Influence of Crystal Structure, Encapsulation, and Annealing on Photochromism in Nd Oxyhydride Thin Films

Thin films of rare earth metal oxyhydrides show a photochromic effect, the precise mechanism of which is yet unknown. Here, we made thin films of NdH3–2xOx and show that we can change the band gap, crystal structure, and photochromic contrast by tuning the composition (O2–:H–) via the sputtering deposition pressure. To protect these films from rapid oxidation, we add a thin ALD coating of Al2O3, which increases the lifetime of the films from 1 day to several months. Encapsulation of the films also influences photochromic bleaching, changing the time dependency from first-order kinetics. As well, the partial annealing which occurs during the ALD process results in a dramatically slower bleaching speed, revealing the importance of defects for the reversibility (bleaching speed) of photochromism.

(limited only by the substrate) since this energy is not sufficient to excite carriers across the band gap. Once this threshold (band gap) is reached, however, the transparency decreases to zero and the light is absorbed.
It has been shown in previous work that the optical band gap of RE-oxyhydrides is related to the O 2− :H − ratio, where more oxidised samples lead to larger band gaps [1]. Eventually, a fully oxidised sample (e.g., Nd 2 O 3 or Nd(OH) 3 ) have band gaps so large that they are outside the measurement range of the equipment used here.
The evolution of the optical band gaps of uncoated Nd-oxyhydride films, thus, shows that the band gap expands rapidly, and a fully oxidised film is formed within 2 days of air-exposure. Nd-oxyhydride films coated by ALD maintain a more stable optical band gap (therefore, composition) for at least 5 months.
We can compare a freshly ALD coated film to the change in transmission for an uncoated film. Although it appears that the band gap of the material expands during the ALD process, this new composition is maintained for at least 5 months.  Figure S2: Optical transmission spectra for some Nd-oxyhydride thin films produced at different deposition pressures (p dep = 0.6 − 0.9 Pa with a protective coating of Al 2 O 3 deposited onto the films by ALD. Unlike the uncoated films, these Nd-oxyhydrides are stable in ambient air for at least 138 days since the absorption edge shows little-to-no shift over time. This implies that no signficant oxidation occurs for these films, substantially extending their lifetime. Figure S3: (left) A comparison of uncoated and coated Nd-oxyhydride films deposited at 0.6

S5
Pa. The black line is the transmission spectrum of a film before ALD, and the blue line is after ALD. The band gap opens slightly during this process. Despite this oxidation, an ALD coated sample will maintain this composition for at least 138 days, unlike the uncoated film. (right) Tauc plots for the transmission data of ALD-coated Nd-oxyhydride films deposited at 0.6 Pa.
Fitting lines for day 0 and day 138 are shown, where the x-intercepts (indicating the optical band gap) are similar. Shifts towards the right are related to an expansion of the optical band gap.

II. Microscopy & imaging of ALD coating
Our NdH 3−2x O x films were coated by an ALD layer of Al 2 O 3 to protect them from rapid oxidation. Below, we show the characterisation of this layer by atomic force microscopy (AFM) where we show that it is conformal (Fig. S4).
However, although this coating indeed protects our films from complete oxidation for at least 5 months, we noticed that imperfections can occur sometimes, for example, in the presence of dust (Fig. S5). The coating will deposit on the dust particle, which can fall off later and reveal either the substrate or a part of unprotected NdH 3−2x O x . This results in pinholes in the coating which act as centres of oxidation. If there are enough pinholes, complete oxidation can take place. This is also a further testament to the positive function of the coating, without which, the samples cannot retain their composition.

S8
III. Air-oxidation conditions for measured films Table SI: List of all the oxyhydride samples considered in this work. p dep is the deposition pressure used to sputter the sample, while "s" and "f" are used to denote which samples exhibited slow or fast bleaching kinetics, respectively. "Heat" refers to the set of samples which were treated for controlled amounts of time on the heated ALD deposition stage. The temperature is the average temperature during air-oxidation, and the transfer time is the time spent transferring the samples from the sputtering vacuum chamber to the ALD vacuum chamber (where a protective layer was deposited).  shown that there is a critical deposition pressure (p * ), above which, the oxyhydride phase is formed [1,2]. Below we show the transmission spectra for two films made below this critical deposition pressure. Films made below p * should largely maintain their RE-dihydride composition and oxidise very minimally. This is visible especially for the ALD coated films, whose transmission spectra do not show any changes for at least 21 days. For the uncoated films, however, the transmission spectra in the initial days after removal from the vacuum chamber show very low transmission (characteristic of a RE-dihydride), but later oxidise a bit. This is true primarily for the sample deposited at 0.  Figure S10 for the same sample. According to the literature [3], the intensity of the (101) peak should be approximately 10 times lower than that of the (200) reflection. Based on that, we would expect the (101) reflection here to have an intensity of around 8 cps, which may be hidden under the substrate signal for incident angle 3.2 • , but visible for 1.2 • . However, others report an even lower intensity of the (101) [4,5].

VIII. Thin film texture & strain
Thin films can be analysed for their texture and macro-stress by measuring diffraction patterns at different values of ψ. The angle ψ defines the tilt of the sample perpendicular to the X-ray beam, allowing for the measurement of crystallites of different orientations.
We measured four NdH 3−2x O x films with Bragg-Brentano (θ-2θ) geometry and a ψ angle varying between 0-80 • . The full XRD patterns for this are shown in Figure S12. These four films can be compared to assess the influence of not only p dep , but also τ B,50% since two of the films showed "slow" kinetics, and two showed "fast" kinetics (specified in Table SII).
Growing thin films can develop a "texture", meaning that, although the film is polycrystalline, these crystals have a preferred out-of-plane orientation. This increases the intensity of some reflections, while decreasing the intensity of others. In extreme cases, some reflections may even disappear, as we see for our NdH 1.9+δ film (Fig. S9).
To examine if our ALD coated NdH 3−2x O x films are indeed textured, we compared the intensity ratios of the (200) and (111) Figure S13. In case of random orientation, we expect a value of 0.55-0.67 for δ based on the structure factor and depending on composition. We find that δ changes slightly with p dep which is a consequence of the differing compositions (structure factors) of these samples. As well, δ changes with ψ as different planes satisfy diffraction conditions with the change in tilt. Thus, we conclude that our films are only slightly textured.
Macro-strain is quantified by assessing the peak shifts for various ψ angles. This is done in Figure S14, where the peak shifts are displayed as (d − d o )/d o , and d o is the value at ψ = 0 • . This can be calculated for all the reflections which appear in the XRD pattern, however, the (220) and (222) reflection intensities were too low and reliable results could not be obtained. The slopes of the lines in Figure S14 give an indication of the macro-strain that is present in our films. Notably, the determination of macro-strain by this method relies on the assumption that the stress is uniform, isotropic, and biaxial. These assumptions are not true for weakly textured films, even so, they provide some valuable insights into the microstructural properties of our films. S17   Gd-based films shown in a double logarithm plot. Previous reports have used the slope of such a graph to obtain the negative reciprocal of the first-order bleaching rate constant [1,2,6]. This is in agreement with the bleaching of the uncoated sample, but does not describe the coated film. Additionally, the Y-and Gd-films exhibit a cubic crystal structure, so only one lattice constant (a) is given for them.