Environmental dependence of the photochromic effect of oxygen-containing rare-earth metal hydrides

We study the dependence of the photochromic effect on environment and triggering light. We demonstrate that the first darkening/bleaching cycle of freshly grown films is accompanied by a release of weakly bound hydrogen, most likely present at the grain boundaries. For consecutive photochromic cycles, we do not find further exchange of material with the environment. Moreover, we report bleaching kinetics dependent on the gas environment after darkening with light of energies below the optical bandgap of the film. For darkening with photon energies above the bandgap of the film, we report a linear relation between the degree of darkening and bleaching relaxation time irrespective of gas environment.


I. INTRODUCTION
Metal hydrides are often used as catalysts or reducing agents 1 and battery materials, 2,3 but they have a strong tendency to oxidize and therefore are typically covered with a protective layer.It has been demonstrated that oxygen-containing yttrium hydride thin films are photochromic in air, i.e., reversibly change their absorption spectrum under illumination. 4Later, a photochromic effect was also confirmed for scandium, dysprosium, gadolinium, erbium, and neodymium oxygen-containing hydrides. 5,6hotochromism is observed in a composition range δ (oxygen to yttrium ratio) between 0.5 and 1.5, with the sample composition following the overall non-stoichiometric equation: MeH 2Àδ O δ . 6,7ased on charge neutrality considerations assuming single-phase metal oxyhydrides, 8,9 Cornelius et al. 10 proposed a different formula (MeH 3À2δ O δ ).Both these studies agree on the composition range in which photochromism is found.However, the typically columnar structure of the films suggests lateral concentration fluctuations, and indeed a more detailed and atomically resolved compositional analysis study using atom probe tomography and energy-dispersive x-ray spectroscopy revealed the presence of two phases (hydrogen-and oxygen-enriched, respectively) for a photochromic GdHO film. 11hotochromic films are commonly produced by depositing the rare earth metal dihydride either by reactive sputtering under an argon/hydrogen atmosphere 12 or by e À -beam evaporation. 13he resulting MeH 2 (Me=Y, Gd, Dy, Sc, Er, or Nd) is subsequently oxidized in air.A polycrystalline columnar structure observed by transmission electron microscopy 11 facilitates oxidation and ensures a uniform distribution of oxygen atoms throughout the whole thickness of the film. 14The performance of the photochromic films strongly depends on the microstructure and porosity as these parameters critically influence the oxidation behavior. 14Films, which are kept in air, develop from opaque metal hydride over photochromic oxygen-containing metal hydride into transparent metal oxide.
Despite intensive studies of these materials over the last decade, the mechanism behind the photochromic effect remains disputable.In situ composition analysis under photodarkening and bleaching revealed no changes in composition larger than the sensitivity of the method that is limited by statistical uncertainty and on

Journal of Applied Physics
the order of one atomic percent. 15On the other hand, x-ray diffraction studies showed a reversible shift of the powder diffraction peaks toward larger Q values under illumination. 16,17Moreover, the authors observed that substantial bleaching was only found under exposure to air but not in nitrogen atmosphere and concluded that the films may breathe oxygen.However, a closer look at the diffraction data shows that the apparent shift of diffraction peaks may be explained by material transport between the sub-phases rather than with the environment.The Bragg reflections of the oxide and hydride phases are very close to each other (see, for instance, ICDD powder diffraction files 00-050-1107 and 00-012-0797).Thus, the diffracted intensity may just be redistributed among these two phases, e.g., hydrogen and/or oxygen may migrate between the subphases resulting in the photochromic effect as it was reported in a pressurized photochromic yttrium foil. 18This theory is further supported by Remhof et al. 19 who showed that tiny changes in hydrogen concentration, in the range from YH 1:9 (higher transmission) to YH 2:1 (lower transmission), in the fcc, β-phase go in hand with strong relative changes in optical transmission.
In this article, we shed further light on the role of material exchange with the environment.We investigate the darkening and bleaching kinetics of oxygen-containing rare earth metal hydrides in different gas environments.We show that samples darkened for the first time release hydrogen under illumination.Moreover, we demonstrate that the dependency of bleaching on gas environment is only found if the films are darkened with photons of energy smaller than the bandgap but not for photons of energy larger than the bandgap.

II. EXPERIMENTAL DETAILS
Dysprosium and yttrium hydride films were produced by reactive magnetron sputtering in Ar:H 2 atmosphere on a soda-lime substrate (microscope slides, 10 Â 10 mm 2 and 1 mm thick) and oxidized in air.Commercially available Dy and Y targets (with nominal purity of 99.99% and 99.9%, respectively) were used, and the target-substrate distance was varied from 4 to 8.5 cm.The plasma sputtering current was 120 mA without intentional heating of the substrate.For further details regarding sample preparation, we refer the reader to Ref. 6.
The chemical compositions and thicknesses of the resulting photochromic films were extracted from iterative and self-consistent ion beam analysis (IBA), combining Rutherford backscattering spectrometry, coincidence time-of-flight energy elastic recoil detection analysis, and nuclear reaction analysis.For the measurements, 2 MeV 4 He þ , 36 MeV 127 I 8þ , and 6.4-6.7 MeV 15 N 2þ ions were used as probing beams, respectively.More details on the composition analysis of photochromic oxygen-containing rare earth metal hydride films can be found elsewhere. 6,15ransmission spectra of the samples were measured using a Perkin Elmer Lambda 35 UV/Vis spectrophotometer and normalized to the transmission of air.The photochromic response, ΔT, is defined as the relative change of averaged optical transmission before and after illumination (integrated over the wavelengths' interval between 500 and 900 nm).The effect of wavelength and intensity on photodarkening and bleaching in air was studied using an in-house-developed modular optical diffractometer (see Fig. S1 in the supplementary material), equipped with a supercontinuum laser (SC Fianium SC-400-2) delivering light in the wavelength range from 370 to 2600 nm.The output of the light source is coupled to a dual acousto-optic tunable filter (AOTF) acting as a monochromator with a bandwidth of 2-5 nm for the visible (AOTF VIS 400-760 nm) and near-IR ranges (AOTF NIR 550-1100 nm).The AOTF NIR capable to provide eight different wavelengths simultaneously was used as probing light, whereas one specific wavelength delivered by AOTF VIS was utilized for photodarkening.The probing light was splitted by a beam splitter, and both beams were detected using Si photodiodes (Thorlabs DET100A) and lock-in detection, allowing for on-the-fly normalization of transmittance.To measure transmittance averaged over the VIS-NIR part of the spectrum, the following wavelengths were used as probing light: 520, 560, 600, 640, 680, 720, 760, and 800 nm.To minimize possible parasitic photodarkening from probing light, its intensity was optimized by measuring transmittance over time.A continuously variable neutraldensity filter (Thorlabs NDC-50C-4M) was installed to modulate the intensity of the triggering light.
Bleaching in different environments was studied using a vacuum chamber connected to a turbo-molecular pump and equipped by two view ports allowing for a measurement of optical transmittance.A SPECTRA satellite residual gas analyzer was installed at the inlet of the turbo pump to monitor changes in gas composition.
All samples studied in this paper are summarized with their respective composition, δ, direct optical bandgap, E dir g , as extracted from Tauc plots 20 (see Fig. S2 in the supplementary material), and thickness, D, in Table I.The film thicknesses in nanometers were deduced from IBA assuming bulk densities; for details, see Ref. 21.The oxygen and hydrogen concentrations, δ and 2 À δ, respectively, follow the composition suggested in Ref. 7.

III. RESULTS
Figure 1(a) depicts the transmission spectrum of sample YI in the transparent state together with the spectra of light from the AOTF VIS used for photodarkening in air.Photochromic response ΔT darkening and bleaching ΔT bleaching for different transparency levels of the neutral-density filter and thereby for different intensities of triggering light (λ ¼ 430 nm) are presented in Fig. 1(b).A nearly linear dependence is observed for both forward and backward reactions, which is typical for a single-photon process.The ratio between the two contrasts remains constant with a change in light intensity (ΔT darkening =ΔT bleaching % 2), indicating that the processes  Using light sources of same intensity would yield the same trend, but the wavelength dependence would be even more significant.
To study the effect of gas exposure, sample YII was loaded into a vacuum chamber and exposed to blue light emitted from a LED-array with λ ¼ 455 nm (E ¼ 2:72 eV) and W flux % 500 mW cm À2 and purple light with λ ¼ 400 nm (E ¼ 3:1 eV) and W flux % 10 mW cm À2 .Figure S3 in the supplementary material shows the optical transmission of sample YII together with the spectra of both light sources.While the wavelength of blue light is at the absorption edge of the film, the purple light is in a regime in which the sample is strongly absorbing.The same holds true for all samples studied in this work: the purple light (3.1 eV) and the blue light (2.72 eV) are above and below the band gap, respectively (see Table I).
Sample YII was photodarkened in high vacuum to the same level of ΔT % 0:23, which took 4 h for the purple light and 33 min for the blue one, as a consequence of the different power of the light sources.The total energy of the purple light (40 mWh cm À2 ) is about seven times smaller as compared to the blue one (275 mWh cm À2 ) demonstrating again that light of a shorter wavelength is more efficient in photodarkening the films.Figure 2 shows the bleaching of the sample.It turns out that for the film darkened by purple light (slower darkening and lower total energy), the bleaching in HV is significantly faster.
In order to address a possible gas release and absorption during photodarkening and bleaching, we have done experiments on samples YII and DI inside an ultra-high vacuum (UHV) system.The photochromic films were darkened, while the composition of the gas was monitored by a residual gas analyzer (RGA), attached at the inlet of a turbo molecular pump.Such a measurement allows qualitative information on the composition and the amount of the released material.Figure 3 summarizes the results by plotting the partial hydrogen gas pressure detected by the RGA vs time during and after illumination.To exclude other sources of gas release, e.g., degassing from the chamber walls or the substrate, a series of background measurements were done with the substrate (orange closed

Journal of Applied Physics
circles) only, no film deposited, and the empty chamber (red closed circles).No gas release was detected above 10 À9 mbar, which is the sensitivity of the RGA.
Sample YII was illuminated with the blue LED directly after growth and photodarkened for the first time.During the first 5 min of illumination, the hydrogen partial pressure steeply raises to values of % 2:1 Â 10 À7 mbar, which is more than 400 times the base pressure.For times longer than 5 min, the partial hydrogen pressure decreases.After switching, the LED off the partial hydrogen pressure quickly drops to the base pressure.Considering a pumping speed of 170 L/s for H 2 and integrating pressure over the time of illumination, this results in 10 16 hydrogen atoms, which is about 1% of the total hydrogen % 10 18 , contained in the film.Note that for our estimate we assume that all gases in the chamber are extracted via the turbo pump.
To reproduce our results, we have repeated this experiment with sample DI and find qualitatively the same result.After the first photochromic cycle, films were exposed to air and then the experiment was repeated (green and blue stars).During the second darkening of the films, still a slight increase in hydrogen partial pressure is visible, however, at a much lower level than during the first cycle.Our results can be explained by assuming some of the hydrogen in the film, which is weakly bound and very mobile, becoming released under illumination.In parallel to the hydrogen release, we also detected a weaker release of O 2 and H 2 O resulting in a 2.5 and 3.7 times (above base pressure) increased partial pressure, respectively, and indicating weakly bound O and possibly H 2 O, with total quantities equivalent to sub-mono-layers.Note that both samples showed persistent photochromism after these two photodarkening cycles.Therefore, a significantly lower release of H 2 during the second cycle cannot be attributed to the degradation of the photochromic properties.
To check whether the hydrogen released from the film during photodarkening is reabsorbed during relaxation, we have done bleaching experiments in H 2 and in other environments like UHV, O 2 , and technical air.The experimental procedure was as follows.Samples YII and DI were photodarkened in UHV by blue light.Then, the light was switched off, and the samples were exposed to different gases.Thereupon, the vacuum chamber was vented and the samples were bleached in air.Transmittance of the samples was monitored at every stage of bleaching.Figure 4 summarizes the results.The optical transmittance of samples YII and DI is plotted vs bleaching time.The bleaching process in UHV is, if it exists at all, very slow [see also Fig. 2(b)].Exposure to pure O 2 , pressure 1.5 bar (which is 7 times higher than the partial pressure of O 2 in air), and H 2 does not lead to substantial bleaching either, as seen in Figs.4(a), 4(b), and 4(d).Figure 4(d) covers much longer bleaching times.A similar behavior of the YHO film was earlier reported in N 2 atmosphere. 17Once the chamber is vented and the sample is exposed to air, however, the film bleaches quickly, which is especially evident from Figs. 4(a) and 4(d).Significant bleaching is also observed in technical/dry air [composition of technical air: N 2 þ 20(1)% O 2 , H 2 O , 3 ppm, C n H m , 0:1 ppm, CO , 1:0 ppm, CO 2 , 1:0 ppm], containing mostly nitrogen and oxygen, both at pressures of 1 bar [see Fig. 4(c)].In this case, subsequent venting of the chamber does almost not change the bleaching kinetics, demonstrating that water, present in the ambient air, does not play a key role in the bleaching process.
The data in Fig. 2 show that the bleaching process in vacuum is depending on the wavelength used for photodarkening.To investigate this effect together with the gas release and re-absorption, we illuminated sample YIII for 4 h with purple light in UHV and studied the bleaching kinetics.Figure 5(a) depicts the transmission of the film plotted vs time under exposure to different gases.The bleaching follows an exponential increase in transmitted intensity, and a fit of the relaxation time [see Eq. (1) in supplementary material] shows that the sample bleaches on very similar time scales independent of environment: UHV, air, or H 2 (τ HV ¼ 1:87 h, τ air ¼ 2:60 h, τ H2 ¼ 2:76 h).
To get a better handle on the bleaching kinetics, we have done systematic studies on sample YIII photodarkened in UHV to various degrees with purple light and subsequently exposed to different gases.Figure 5(b) summarizes the results.For films photodarkened to a photochromic response of up to about 22%-25%, the relaxation time is less than 3 h.For stronger darkening, the relaxation time linearly increases to about 18 h for photodarkening of about 42% again independent of the environment.

IV. DISCUSSION
Studies on bleaching in different environments show that the photochromic effect may be significantly different depending on the spectrum of the triggering light.For photodarkening in UHV with purple light, bleaching is almost independent of the gas environment.This situation completely changes when illuminating the films with blue light.In this case, we find a significant difference in bleaching times depending on whether bleaching is done in air and technical air or UHV, H 2 , and O 2 , indicating a possible material At this point, the source of released H 2 remains unclear and may be originating from either the bulk of the film or as a product of photo-catalytic water splitting happening at the surface.The latter scenario is reported, for instance, for photochromic transition metal oxides, [23][24][25][26] which undergo a color change under proton injection from water molecules.Light-induced proton injection could also explain the lattice contraction under illumination: 16,17 H is located in the tetrahedral sites in YH 1:9 , whereas an increase in H concentration up to YH 2:1 leads to a filling of octahedral sites, causing a compression of the crystal lattice. 27The verification of this hypothesis goes beyond the scope of this paper but can be done by probing the local structure with EXAFS before and after illumination.The significantly lower RGA signal resulting from H 2 during the second cycle may be related to the so-called memory effect, i.e., accelerated photodarkening of the previously illuminated and bleached sample, reported in Ref. 4. The second illumination results in a faster photochromic response since part of the ions (e.g., protons) migrate between the subphases as proposed in Refs.11 and 18, and thus, no gas release is observed.Nuclear magnetic resonance studies showed that 3% of hydrogen atoms are highly mobile and a corresponding signal of this species vanishes upon illumination, which can be related to H 2 release observed in the present work. 28

V. SUMMARY AND CONCLUSIONS
We show that the photochromic effect is triggered by all wavelengths ranging from 400 to 600 nm used in the present study, even though some of them are of energies smaller than the bandgap of the sample.Light of shorter wavelength is more efficient for photodarkening as it results in larger photochromic contrast despite its lower intensity in comparison to the sources of longer wavelength.The photochromic response scales linearly with the number of photons, which indicates a single-photon nature of the effect.For photons with energies exceeding the bandgap of the film, we find bleaching that is independent of the environment, while films bleached at photon energies on the order of bandgap only bleach once exposed to air.Films photodarkened for the first time release detectable amounts of hydrogen, indicating material transport during photodarkening.Our results are in line with the assumption of hydrogen transport from the oxide to the hydride phase, which has the potential to explain the photochromic effect.
. The energy of the photons affects the photochromic reaction in a similar manner as their flux [see Fig.1(c)].The strongest/fastest photochromic effect is observed for light with a wavelength of 450 nm.The decrease in energy yields weakening of the response, which almost fully vanishes for orange light (600 nm).A similar trend of photochromic response decreasing for a longer wavelength was observed in Ref.22.Note that the AOTF VIS delivers higher intensity for longer wavelength and photochromic response scales with the number of photons at a given wavelength [see Fig.1(b)].

FIG. 1 .
FIG. 1. Left panel (a): Transmission spectrum of sample YI plotted together with the wavelength from the super-continuum laser used for photodarkening in air.The dashed black line marks the wavelength corresponding to E dir g of the film.Central panel (b): ΔT after 180 s of photodarkening with 430 nm photons and after 180 s of bleaching plotted vs intensity.Blue and red dashed lines represent linear fits of ΔT darkening and ΔT bleaching , respectively.Right panel (c): Photodarkening and bleaching for different wavelengths plotted vs time.

FIG. 3 .
FIG. 3. Partial pressure of H 2 for photochromic films darkened with blue light (455 nm) in UHV.Both samples were darkened for the first time after production.The pressure drops rapidly toward the base pressure once the light source is switched off.During a second cycle of photodarkening, a significantly smaller gas release is found.

FIG. 4 .FIG. 5 .
FIG. 4. Normalized optical transmission of YII (a), (b), (c), and DI (d) after photodarkening 33 min with blue light in HV plotted as a function of bleaching time in HV (gray), H 2 (yellow), and O 2 (blue) as well as technical gas (green) and air (white).The red dashed line illustrates transmission after illumination.

TABLE I .
Photochromic films investigated in this study.D and Y symbolize dysprosium-and yttrium-based films.Values for samples YI and YIII are taken from Refs. 7 and 15, respectively.