Defect generation in Pd layers by ‘smart’ films with high H-affinity

In this paper, we demonstrate that the microstructure and the surface of a thin palladium (Pd) film can be intentionally altered by the presence of a subjacent niobium (Nb) film. Depending on the thickness of the Nb film and on the hydrogen gas pressure, defects in the Pd film can be healed or created. To demonstrate this effect, Pd/Nb/sapphire (Al2O3) stacks are studied during hydrogen gas exposure at room temperature by using scanning tunneling microscopy (STM), X-ray diffraction (XRD) and environmental transmission electron microscopy (ETEM). STM shows that hydrogen-induced topography changes in the Nb films depend on the film thickness which affects the height of the Nb surface corrugations, their lateral size and distribution. XRD measurements show that these changes in the Nb hydride film influence the microstructure of the overlaying Pd film. ETEM reveals that the modifications of the Pd film occur due to the precipitation and growth of the Nb hydride phase. The appearance of new defects, interface and surface roughening is observed in the Pd film above locally grown Nb hydride grains. These results can open a new route to design ‘smart’ catalysts or membranes, which may accommodate their microstructure depending on the gaseous environment.

thickness 37,40 , as the coherency between the hydride and the α-matrix changes. Precipitates are coherent with the matrix for 10 nm < d 0 ≤ 37 nm 40 and incoherent for d 0 ≥ 40 nm 37 . For coherent precipitates, the lattice planes between the hydride precipitates and the α-matrix match by local strain, while for incoherent precipitates dislocations are present to adjust for the mismatch between the lattice planes 41 . The two states result in different surface topographies in the two-phase region of the Nb-H thin films system, with changes in lateral spreading as well as in the local topographical heights 33,37 .
The landscape of the morphological changes on the Nb film is predicted by linear elastic theory 41,42 ., Therein, an out-of-plane film expansion z for a hydrogen-absorbing Nb-film grown on a rigid substrate is described by ∆ = .
⋅ ∆ ⋅ z c d 0 136 theor H o 29,39 , where Δc H is the change of the hydrogen concentration within the film. This can be used to determine the surface topography changes expected for an incoherent phase transformation 41,42 . In this case, Δc H corresponds to the width of the miscibility gap between two related phases. In the coherent regime, the out-of-plane expansion is less than expected in the incoherent regime (∆z theor ) 41 , thus it turns out that the amplitude of local topographic changes caused by hydrogen absorption can be maximized by choosing the incoherent regime (d Nb > 40 nm). The amount and length (perimeter) of bordering regions (modified regions of the interface) can be tuned by changing the lateral spreading of hydrides. For example, a maximal length of bordering regions can be created by focusing on the coherent phase transformation regime.
In the present study, we demonstrate experimentally that the lattice expansion during hydrogen absorption in Nb-films modifies the microstructure of an overlying Pd layer in the Pd/Nb/ sapphire (Al 2 O 3 ) system. Variation of the Nb-film ('smart layer') thickness allows for tuning of the out-of-plane expansion and the Pd/Nb interface roughness, causing local changes (bending) in the Pd top-layer. The border regions between the Nb α-phase and the Nb-hydride phase are suggested as origins for defect generation in the Pd top-layer and bending of its surface. We use a combination of in situ characterization techniques to support our hypothesis.

Results and Discussion
The results of STM measurements performed during hydrogen gas exposure of a Pd0.2nm/Nb40nm/Al 2 O 3 film (a)-(c), (g) in the incoherent regime, and a Pd0.2nm/Nb25nm/Al 2 O 3 film (d)-(f), (h) in the coherent regime are shown in Fig. 1. Upon hydrogen gas exposure at p H = 1.6 × 10 −6 mbar, different hydride-related local topographical changes (bright regions) occur. Figure 1(a) shows the surface topography before the hydrogen exposure, Fig. 1(b) -during hydrogen exposure at p H = 1.6 × 10 −6 mbar (t e = 100 min), Fig. 1(c) -during hydrogen exposure at p H = 3 × 10 −6 mbar (t e = 135 min), and Fig. 1(g) -a line scan taken along the black dotted line shown in Fig. 1(b). The lateral size of the surface corrugations visible in Fig. 1(b) and (c) varies from 100 nm to about 1000 nm. We note that the lateral sizes and the distribution of initial film corrugations (dark dots in Fig. 1(a)) do not correlate with the lateral size and the distribution of the hydrides in Fig. 1(b). Thus, the initial film corrugation influence is here regarded as being of minor importance. The average distance between the neighboring elevated regions is large in Fig. 1(b), and ranges from 500 nm to about 750 nm. An example of a line scan taken across one of the hydrides shown in Fig. 1(b) is given in Fig. 1 For the 40 nm film, only a few relatively large hydride regions are detected in the area of interest. The chosen frame is representative of the complete film. Similar results were obtained for thicker Nb films 30,33 . The perimeter of the border region between the hydride and the α-phase region can be determined by analyzing STM images Thus, the spatial distribution of hydride-induced surface corrugations as well as their typical height amplitude and lateral size can be changed by varying only the thickness of the Nb film. In comparison to the 40 nm film, the 25 nm film has a higher density of hydrides, as visible in Fig. 1(c and f). The volume fraction of the hydride phase in Fig. 1(f) is about 30% and the length of the border region is about 2.3 • 10 −2 nm −1 . Extrapolation to a hydride volume fraction of 41% (resembling the hydride volume fraction of the 40 nm film) would result in a length of the border region of about 3.2 • 10 −2 nm −1 . Thus, for similar volume fractions of the hydride phase, the relative length of the border regions is approximately 5 times higher for the 25 nm film as compared to the 40 nm film. This means that the length of the border region increases by reducing the film thickness.
Due to the phase transformation in the Nb-film, microstructural and topographical changes will appear in the Pd layer. XRD measurements of these changes are given in Fig. 2. Figure 2(a) shows XRD patterns obtained for different stages of hydrogen exposure of a Pd20nm/Nb55nm/Al 2 O 3 thin film system. The black arrow added on the right side of this graph provides the time axis of the experiment with arbitrary scale and the hydrogen pressure. The bottom curve shows only the Nb (110) peak at 2θ = 26.35° and the Pd (111) peak at 2θ = 27.25° (Fig. 2a). They correspond to the Nb-H α-phase and the Pd-H α-phase, respectively. During hydrogen exposure, the Nb-H α-phase peak shifts towards smaller angles, reflecting the lattice expansion during hydrogen absorption. Starting with the third pattern from the bottom, an additional peak appears, weak at first, at 2θ = 24.96° corresponding to the Nb-H hydride phase. This peak's intensity increases while the intensity of Nb-H α-phase related peak at Scientific RePORts | 7: 9564 | DOI:10.1038/s41598-017-09900-z 2θ = 25.95° decreases, as commonly observed during the phase transformation. The phase transformation from the Nb-H α-phase to the Nb-H hydride-phase occurs in the pressure range from p H2 = 1 × 10 −3 mbar to p H2 = 5 mbar. At p H2 = 5 mbar the phase transformation was completed as confirmed by a disappearance of the Nb-H α-phase peak. It should be noted that this hydrogen pressure is much higher than the phase transition pressure of about p H2 = 1 × 10 −6 mbar found by STM measurements, even for films with similar (20 nm) Pd layer thickness. We attribute the pressure difference to the Pd surface or interface conditions that change when samples encounter environmental conditions outside the vacuum chamber. The position and the full width at half maximum (FWHM) of the Pd (111) peak at 2θ = 27.25° change during hydrogen loading, despite hydrogen absorption occurring mainly in the Nb film (Fig. 2b). The Pd-peak shifts from 27.257° to 27.278°, while the FWHM changes from 0.333° to 0.337° (marked by yellow color). The FWHM is determined by a Gaussian fit to the Pd (111) peaks, shown in Fig. 2(a).
By further hydrogen exposure, to p H2 = 10 mbar and p H2 = 20 mbar (two upper curves in Fig. 1(a), hydrogen absorption in the Pd film (Pd-H α-phase) results in a reverse peak shift to 27.21° and an increase of the FWHM to about 0.372°. Consequently, the results of the XRD measurements suggest that the Pd layer undergoes microstructural changes resulting from the phase transformation occurring within the Nb-H film (marked by yellow color in Fig. 2(a) and (b)).
Decreasing the Nb film thickness causes different microstructural changes in the Pd film. Figure 2(c) and (d) present the results of XRD measurements on the Pd20nm/Nb28nm/Al 2 O 3 exposed to a hydrogen gas pressure of p H2 ≤ 5 mbar. The phase transformation occurring in the Nb-H film causes a peak shift of the Nb (110) peak from 26.18° to 24.64°. Here, during the phase transformation only an intermediate peak broadening is detected with no separated hydride-related peak. This result is related to the coherent phase transformation in the Nb-H film 37 . At the same time, the Pd (111) peak slightly shifts from 27.221° to 27.238°. The FWHM also changes, from 0.316° to 0.349°, first decreasing from 0.316° to 0.309°, followed by an increase to 0.349°. However, the relative Pd peak shift for both samples (Fig. 2) stays almost the same and corresponds to a lattice contraction of d < 0.1%. At the same time, the change of the FWHM of the Pd peak, ΔFWHM, is 9 times higher for the Pd20nm/Nb28nm/Al 2 O 3 (ΔFWHM = 0.0325°) than that for the Pd20nm/Nb55nm/Al 2 O 3 (ΔFWHM = 0.0037°). This indicates a higher defect density in the Pd film of the Pd20nm/Nb28nm/Al 2 O 3 system.
In order to characterize the microstructural changes in details, HRTEM measurements using cross-section specimens were performed during hydrogen gas exposure. Figure 3 shows HRTEM images of the Pd20nm/ Nb50nm/Al 2 O 3 taken before (Fig. 3a) and during hydrogen exposure at p H2 = 5 mbar (Fig. 3b). Fast Fourier transform (FFT) analysis of the rectangular regions highlighted in both images (see insets) clearly shows additional weak spots corresponding to an ordered hydride phase in Fig. 3(b). These spots were used to localize the hydride precipitate, as marked in red color in Fig. 3(c). The precipitate starts to grow at the Pd/Nb interface towards Nb/ Al 2 O 3 interface. The hydride phase causes the local expansion of the Nb film from 50.3 nm to 51.4 nm (Fig. 3b).
The expansion of the Nb film results in the modification of Pd/Nb interface and in the microstructural changes of the Pd film. The enlarged images of the Pd/Nb interface (selected area in Fig. 3c) before hydrogen exposure and at 5 mbar H 2 are shown in Fig. 3(d and e), respectively. As compared to the initial position of the Pd/Nb interface marked with a yellow line in Fig. 3(d), the topographical changes of the interface of about 1-2 nm height are visible in Fig. 1(e). The interface is strongly roughened by the growth of the hydride. Moreover, the appearance of stacking faults in the Pd layer is observed close to the interface, as shown by the yellow arrows (Fig. 3(d and e)). The lateral distance between the newly formed stacking faults is similar to that between of the surface hillocks on the surface of the hydride precipitate. Additionally, the number of low angle grain boundaries decreases, as shown by red arrows in Fig. 3(d) and (e). The HRTEM images also have a typical contrast caused by the appearance of strain induced by the bending of the Pd/Nb interface Fig. 3(e). An additional effect that should be considered is that the Pd film also undergoes microstructural changes because of Pd interaction with H 2 . Hydrogen environment gives rise to an enhanced mobility of Pd-atoms at the surface as well as dislocation mobility and defect healing. A particular example of hydrogen-induced defect healing in Pd films is shown in Fig. 4. It shows HRTEM images of the Pd12nm/Nb15nm/Al 2 O 3 taken before the hydrogen exposure (Fig. 4a) and during hydrogen exposure at p H2 = 1 mbar (Fig. 4b) in a sample area without border regions. A twin boundary between two adjacent domains (red arrow) and the stacking faults (yellow arrows) are highlighted (Fig. 4a). In the pressure range from 0.01 mbar to 0.1 mbar the stacking faults parallel to the surface of Pd film start to disappear (See Figure Sl) and they are not visible anymore at p H2 = 1 mbar (Fig. 4b).
This result suggests that also the leading partial dislocations of the stacking faults become mobile with the presence of hydrogen. The vertical twin boundary (Fig. 4) doesn't move but its morphology is remarkably changed, as can be seen in Fig. 4(a) and (b). This may be attributed to partial dislocations being retracted into the boundary thereby removing the stacking faults. We note that the contrast change of the Nb film between Fig. 4(a) and (b) indicates a change of the stress state caused by hydrogen inside the Nb film.
Morphological changes of Pd film may be enhanced by decreasing the thickness ratio d Pd /d Nb. A particular example with d Pd /d Nb = 2 nm/55 nm = 0.04 is shown in Fig. 5, which summarizes STM measurements performed during hydrogen gas exposure of a Pd2nm/Nb55nm/Al 2 O 3 . Figure 5(a) shows the surface topography before hydrogen exposure, Fig. 5(b) -during hydrogen exposure at p H = 1.6 × 10 −6 mbar (t e = 1220 min), and Fig. 5(c) -line scans taken along the black dotted line in (a) and (b). The STM measurements (Fig. 5b) show that the phase transition from Nb-α phase to the Nb-hydride phase was mainly completed in the loading stage, as the majority of the surface observed is expanded with respect to the initial stage. A significant surface roughening from 0.4 nm to 0.9 nm was detected upon hydrogen exposure to p H = 1.6 × 10 −6 mbar. Line scans taken along the dotted lines in Fig. 5(a) and (b) provided in Fig. 5(c) (black line: before hydrogen exposure, red line: after exposure to p H = 1.6 × 10 −6 mbar) shows the appearance of high topographic corrugations, e.g. steps and hillocks. As hydrogen-related lattice expansion in Pd is negligible at this pressure, the measured height change on the Pd-surface is mainly related to the Nb-lattice expansion.
In summary, our experiments show that the microstructural and topographic changes in the Pd-films used as a catalytic layer in Pd/Nb/Al 2 O 3 systems are caused by the underlying Nb film and depend on its thickness. We suggest that similar effects will also show up in different multilayer systems containing hydrogen-absorbing materials. These materials can change their structure or phases depending on the environment, for example in gas sensors applications. The border regions between the different phases (present during the phase transition) appear to act as centers for bending and defect generation in the overlaying Pd-film. 'Smart layers' that can form hydrides may provide a means of tuning the microstructure of a coupled metal with low H affinity and may be interesting for a variety of practical applications. For example, for membrane applications the microstructural changes can be prevented by choosing an appropriate 'smart' layer thickness. Likewise, the microstructural features can be designed to be pronounced for the applications, such as recovery of surface activity of catalysts.
Characterization. STM measurements on Pd/Nb/Al 2 O 3 films were carried out using a UHV Micro-STM (Omicron). For hydrogen loading experiments, a Pd catalyst of 0.2 nm (islands) and a Pd film of 2 nm (closed layer) thicknesses were deposited onto Nb films. Measurements were performed directly after film preparation and without breaking the UHV conditions 30 . The STM images were obtained without changing the position Scientific RePORts | 7: 9564 | DOI:10.1038/s41598-017-09900-z of analysis (in situ). During experiments, the hydrogen pressure was increased in steps from 1 × 10 −9 mbar to 5 × 10 −6 mbar.
XRD measurements were performed on Pd/Nb/Al 2 O 3 at the synchrotron facilities at the ESRF (Beamline BM20) and at Petra III (Beamline P08). During these measurements, the hydrogen pressure was changed stepwise from 1 × 10 −6 mbar to 20 mbar. XRD patterns were measured in situ at different hydrogen pressures and loading times 30 .
Transmission Electron Microscopy (TEM) measurements were performed by using a FEI Titan 80-300 environmental transmission electron microscope (ETEM) operated at 300 kV, and equipped with C s -image corrector. During the measurements, hydrogen pressures ranging from 5 × 10 −7 mbar to 5 mbar were applied. Most of time the electron beam was blanked to minimize its influence on the materials. For XRD and ETEM measurements, a 20 nm-thick protective Pd layer was deposited to account for sample transport in air.