2 for 100 keV N ion impact, respectively.Lattice compaction is obtained to be 0.5% at 0.5 × 10 15 cm −2 and 0.54% at 0.1 × 10 15 cm −2 for 100 keV N and 100 keV Ne ions.
Sputtering yields of Cu3N by 100 keV N and Ne ions are carefully reanalyzed, and the revised yields are obtained to be 3.2 and 7.6, respectively (Table 5).Here, the maximum ion fluence for the sputtering yield evaluation is 6 and 31 × 10 15 cm -2 for 100 keV N and Ne, and the fluence is larger than that of the Cu phase appearance in the XRD patterns, as mentioned above.The sputtering yields are much larger than those of Cu2O.The ratio, Ysp (100 keV Ne)/Ysp (100 keV N), is obtained to be 2.38, and this agrees with the nuclear stopping power ratio of 2.22.Again, the so-called α-factor (order of unity, representing the efficiency of the nuclear stopping power contributing to nuclear sputtering [55]) is discarded.As described in Section 2, the estimated macroscopic temperature rise is 40 °C [44].This temperature rise is smaller than the thermal decomposition temperature, which ranges from 100 to 470 °C [57].However, the possibility that the large sputtering yields are partly due to decomposition is not ruled out, even though the temperature rise is lower than the decomposition temperature.Stopping powers are calculated using TRIM1997.Se* calculated using SRIM2013 is provided in the parenthesis after Se* (TRIM1997).Normalized XRD intensity as a function of ion fluence for 100 KeV N and 100 keV Ne ions is shown in Figure 13.The XRD degradation yield per unit ion fluence (Y XD ) is obtained, and the results are provided in Table 5.From Table 5, it is found that the ratio of Y XD (100 keV Ne) over Y XD (100 keV N) is obtained to be 2.45, and this ratio reasonably agrees with the average nuclear stopping power ratio of 2.2.Hence, the lattice disordering is nearly proportional to the nuclear stopping power.The Cu peak in the XRD appears at 0.82 × 10 15 cm −2 (~1 × 10 15 cm −2 [40]) for 100 keV Ne ion impact and 4.5 × 10 15 cm −2 for 100 keV N ion impact, respectively.Lattice compaction is obtained to be 0.5% at 0.5 × 10 15 cm −2 and 0.54% at 0.1 × 10 15 cm −2 for 100 keV N and 100 keV Ne ions.

Ion
Sputtering yields of Cu 3 N by 100 keV N and Ne ions are carefully reanalyzed, and the revised yields are obtained to be 3.2 and 7.6, respectively (Table 5).Here, the maximum ion fluence for the sputtering yield evaluation is 6 and 31 × 10 15 cm −2 for 100 keV N and Ne, and the fluence is larger than that of the Cu phase appearance in the XRD patterns, as mentioned above.The sputtering yields are much larger than those of Cu 2 O.The ratio, Y sp (100 keV Ne)/Y sp (100 keV N), is obtained to be 2.38, and this agrees with the nuclear stopping power ratio of 2.22.Again, the so-called α-factor (order of unity, representing the efficiency of the nuclear stopping power contributing to nuclear sputtering [55]) is discarded.As described in Section 2, the estimated macroscopic temperature rise is 40 • C [44].This temperature rise is smaller than the thermal decomposition temperature, which ranges from 100 to 470 • C [57].However, the possibility that the large sputtering yields are partly due to decomposition is not ruled out, even though the temperature rise is lower than the decomposition temperature.
The optical absorption spectra are shown in Figure 14 for unirradiated and irradiated Cu3N films by 100 MeV Xe ions.Oscillations are seen in the unirradiated film, and these disappear upon ion irradiation, as in the case of 100 keV Ne impact [40].The corresponding bandgap is obtained to be 1.38 and 1.22 eV, using the relation where the square root of the absorbance times the photon energy is proportional to the photon energy-bandgap (Figure 15).The optical bandgap modification by ion impact is small.The absorbance of unirradiated films at 0.48 µm (photon energy of 2.58 eV), divided by the film thickness, is obtained to be 0.02 nm −1 after subtracting the OA of Al2O3 (0.065), and the absorption coefficient is found to be 4.6 × 10 5 cm −1 , in reasonable agreement with 5.2 × 10 5 cm −1 [58].
Figure 12.XRD intensity degradation YXD (10 -15 cm 2 ) (o) and sputtering yield Ysp (∆) as a function of electronic stopping power, Se (keV/nm).Also shown is power-law fit: YXD (10 -15 cm 2 ) = (0.54Se) 1.75 and Ysp = (2.6 Se) 1.6 .An estimated error of YXD and Ysp is 10%.The optical absorption spectra are shown in Figure 14 for unirradiated and irradiated Cu3N films by 100 MeV Xe ions.Oscillations are seen in the unirradiated film, and these disappear upon ion irradiation, as in the case of 100 keV Ne impact [40].The corresponding bandgap is obtained to be 1.38 and 1.22 eV, using the relation where the square root of the absorbance times the photon energy is proportional to the photon energy-bandgap (Figure 15).The optical bandgap modification by ion impact is small.The absorbance of unirradiated films at 0.48 µm (photon energy of 2.58 eV), divided by the film thickness, is obtained to be 0.02 nm −1 after subtracting the OA of Al2O3 (0.065), and the absorption coefficient is found to be 4.6 × 10 5 cm −1 , in reasonable agreement with 5.2 × 10 5 cm −1 [58].
Figure 12.XRD intensity degradation YXD (10 -15 cm 2 ) (o) and sputtering yield Ysp (∆) as a function of electronic stopping power, Se (keV/nm).Also shown is power-law fit: YXD (10 -15 cm 2 ) = (0.54Se) 1.75 and Ysp = (2.6 Se) 1.6 .An estimated error of YXD and Ysp is 10%.The optical absorption spectra are shown in Figure 14 for unirradiated and irradiated Cu 3 N films by 100 MeV Xe ions.Oscillations are seen in the unirradiated film, and these disappear upon ion irradiation, as in the case of 100 keV Ne impact [40].The corresponding bandgap is obtained to be 1.38 and 1.22 eV, using the relation where the square root of the absorbance times the photon energy is proportional to the photon energy-bandgap (Figure 15).The optical bandgap modification by ion impact is small.The absorbance of unirradiated films at 0.48 µm (photon energy of 2.58 eV), divided by the film thickness, is obtained to be 0.02 nm −1 after subtracting the OA of Al 2 O 3 (0.065), and the absorption coefficient is found to be 4.6 × 10 5 cm −1 , in reasonable agreement with 5.2 × 10 5 cm −1 [58].
Table 5. XRD and sputtering data of Cu 3 N films for low-energy ions.Ion, incident energy (E, keV), film thickness L (nm), projected range (R p nm), nuclear (<S n >) and electronic (<S e >) stopping powers (keV/nm) averaged over the film thickness relevant to XRD intensity degradation (Y XD ), Y XD (10 −15 cm 2 ), retained fraction of ions in film (F R %), nuclear (S n ) and electronic (S e ) stopping powers at surface relevant to sputtering, sputtering yield (Y SP ).R p , F R , and stopping powers are calculated using TRIM1997.

Discussion
The efficiency of electronic energy deposition is compared with that of nuclear energy deposition.Given the linear dependence of the electronic sputtering yield on S e in the case of Cu 2 O (Section 3.1), the effects of electronic and nuclear energy deposition on sputtering can be well defined for Cu 2 O.The average of Y sp /S n (the effect of energy deposition) for two ion species is obtained as 7.9 (kev/nm) −1 from Table 3, and this value is comparable with 3.67 (keV/nm) −1 for electronic sputtering (Figure 3).The energy deposition efficiency is summarized in Table 6.Surprisingly, the results indicate that the energy deposition efficiency for sputtering is comparable within a factor of 2, irrespective of nuclear or electronic energy deposition, even though the effect of electronic energy deposition is an indirect process.Similarly, the average efficiency of nuclear energy deposition on lattice disordering of Cu 2 O is obtained to be 1.3 × 10 −16 cm 2 (keVnm) −1 for low-energy ion impact from Table 3.For high-energy ion impact (Table 2 and Figure 5), a crude estimation of the efficiency of electronic energy deposition is obtained as 1.86 × 10 −16 cm 2 /15 keV/nm = 0.124 × 10 −16 cm 2 (keV/nm) −1 at S e = 15 keV/nm.The efficiency is obtained as 0.256 × 10 −16 cm 2 (keV/nm) −1 at S e = 25 keV/nm.Hence, the effect of electronic energy deposition is much smaller than that of nuclear energy deposition (less efficient than electronic energy deposition).It might be surprising that for track formation in amorphous SiO 2 , the same effect of nuclear and electronic energy deposition has been argued [26], in contrast to the above results, noting that the track area and lattice disordering cross-section are similar for SiO 2 at high S e [16].
For Cu 3 N, the average energy deposition efficiency of sputtering for low-energy ions, Y sp /S n for two ion species, is obtained as 31 (keV/nm) −1 from Table 5, and this value is much larger than that of Cu 2 O.Because of the super-linear dependence of electronic sputtering on S e for high-energy ion impact on Cu 3 N, a crude estimation is applied here, as mentioned above.The energy deposition efficiency, Y sp /S e for high-energy ions, is obtained as 23 and 32 (keV/nm) −1 at S e = 15 and 25 keV/nm (Table 4 and Figure 12), and thus, the electronic energy deposition efficiency is comparable with the nuclear energy deposition efficiency for sputtering.The average energy deposition efficiency of lattice disordering for low-energy ions, Y XD /S n for two ion species, is obtained as 14 × 10 −15 cm 2 (keV/nm) −1 from Table 5, and this value is much larger than that of Cu 2 O.The energy deposition efficiency, Y XD /S e for high-energy ions is obtained as 2.6 and 3.8 × 10 −15 cm 2 (keV/nm) −1 at S e = 15 and 25 keV/nm (Table 4 and Figure 12), and thus, the electronic energy deposition efficiency is much smaller than the nuclear energy deposition efficiency for lattice disordering.The electronic and nuclear energy deposition efficiencies are comparable for sputtering, and the former is smaller (but less than an order of magnitude) than the latter for lattice disordering.

Conclusions
We have measured lattice disordering in polycrystalline Cu 2 O and Cu 3 N films using high-and low-energy ions.We find that lattice disordering is caused by electronic excitation for high-energy ion impacts, and the degradation of XRD intensity (lattice disordering crosssection) fits a power-law relationship with electronic stopping power.The degradation of XRD intensity by low-energy ion impact (elastic collisions dominate) is found to be nearly proportional to the nuclear stopping power.We have discussed the efficiency of electronic and nuclear energy deposition on lattice disordering as well as sputtering.It appears that the efficiency of electronic and nuclear energy deposition is comparable for the sputtering of Cu 2 O and Cu 3 N.In the case of lattice disordering of Cu 2 O and Cu 3 N, we find that the electronic energy deposition efficiency is smaller than the nuclear energy deposition efficiency.

Table 2 .
XRD and sputtering data of Cu 2 O films for high-energy ions.Ion, incident energy (E in MeV), XRD intensity degradation (Y XD ), sputtering yield (Y SP ), appropriate energy E* (MeV) considering the energy loss in the film or carbon foil collector, electronic stopping power (S e *), nuclear stopping power (S e *) in keV/nm at E*.The deviation ∆S e * = (S e */S e (E) − 1) × 100 is also provided.Stopping powers are calculated using TRIM1997.S e * calculated using SRIM2013 is provided in the parenthesis after S e * (TRIM1997).

Figure 5 .
Figure 5. XRD intensity degradation Y XD (10 −15 cm 2 ) (o) and sputtering yield Y sp (∆) as a function of electronic stopping power, S e (keV/nm).Also shown is power-law fit: Y XD = (0.223S e ) 2.42 and Y sp = 3.67 S e .An estimated error of Y XD and Y sp is 10%.

Figure 7 .
Figure 7. Square of optical absorbance times photon energy vs. photon energy of Cu 2 O for unirradiated Cu 2 O (blue) and Cu 2 O irradiated by 200 MeV Xe at 7.2 × 10 12 cm −2 (red), illustrating the bandgap.

Quantum 18 Figure 8 .
Figure 8. XRD intensity from (111) diffraction plane at 36-37°, normalized to unirradiated films of Cu2O as a function of ion fluence for 100 keV N (∆) and 100 keV Ne (o).Data of 100 keV Ne from [39].Linear fit is indicated by dotted lines.An estimated error of XRD intensity is 10%, as indicated by error bars.
Cu3N films are shown in Figures 9 and 10 for irradiation by 200 MeV Xe and 100 MeV Xe ions.It is seen that the XRD intensity decreases with the ion fluence, and the diffraction angle remains unchanged.XRD intensity of Cu3N films on R-cut Al2O3 at a diffraction angle of 23° (the most intensive (100) diffraction of cubic structure), normalized to that of unirradiated films (normalized intensity), is shown in Figure 11 as a function of ion fluence for 90 MeV Ni +10 , 100 MeV Xe +14 , and 200 MeV Xe +14 ions.It is

Figure 8 .Table 3 .
Figure 8. XRD intensity from (111) diffraction plane at 36-37 • , normalized to unirradiated films of Cu 2 O as a function of ion fluence for 100 keV N (∆) and 100 keV Ne (o).Data of 100 keV Ne from [39].Linear fit is indicated by dotted lines.An estimated error of XRD intensity is 10%, as indicated by error bars.
XRD patterns of Cu 3 N films are shown in Figures 9 and 10 for irradiation by 200 MeV Xe and 100 MeV Xe ions.It is seen that the XRD intensity decreases with the ion fluence, and the diffraction angle remains unchanged.XRD intensity of Cu 3 N films on R-cut Al 2 O 3 at a diffraction angle of 23 • (the most intensive (100) diffraction of cubic structure), normalized to that of unirradiated films (normalized intensity), is shown in Figure 11 as a function of ion fluence for 90 MeV Ni +10 , 100 MeV Xe +14 , and 200 MeV Xe +14 ions.It is reminded that the Cu phase has been observed at large fluence, meaning decomposition

Figure 11 .
Figure 11.XRD intensity from (100) diffraction plane at nearly 23 • , normalized to unirradiated films of Cu 3 N as a function of ion fluence for 90 MeV Ni (o), 100 MeV Xe (□, x), and 200 MeV Xe (∆) ions.Data of 100 MeV Xe (x) from [40].Linear fit is indicated by dotted lines.The estimated error of XRD intensity is 10%, as indicated by error bars.

Table 4 .
XRD and sputtering data of Cu 3 N/R-Al 2 O 3 films for high-energy ions.Ion, incident energy (E in MeV), XRD intensity degradation (Y XD ), sputtering yield (Y SP ), appropriate energy E* (MeV) considering the energy loss in the film or carbon foil collector, electronic stopping power (S e *), nuclear stopping power (S n *) in keV/nm at E*.The deviation ∆S e * = (S e */S e (E) − 1) × 100 is also provided.Stopping powers are calculated using TRIM1997.S e * calculated using SRIM2013 is provided in the parenthesis after S e * (TRIM1997).

Figure 13 .
Figure 13.XRD intensity from (100) diffraction plane at nearly 23°, normalized to unirradiated films of Cu3N as a function of ion fluence for 100 keV N (o) and 100 keV Ne (□, x) ions.Data of 100 keV Ne (x) from [40].Linear fit is indicated by dotted lines.An estimated error of XRD intensity is 10%, as indicated by error bars.

Figure 13 .
Figure 13.XRD intensity from (100) diffraction plane at nearly 23°, normalized to unirradiated films of Cu3N as a function of ion fluence for 100 keV N (o) and 100 keV Ne (□, x) ions.Data of 100 keV Ne (x) from [40].Linear fit is indicated by dotted lines.An estimated error of XRD intensity is 10%, as indicated by error bars.

Figure 13 .
Figure 13.XRD intensity from (100) diffraction plane at nearly 23 • , normalized to unirradiated films of Cu 3 N as a function of ion fluence for 100 keV N (o) and 100 keV Ne (∆, x) ions.Data of 100 keV Ne (x) from [40].Linear fit is indicated by dotted lines.An estimated error of XRD intensity is 10%, as indicated by error bars.

Figure 15 .
Figure 15.Square root of optical absorbance times photon energy vs. photon energy of Cu 3 N for unirradiated Cu 3 N (blue) and irradiated Cu 3 N by 100 MeV Xe at 1.2 × 10 12 cm −2 (red), illustrating the bandgap.

Table 2 .
XRD and sputtering data of Cu2O films for high-energy ions.Ion, incident energy (E in MeV), XRD intensity degradation (YXD), sputtering yield (YSP), appropriate energy E* (MeV) considering the energy loss in the film or carbon foil collector, electronic stopping power (Se*), nuclear stopping power (Se*) in keV/nm at E*.The deviation ∆Se* = (Se*/Se(E) − 1) × 100 is also provided.Stopping powers are calculated using TRIM1997.Se* calculated using SRIM2013 is provided in the parenthesis after Se* (TRIM1997).

Table 2 .
XRD and sputtering data of Cu2O films for high-energy ions.Ion, incident energy (E in MeV), XRD intensity degradation (YXD), sputtering yield (YSP), appropriate energy E * (MeV) considering the energy loss in the film or carbon foil collector, electronic stopping power (Se*), nuclear stopping power (Se*) in keV/nm at E*.The deviation ∆Se* = (Se*/Se(E) − 1) × 100 is also provided.Stopping powers are calculated using TRIM1997.Se* calculated using SRIM2013 is provided in the parenthesis after Se* (TRIM1997).

Table 6 .
A summary of the energy deposition efficiency.R XD = (Y XD /S e )/(Y XD /S n ) and R sp = (Y sp /S e )/(Y sp /S n ) are the efficiency ratios for lattice disordering and sputtering due to electronic and nuclear energy deposition.The values in the parenthesis are S e (keV/nm), used for Y XD /S e evaluation.