Challenges in quantifying Pt concentrations in Pd alloys by using secondary ion mass spectrometry: Strong grain orientation effects

Palladium–platinum alloys were analysed by dynamic secondary ion mass spectrometry (SIMS) to investigate grain orientation effects that gave differences of up to 400% in the Pt/Pd count rate ratios, even within the same grain upon small rotations of a Pd sample with 1 wt% Pt. The sample had been homogenized by annealing, and the homogeneity was confirmed by X‐ray analysis in scanning electron microscopy (SEM). Grain orientations were determined by electron backscatter diffraction (EBSD). Crater depths were measured by white light interferometry (WLI). SEM images from the bottom of SIMS craters made in the same grain after small rotations around the sample surface normally showed different patterns of microfaceting for some rotation angles, probably exposing low‐index crystallographic planes. A complete understanding of the observed grain orientation effect is still lacking. However, factors such as ion mass, sputter rate, ion channelling, ion focusing, preferential sputtering, surface height, crater microfaceting and/or angle‐dependent sputtering seem to play a role. For these Pd–Pt alloys, the strong grain orientation effect adds another level of complexity when attempting to quantify concentrations and obtain depth profiles by SIMS. Without proper sampling and/or averaging, one could reach very wrong conclusions when comparing results from different samples.

Palladium-platinum alloys were analysed by dynamic secondary ion mass spectrometry (SIMS) to investigate grain orientation effects that gave differences of up to 400% in the Pt/Pd count rate ratios, even within the same grain upon small rotations of a Pd sample with 1 wt% Pt.The sample had been homogenized by annealing, and the homogeneity was confirmed by X-ray analysis in scanning electron microscopy (SEM).Grain orientations were determined by electron backscatter diffraction (EBSD).Crater depths were measured by white light interferometry (WLI).SEM images from the bottom of SIMS craters made in the same grain after small rotations around the sample surface normally showed different patterns of microfaceting for some rotation angles, probably exposing low-index crystallographic planes.A complete understanding of the observed grain orientation effect is still lacking.However, factors such as ion mass, sputter rate, ion channelling, ion focusing, preferential sputtering, surface height, crater microfaceting and/or angle-dependent sputtering seem to play a role.For these Pd-Pt alloys, the strong grain orientation effect adds another level of complexity when attempting to quantify concentrations and obtain depth profiles by SIMS.Without proper sampling and/or averaging, one could reach very wrong conclusions when comparing results from different samples.

| INTRODUCTION
Secondary ion mass spectrometry (SIMS) is a common technique for studying surfaces and near-surface properties of materials.In dynamic SIMS, detection levels of trace elements in a matrix can be down to the ppb (parts per billion) level. 1 However, SIMS is often described as a 'semi-quantitative' method in the sense that one can easily obtain images showing regions with low or high amounts of some element, but that quantifying the concentration is difficult.In all applications where SIMS is used for quantification of concentrations, it is important that the results are both correct and repeatable, within certain limits.We here present evidence that quantification of Pt by SIMS analyses of Pd-Pt alloys may give differences in concentrations by a factor of four, depending on small changes in the sample orientation.Without being aware of such effects, one could easily reach wrong conclusions when comparing bulk concentrations or depth profiles from different samples of Pt in Pd.It is not unlikely that the observed effects may apply to other types of materials as well.M.T.D. Wingate and W. Compston 2 found that the 206 Pb/ 238 U ratio measured by ion microprobe in Baddeleyite (ZrO 2 with small amounts of HfO 2 , TiO 2 , FeO 2 and SiO 2 ) could vary by up to ±10% or more in a systematic way, depending on the direction of the primary ion beam relative to the Baddeleyite crystals.This grain orientation effect caused problems in using the lead/uranium ratio to determine the age of rocks containing Baddeleyite.Wingate and Compston suggest that 'channelling of primary ions into the crystal, emission of secondary ions along preferred directions, and/or differential ionization of secondary species' might be involved in the crystal orientation effect.They did not explain how these phenomena affected 206 Pb and 238 U differently.However, they found some systematic trends in the 206 Pb/ 238 U ratio with crystal angle relative to the ion microprobe.
Schmitt et al. also found grain orientation effects in Baddeleyite. 3ey showed that a correlation in the 206 Pb/ 238 U ratios for repeated analysis of the same grains disappeared when the sample was rotated 90 in the SIMS instrument.
Jason M. Huberty et al. 4 found a small but consistent spread of about 6‰ (±2 Standard Deviations) in the 18 O/ 16 O ratios among different grains of magnetite, compared with 0.8‰ for measurements within single grains.As an explanation for this grain orientation effect, they 'hypothesize that secondary ions emitted along channelling and focusing directions may have their trajectories deviated by the electrostatic field and could preferentially select the light isotope ( 16 O)'.
They also mention the possibility that 'surface topography can deform the equipotential surfaces of the electrostatic field parallel to the sample surface, which in turn deviate the path of secondary ions'.
In addition to these relatively small grain orientation effects, a commonly quoted challenge when quantifying SIMS data is the socalled 'matrix effect'. 5This means that the SIMS count rate for a given isotope depends on the presence of all other elements in the sample.This complicates quantification since one cannot simply convert count rates into concentrations by a general formula.Instead, one ideally needs reference samples of known concentrations that are similar to the 'unknown' samples under study to get reasonably good quantification.The spread in SIMS count rates described here represents yet another level of complexity when quantifying SIMS results.
A third effect that will complicate SIMS quantification is surface height.A paper by Jonathan W. Angle et al. 6 describes a SIMS study of nitrogen in niobium where an apparently large grain orientation effect is strongly reduced by careful centring of the secondary ion beam relative to the SIMS instrument.They used a Cameca F7 Geo magnetic sector SIMS instrument equipped with deflection plates in the ion transfer optics, which allowed the secondary ion beam to be centred in the contrast aperture.By calculating relative sensitivity factors (RSFs) of ion-implanted NbN standards based on data without and with ion beam centring, they could in the most extreme case lower the spread in RSF values (Standard Deviation/Average value) from 40% to 10%.The authors note that the sample flatness and crater topography are also important.A height difference of as little as 1.5 μm is enough to change the trajectory of the secondary ion beam, potentially affecting the ion count rates if the beam is not centred.
In a recent follow-up paper, J. W. Angle et al. 7 further investigate the effect of surface height and grain orientation in NbN samples.They use a special sample holder that reduces the expected deflection in the face plate of 10 μm when using a standard sample holder to about 5 nm.They obtain a four-fold reduction in the RSF uncertainty with this new holder.In a systematic study of the grain orientation effect where a bicrystal sample is analysed by SIMS in steps of 15 around the full circle, the authors find that a randomly oriented crystal displays a clear peak in the RSFs around one specific angle, with a four-fold increase compared to outside the peak.This result was attributed to channelling of the primary beam ions along a {101} plane.The channelling effect was further accompanied by a drastic reduction in the sputter rate.
The current work was part of a larger study on the use of palladium wire gauzes to catch platinum which escapes as PtO 2 (g) during industrial catalytic oxidation of ammonia.The initial aim of the study was to use dynamic SIMS to obtain quantified depth profiles of Pt in Pd model samples after exposure to PtO 2 vapour under various conditions.Given the results presented here, we decided to rather use secondary neutral mass spectrometry (SNMS) 8 to quantify our samples since SNMS does not have the same matrix effect that makes SIMS quantification difficult.Further, the analysis area in SNMS was 2 mm in diameter, which will average the contribution from many Pd grains in our polycrystalline samples.

| Samples
The main sample was shaped like a cylinder with 5.6 mm diameter and 2 mm thickness.It was made of an alloy containing nominally 99 wt% (percent by weight) palladium and 1 wt% platinum.The sample had been cast, cut into a disk and polished to mirror finish on one side.An arrow was engraved in the periphery of the polished surface to facilitate the orientation of the sample during microscopy.In addition, 10 indentations were made with a Vickers hardness tester (50 g load, 5 s duration, 0.2 mm separation) to uniquely mark this sample relative to other samples with different compositions.This sample got nine indentations in a 3 by 3 grid, plus one extra indentation 0.5 mm away to break the four-fold rotational symmetry.
The sample was homogenized inside double, evacuated quartz tubes for 1 week at 1100 C in order to even out any compositional inhomogeneities from the casting process.During this heat treatment, a weak restructuring of the polished surface took place.Narrow valleys formed along the grain boundaries and some waviness developed in certain grains.
In the initial stage of this study, samples similar to the one described above, but with different concentrations of Pt in Pd, were analysed.The nominal concentrations for the full set of reference samples were 0.1, 1, 5, 20, 40, 55 and 70 wt% Pt in Pd.This corresponds to atomic percentages for Pt of 0.05, 0.55, 2.8, 12, 27, 40 and 56 at%, respectively.
For the later stages of the investigation, the main sample with 1 wt% Pt was exposed for 24 h to low concentrations of PtO 2 gas at 900 C in order to deposit a layer of platinum on the sample surface and study the diffusion profiles of Pt into the Pd crystals.

| Microscopy
The grain orientations were initially analysed by electron backscatter diffraction (EBSD) using a scanning electron microscope (SEM) with an EDAX system running the OIM software for determining the crystal orientations.In addition to producing images where the grain orientations were colour coded, the software also generated text files that for each grain gave the Euler angles of the cubic Pd crystals, thus specifying the directions of the crystallographic axes relative to the sample coordinate system.We developed a Microsoft Excel sheet where these Euler angles allowed us to translate the coordinates of a vector given in the sample coordinate system to the crystal's coordinate system.Using this Excel sheet to calculate the crystallographic direction in each Pd grain of the sample surface normal, we could confirm that the angle calculations gave orientations consistent with the colours in the EBSD maps.
As a double check, the main sample was later analysed again by EBSD in a different SEM instrument, Nova NanoSEM 650 field emission gun scanning electron microscope (FEG-SEM) from FEI.Chemical compositions were measured using an X-max-50 SDD energy dispersive spectrometer (EDS) from Oxford Instruments.Crystallographic information was obtained using a Nordlys detector, also from Oxford Instruments.When compensating for differences in the Euler angles that were multiples of 90 -which are equivalent for a cubic crystal like palladium-the two EBSD analyses gave the same grain orientations within less than 2.5 .
After EBSD maps had been obtained for the entire sample surface by putting together the individual EBSD images into a mosaic, the centre of each grain was marked with a Vickers hardness indenter.This was to ensure correct positioning of the craters in the subsequent SIMS analysis.Cs + primary beam that promotes the creation of negative secondary ions.Since the Cesium gun was not available during the first part of the study, we used the oxygen gun and counted Pd + and Pt + ions.
Although the count rates for Pt + were one to four orders of magnitude lower than the Pd + rates and more noisy, we still obtained acceptable counting statistics even for the Pt + signal.
Crater depths were measured using a WYKO NT9800 white light interferometer (WLI) in the vertical scanning interferometry (VSI) mode.In most cases, all three craters within one grain on the main sample were imaged using the 10Â magnification lens.But in cases where more lateral details were required, for example, if a grain boundary passed through a crater, the 50Â lens with field of view selector 0.55Â was used, giving images with 230 μm by 172 μm image size.Because of the flat, polished surface, it was easy to measure the depth of each crater: Masking the craters, fitting a plane to the remaining surface, and subtracting this plane from the original image would effectively reset the polished surface height to zero.The NT9800 software has a feature where one can get the average height of a rectangle within the image.This was used to find the depth of the flat parts of each crater with an accuracy of about 0.01 μm.

| RESULTS AND DISCUSSION
The analysis proceeded in three stages, with ever more targeted experiments in an attempt to understand the variability observed in the initial study of reference samples.The three stages of SIMS analyses were as follows: A. SIMS depth profiling of homogeneous reference samples with different concentrations of Pt in Pd from 0.1 to 70 wt%.This analysis did not give the expected linear relationship between concentrations and Pt/Pd count rate ratios from SIMS.
B. In order to check if grain orientation effects could explain the nonlinearity, we focused on the 1 wt% Pt sample, which had many grains of different orientations as determined by EBSD.Three craters in each of 20 selected grains were analysed by SIMS.The Pt/Pd count rate ratio within each grain was nearly constant, but we found large differences between grains.The Pt/Pd ratio differed by as much as a factor of four.This strong, grain-specific effect could be due to the crystal orientation itself (i.e., the crystallographic surface normal of each grain) or more specifically the direction of the incoming and outgoing ion beams in the SIMS relative to the Pd crystal axes.
C. To distinguish these two effects, we measured the Pt/Pd ratio within the same grain, but with different rotations of the sample around its surface normal.This gave different orientations relative to the SIMS primary beam (which came in from one side at a 38 angle as mentioned above).We chose a grain with an orientation close to h001i where a 90 rotational symmetry was anticipated.
Also within this one grain, the Pt/Pd count rate ratio differed by a factor of four, even over small angular rotations.No 90 rotational symmetry was found.SEM images that were made inside the SIMS craters revealed microfaceting for certain sample rotation angles.
The below figures summarize the findings we made during these three stages of analyses.Several hypotheses were tested to explain the observations, but most of the hypotheses could be dismissed.
Detailed descriptions of these tests have been put into four separate supplementary sections and the results mentioned only briefly in the main text.

| Stage A
Figure 1 shows the SIMS count rates as a function of nominal Pt concentration for the series of homogenized reference samples.Three craters were sputtered in each sample.The upper clusters of data points (blue), in the range of 100,000-3,000,000 counts/s, are from Pd isotopes, whereas the lower count rates between 30 and 3000 counts/s (red) belong to Pt isotopes.With a logarithmic scale on the vertical axis, the data points for the different isotopes more or less overlap since they differ by less than a factor of three for the Pd isotopes and less than five for the chosen isotopes of Pt.The black, diamond symbols show the corresponding sputter rates for each crater, as measured afterwards by WLI.The sputter rates are given in depth per total charge of primary ions (millimetres per Coulomb) to compensate for any variability in the primary ion current for different craters.This convention is used here to make the analyses comparable.The crater sputter rates are roughly the same within a factor of three.However, the data for Pd and Pt count rates do not fully show the expected steady increase for the Pt signal and the corresponding decrease for the Pd signal as the samples get richer in Pt.Instead, the Pd tends to fluctuate around 300,000 counts/s even if the Pt concentration increases from 20 to 70 wt%.
This was not promising for finding general RSFs that could be used to quantify the amount of Pt in Pd from our SIMS data.Section S1 describes how RSFs were derived, and how assuming one common RSF for Pt in Pd did not give satisfactory quantification in this case.
Firstly, for the same sample, the spread in calculated concentration was typically 20 at%, depending on where in the sample the data were obtained (Figure S10).Secondly, the average atomic percentage did not follow the expected curve, deviating erratically by up to 20 percentage points.

| Stage B
To investigate this double scatter in the Pt concentrations, which was detrimental to our attempts of obtaining quantified depth profiles in Pd-Pt samples, we studied the 99 wt% Pd + 1 wt% Pt sample in more detail.This sample was chosen because the matrix effect should be much smaller than in the high-Pt samples and because EBSD analysis showed that this sample had many grains with a large variety of crystal orientations; see Figure 2. A WLI image that covered the full diameter of the sample showed that the polished surface was smooth, but slightly convex with a height difference of 2.6 μm between centre and periphery.
Twenty grains with a large enough size and a representative spread in crystal orientations were selected for the second SIMS F I G U R E 1 Raw SIMS count rates for the seven Pd-Pt reference samples with nominal Pt concentrations of 0.1, 1, 5, 20, 40, 55 and 70 wt% Pt.The three samples with the lowest concentrations were analysed with a larger contrast aperture (CA2) than the samples with 20-70 wt% Pt (CA3), hence the higher count rates.
analysis.Three craters were sputtered inside each grain and marked by the primary ion beam so each crater could be uniquely identified later.The markings are visible as white dots in Figure 3.
In Section S2, we show how we made deep pits at two locations on this sample by keeping the primary beam stationary and sputtering for 1 h.These pits were used to confirm that the direction and angle of the primary beam were as expected relative to the sample.This primary beam direction was used together with the EBSD grain orientations to look for systematic trends between directions in the crystal and the primary beam, but no clear trends were found.By dividing all count rates by each isotope's abundance fraction, the signal intensity for each isotope could be plotted as if that isotope Mosaic of EBSD maps from the initial grain analysis of the 1 wt% Pt sample, with grain numbers as defined by the OIM software.This sample had some large, but also many smaller grains of different crystal orientations, covering most of the colours shown in the stereographic triangle at the lower right.
F I G U R E 3 Optical macroscope image of the 1 wt% Pt sample after the second SIMS analysis.Large green squares show the locations of the craters from the first SIMS analysis.Blue squares with differing shades mark three SIMS craters that were made in each of the 20 selected grains for the second investigation.Small white dots below each crater were made by letting a stationary primary ion beam drill a hole for some seconds at one location.The number of dots shows the crater number within each grain.Vickers indentations used for locating the grains are highlighted with yellow diamonds.The red squares show the location of the two deep SIMS pits used to verify the primary beam direction and angle, see Figures S11 and S12 in Section S2.The arrow that was engraved by hand to ensure easy orientation of the sample can be clearly seen at the left.
were present alone; see Figure 5.This figure again shows a reasonable consistency among the isotopes counted in each crater.There is also a nearly as good consistency in the data for each of the three craters measured in the same grain.However, from one grain to the next, both the Pd count rate and the Pt count rate may differ significantly.
For some grains (e.g., the three grains of craters 7 to 15), the Pt rates are quite similar, whereas the Pd count rates differ by a factor of three.For other grains (like craters 28 to 36), the Pd signal is nearly constant, whereas the Pt count rates differ significantly.These data were obtained on two consecutive days with the same microscope settings, except for the daily mass calibration.Thus, there must have been something in the sample that from one grain to another affected the detection probability of Pd ions and Pt ions differently and independently.
A subsequent SEM study with X-ray analysis (EDS) inside each of the 60 SIMS craters found no evidence of real concentration differences in the homogenized sample.On the contrary, based on three EDS analyses within each crater, the overall Pt concentration was (0.97 ± 0.07) wt% (average ± one sample standard deviation).There were no significant differences in the EDS concentration among the grains at a confidence level of α = 0.01.Thus, sample inhomogeneity could not explain the large differences in the SIMS signals for Pd and Pt in different grains.Still, when averaging over the isotopes and then calculating the ratio of the Pt count rate and Pd count rate, we get a scatter summarized in Figure 6 that would correspond to SIMS concentration differences of a factor of four for Pt in the same sample.In Figure 6, we see that the data from within the same grain are much more consistent than between grains, with the largest difference in Pt/Pd for one grain being a factor of 1.4 at craters 22-24.A SEM study of crater 24 (the third crater in grain number 60) revealed a sub-grain boundary in the lower right corner of the crater, which is probably the cause of this anomaly.
The most striking correlation in Figure 5 is perhaps the opposite trend of the Pd signal (top) versus the sputter rate (bottom, black diamonds), where a particularly high Pd count rate is generally associated with a particularly low sputter rate, and vice versa.A more subtle but positive correlation can be seen between the Pt signal and the sputter rate for craters 28-36 where the Pd signal is nearly constant: Here, a higher Pt count rate is associated with a higher sputter rate.These two correlations are investigated further in Section S3.We there find a linear decrease in the Pd signal with sputter rate and a linear increase F I G U R E 4 Depth profiles of all the studied isotopes for crater 1 in grain 158.There is a slight reduction in the intensity over time, but the signal is constant enough that a simple averaging over time between the vertical green lines was used to obtain the count rate values from each grain and crater.

F I G U R E 5
The abundance corrected Pd and Pt count rates for each of the measured isotopes among the 20 selected grains of the main sample with 1 wt% Pt.The sputter rate of each crater is also included, calculated from crater depth measurements by WLI divided by the average primary ion beam current and the sputtering time.
For the rotation angle analysis, we chose EBSD grain number 183 (see Figures 3 and 7), which was a relatively large grain with space for more SIMS craters and which was of red colour.This means that the grain was close to a h001i crystallographic orientation.The closest h001i direction was only 10 from the sample surface normal.As part of our ongoing investigation, this sample had now been heat treated at 900 C for 24 h and Pt was deposited on the surface and allowed to diffuse into the sample.We could thereby obtain diffusion profiles during our subsequent SIMS analysis.We sputtered longer than before to make sure we reached the bulk level in the Pt signal, typically 2500 s, which gave crater depths down to 4 μm.The initial target sample rotations were every 10 from 0 to 120 .After seeing the large spread in the results (Figure 8), we also made craters at 26 and 29 rotation.The latter crater is not shown in Figure 7 since it was in a high region at the edge of the sample that here appears all white.
Figure 8 shows the diffusion profiles of Pt in grain 183, plotted as the Pt/Pd count rate ratio versus sputtering depth.Because of different sputtering rates, the lengths of the profiles vary, but all reach what is assumed to be the bulk level with a constant Pt/Pd ratio.Even within the same grain, the SIMS count rates have large differences, depending on the direction of the primary beam relative to the crystallographic axes of the Pd grain.To summarize the rather noisy depth profiles and investigate the effect of rotation angle, we simply averaged the full length of the profiles as shown in Figure 8 and plotted the results as a function of sample rotation angle.Section S4 shows the result of this analysis in Figure S17.We find rapid, and apparently non-systematic, changes in the Pt/Pd count rate ratios as a function of angle.Even over a 3 rotation, there is a four-fold difference in the Pt/Pd ratio.Thus, the effect of grain orientation is very sensitive to rotation angle.the lowest surface energy of 1.36 J/m 2 , followed by 1.45 J/m 2 for {322} planes, 1.46 J/m 2 for {332} planes, 1.50 J/m 2 for {221} planes, 1.52 J/m 2 for {001} planes, 1.53 J/m 2 for {331} planes and 1.57J/m 2 for {311} planes. 9It is therefore natural that sputtering from a favourable angle will cause {111} surface facets to dominate after a while.
The fact that grain 183 is fairly close to a [001] projection gives a geometry where the opposite {111} traces pair up at nearly 90 to allow such a doubly undulating surface, as shown in Figure 9 The abundance corrected and averaged Pt/Pd count rate ratio for the three craters within each of the 20 different grains of the main sample.The ratio is quite constant within each grain, but it varies by up to a factor of four between the grains.
temperature in the presence of PtO 2 . 10,11Large pits, such as those seen in the image for 78 , were observed in several other craters too but were not included in these high-magnification images.
Despite intense efforts, we have not found any systematic trends in the Pt/Pd ratio as a function of the primary ion beam vector relative to the crystallographic axes of the Pd grains.The abrupt changes in Figure S17 indicate that a mathematical formula that can accurately predict the Pt/Pd ratio would have to be complex and produce large differences over short angular intervals.It is beyond the scope of this investigation to develop such a formula, either based on atomic simulations or by rigorous and systematic SIMS studies on single crystals in an instrument where the sample could easily and precisely be rotated and tilted.In the latter case, a large number of single crystals with different surface normals would have to be analysed in order to get a complete formula.We would welcome such an effort but realize it would be quite demanding and time consuming.
The large differences in the Pt/Pd ratios, even from within the same grain, may be caused by a combination of the local surface topography and the so-called 'channelling effect'. 12Channelling means that ions can reach further into the crystal when moving along 'channels' between rows of atoms in close-packed planes or directions.
The local topography is, as seen in Figure 9, strongly dependent on the direction of the primary beam.When microfacets form, some parts of the surface may experience gracing incidence sputtering, whereas others may be closer to normal incidence.yield versus incoming angle are nearly flat between normal incidence (0 ) and 10 and then increase gradually by up to an order of magnitude with a peak at angles around 80-85 before falling off sharply to zero at 90 . 12Thus, in the microfaceted crater bottoms seen in Figure 9, the sputtering speed will be locally very different.If some form of preferential sputtering of heavy or light ions takes place on some of these facets, or this uneven sputtering somehow affects the ionization probabilities of Pt and Pd in different ways, this may also contribute to our observed scatter in the Pt/Pd ratios.It is therefore not surprising that many SIMS instruments have rotation stages that allow the sample to be rotated around the point of sputtering in order to reduce crater microfaceting and average the count rate over all incoming orientations.
Being a nearly non-destructive technique with high sensitivity, SIMS is well suited for the study of Platinum Group Metals (PGMs), particularly for catalyst research. 13However, there is a limited number of publications where SIMS has been used on PGM samples.Timeof-flight SIMS has been used to study minerals containing Pt and Pd. 14 SIMS was employed to determine depth profiles and diffusion coefficients of gold in iridium single crystals. 15SIMS depth profiling and quantification of Pt have further been performed relative to the Fe and Mn matrix elements of the mineral vernadite. 16Recently, SIMS depth profiling was used to quantify trace levels of Pt in deep sea minerals. 17gether with other advanced analytical methods, such as neutron activated analysis (NAA), transmission electron microscopy (TEM) and electron probe microanalysis (EPMA), SIMS has been proposed as a way to quantify sub-surface mercury contamination in platinum iridium mass standards, similar to the one that defined the kilogram in the SI system until 2019. 18Perhaps the challenges faced by us when trying to use SIMS to quantify Pt in Pd have also applied to other investigations of precious metals, where it was decided to rather use alternative techniques where the quantification was more straightforward.
The grain orientation effects of 6‰ in 18 O/ 16 O ratios found by Huberty and of 20% in 206 Pb/ 238 U ratios reported by Wingate and W. Compston are small compared to the four-fold difference (400%) between the smallest and largest Pt/Pd ratios found in this study.
However, since the recent paper by J. W. Angle et al. 7 also reports grain orientation effects with a four-fold magnitude, this seems to be a real and significant effect that should be considered when performing quantitative SIMS analyses.
J. W. Angle's papers 6,7 further show the importance of sample flatness and crater topography.It cannot be ruled out that small height differences may have affected our measurements, particularly in Stage C when the sample was repeatedly removed from the holder and repositioned after each rotation.However, during Stage B, the sample was fixed in the holder.Then, the overall sample surface height difference of 2.6 μm was larger than the sputtering depths of 0.8-2 μm.Could height differences between the grains explain the four-fold difference in Pt/Pd ratios?For crater 1 in grain 158, where the raw data are shown in Figure 4, there was a 4% increase in the primary ion beam intensity from start to finish of the analysis.However, the count rates for both Pd and Pt were reduced, falling by 17% and 7%, respectively.This opposite effect may have been caused by the changing height of the analysed surface relative to the microscope's secondary ion optics.The crater depth at the end of the analysis reached 1.5 μm.But even at this appreciable depth, the Pt/Pd count rate ratio from beginning to end of the analysis only changed from 84 to 92 ppm, thus a 10% increase.The crater depth was here roughly half the maximum height difference across the polished sample of 2.6 μm.With this simple analysis, if the observed change in Pt/Pd ratio from beginning to end of the 1.5-μmdeep crater was entirely caused by the changing surface height, we would not expect this effect to be larger than about 20% between any two places on the polished sample.This is small compared with the four-fold change we observe here from grain to grain.Further, as a check that the sample was not somehow tilted in the sample holder during the Stage B analysis, a plot of average Pt/Pd ratios as a function of position on the sample showed no clear correlations: High and low Pt/Pd values were scattered over the sample surface in a seemingly random fashion.Thus, height differences alone cannot explain the large differences in Pt/Pd ratios described here.
Regarding the crater microfaceting, the differences in crater topography shown in Figure 9 are independent of the secondary ion optics.So, unless height differences have also affected the primary ion beam and somehow given rise to the different crater microfaceting, it still seems likely that real grain orientation effects are at play in our samples.Although this effect is still not understood at a level where it can be predicted, we hope that the results presented here will be taken as a word of caution for anyone who intends to perform SIMS quantification for palladium-platinum alloys.Without making sure to correct for any height differences and use exactly the same crystal orientation when comparing samples, or averaging over a large enough number of grains, or using a rotation stage during the analysis, one can either get The SIMS analyses were performed with a Cameca IMS-7f magnetic sector instrument at the University of Oslo, Norway.The primary beam contained 10 keV O 2 + ions with an intensity of 30-100 nA.At this energy, the angle between the primary beam and the sample surface normal was 38.4 .The secondary ion energy was 5 keV.Each analysis lasted for 1000-2500 s.The crater raster size was 150 μm during the initial study of different reference samples and 100 μm in the subsequent analysis of the main sample in order to save space.During depth profiling, the ion count was done from a central spot with a diameter of about 1/3 the crater size to avoid ion counts from the tilted crater edges.The most common isotopes of both elements ( 104 Pd, 105 Pd, 106 Pd, 108 Pd, 110 Pd, 194 Pt, 195 Pt, 196 Pt and 198 Pt) were included in the sequential counting of positive ions during depth profiling at each position.For noble metals such as Pd and Pt, it is common in SIMS to use a Platinum was captured and diffused into the polished 1 wt% Pt disc sample in a 10 mm ID quartz reactor in a furnace with six temperature zones.A Pt filament was located upstream with the disc sample in a separate temperature zone.The furnace was heated while flushing the Pt filament and the Pd-Pt disc sample with a flow of nitrogen to a filament temperature of 1000 C, and the temperature in the Pt capture zone was 900 C. A gas mixture of 6% O 2 in N 2 was then fed over the Pt filament, transporting PtO 2 (g) to the 1 wt% Pt disc sample.There, the Pt was captured onto the surface of the disc and diffused into the sample.The experiment lasted for 24 h before the O 2 flow and furnace temperature were switched off.After cooling to room temperature, the sample was removed from the furnace and analysed by SIMS and WLI.

Figure 4
Figure4shows a typical example of SIMS raw data from a crater in one of the 20 selected grains.The SIMS count rate versus sputter time for all analysed isotopes of Pd and Pt are included.There was a slight decrease in the count rates over time after the initial stabilization during the first 30 s.To get a value for the signal intensity (count rate) of each isotope, the data were averaged over times between the vertical, dashed, green lines, thereby excluding non-typical values at the start and end of the sputtering.
for the Pt signal.The same trends are found for data from the reference samples with different Pt concentrations analysed in Stage A. However, for the measurements in Stage C where we analyse within the same grain but with different rotation of the sample, the trend is broken since the Pt signal then decreases with sputter rate.

Figure 9 shows
Figure9shows SEM images from the bottom of the SIMS craters sputtered with different sample rotations, thereby giving different directions of the incoming primary ion beam relative to a stationary sample.The white numbers and arrows show the directions of the insurface component of the incoming beam vector.The crater surface topography is strongly influenced by the direction of the primary beam, with ridges and terraces forming for the lowest angles and at 32 .For the two different À1 craters (see locations in Figure7), the surface after sputtering had a distinct wavy pattern with an orientation not very far from the (010) trace (line with magenta colour).At 11 , the surface relief is quite different with ridges forming V-shaped structures of varying sizes in several places in the image.The straight parts of the 'V's appear to follow the (111) and (111) traces (black and red lines) up and down in the first type of region and possibly follow the (111) and (111) traces (green and blue lines) up and down in the second type of region.The close-packed {111} planes in Pd have rotation in Figure 9, the crater bottom looks very different, with wide, dark bands of upward slopes and narrower lines of bright downward slopes when moving from left to right across the image.(As can be seen from the pit in the 78 image and the two small particles on the surface in the 113 and 122 images, the detector in the SEM was placed in a way that made slopes facing right and slightly upwards look brightest.)The orientation of these rows of facets matches well with the trace of (010) planes in the surface (magenta lines in selected SEM images).It is possible that the dark bands are dominated by (001) planes and the bright ones by (010) planes in this case.Similar geometries are found for 26 and 32 sample rotations.However, the intermediate angle of 29 shows a distinctly different crater bottom topography with shallow cusps and weak ridges roughly along the {111} traces.The 29 crater was made very near the edge of the cylindrical sample where a mechanical impact had compressed the surface and made it bend slightly upwards.It is not known if this may have caused the anomaly in the surface topography compared with the neighbouring images, or if the crater topography would normally and repeatably change so abruptly with angle-as we clearly see it does from 32 to 40 .Grain 183 was so full of craters that this could not be checked, given the remaining space and the accuracy by which it was possible to place the craters.Between sample rotation angles of 40 and 122 , the surface is generally quite smooth, with some shallow ridges along the (111) and (111) traces.These ridges were, however, present also outside the SIMS craters and must have formed on the polished surface during the heat treatment or Pt deposition.Reconstructions towards lowenergy planes are commonly seen on Pd samples kept at high

F I G U R E 9
SEM images from the bottom of 16 SIMS craters in grain 183, sputtered with different sample rotations.The images are all made with the sample in the same orientation (arrow up as in Figures 3, 4, 5 and 7).Both of the craters made at À1 sample rotation are included at the top left in order to show the consistency for similar orientations.The white arrows show the direction of the incoming primary beam in each crater, where the beam came in at a 38 angle from the surface normal.Numbers in white indicate the measured angle of sample rotation during each SIMS analysis.The 1 μm scale bar shown in only the first and last image gives the magnification of all 16 images.Red, black, green and blue lines in the lower left corner of each image show the direction of where the four, close-packed {111} planes of the Pd crystal would cross the flat surface of grain 183.These four crystallographic planes are shown in the octahedron at the lower right, which is drawn with the correct orientation relative to the images and the Pd unit cell shown at the top right.The cube in the middle similarly shows the orientation of the three {001} planes of the face-centred cubic unit cell.The crossing of the (010) plane with the image plane is shown in the first SEM images as a magenta line.See the text for a discussion of which surface facets are exposed in the various cases.

4 |
CONCLUSIONSWhen using isotope abundance corrected and averaged count rate ratios from SIMS for Pt in Pd, we have found differences in the Pt/Pd ratios of up to four times in different alloys of Pd-Pt, among different grains in the same sample (1 wt% Pt) and within the same grain with different rotation of the sample relative to the primary ion beam.This large grain orientation effect adds another layer of difficulty when attempting to get quantitative concentrations or depth profiles of Pt in Pd.Ion channelling and focusing inside the crystal, along with different microfaceting of the SIMS crater bottom that affects the sputter rate, are factors that seem to influence the grain orientation effect.Changes in relative angle of only a few degrees can change the Pt/Pd ratio by a factor of four.
very confused or reach very wrong conclusions because of the large grain orientation effects in Pd-Pt samples.operating the SIMS instrument.Financial support from the Research Council of Norway through the 'industrial Catalysis Science and Innovation Centre' (iCSI, contract number 237922) is gratefully acknowledged.