EXAFS Peaks and TPR Characterizing Bimetallic Interactions : Effects of Impregnation Methods on the Structure of Pt-Ru / C Catalysts

To investigate bimetallic interactions, Pt-Ru/C catalysts were prepared by coimpregnation (Pt-Ruco/C) and successive impregnation (Ru-Ptse/C), while Pt/C, Ru/C, and reduced Pt-Rublack were used as reference. Those samples were characterized by XAS and TPR. When Pt(absorber)-Ru(backscatter) phase-and-amplitude correction is applied to Fourier transformed (FT) EXAFS of Pt-Rublack at Pt edge, the characteristic peak of Pt-Ru interactions appears at 2.70 ́ Å, whereas, when Pt-Pt correction is applied, the peak appears at about 2.5 ́ Å. Detailed EXAFS analysis for Pt-Ruco/C and Pt-Ruse/C confirms the nature of the characteristic peak and further indicates that the interactions can semiquantitatively be determined by the relative intensity between Pt-Ru and Pt-Pt characteristic peaks. This simple method in determining bimetallic interaction can be extended to characterize Pt-Pd/γ-Al2O3. However, for PtRe/γ-Al2O3, Pt-Re interactions cannot be determined by the method because of the overlap of Pt-Pt and Pt-Re characteristic peaks due to similar phase functions.

Several explanations have been offered for the role of M 2nd in improving the catalytic performance: (1) it may change the morphology of Pt clusters and help anchor Pt and maintain it in a state of high dispersion [14,[20][21][22][23][24][25][26]; (2) it may convert the poisonous species into harmless compound in the reaction [2,27]; (3) in bonding with Pt, it may change the electronic properties and geometric features of Pt thereby decreasing the adsorption of poisonous species on Pt [18,24,[28][29][30][31][32][33]; and (4) in association with Pt, it may modify the catalytic properties of Pt [34][35][36][37][38].To discriminate the role of M 2nd in improving the catalytic performances, it is crucial to determine the structure of the catalysts, specifically, the extent of bimetallic interaction.
TPR (temperature programmed reduction) has been widely used to characterize the bimetallic interactions.For Ru-Pt and Re-Pt catalysts, it has been suggested that the TPR characterized by a single peak would indicate a catalytic reduction of M 2nd [39][40][41][42][43]; presumably, M 2nd precursors, which are highly mobile or located nearby Pt, can quickly migrate to the hydrogen-covered Pt where they are catalytically reduced by Pt.This postulate suggests the formation of bimetallic interactions.In contrast, as the M 2nd precursors are immobile or located far away from Pt, isolated peaks are observed in TPR spectrum, one for Pt and one for M 2nd , which are reduced separately; hence, no M 2nd -Pt interactions were suggested.However, the reduced M 2nd with increased mobility could migrate toward Pt, leading to M 2nd -Pt interactions.In addition, decomposition of surface functional groups in hydrogen reduction may change diffusion rate of spillover hydrogen, the reducibility of M 2nd , and the metal-support interactions.These factors will be reflected in TPR spectra but normally are not taken into account in interpretation of TPR results.Since catalyst structure plays a crucial role in determining the catalytic properties, TPR is regarded to be insufficient to characterize the catalyst structure and to rationalize the roles of M 2nd in improving the catalytic performance.
The main parameter controlling the catalyst morphology was the method of preparation.Bimetallic catalysts can be prepared by a number of synthetic methods, each with its own distinct advantages [26,39,[44][45][46][47][48][49][50][51][52].Among them, impregnation method is the commonest one and is used widely in the preparation of commercial catalysts.By coimpregnation methods, Pt and M 2nd ion precursors collide with each other and intimately get mixed together.Thus the metals could aggregate together in the N 2 drying step.Hence, bimetallic interactions are expected upon H 2 reduction.In contrast, in the successive impregnation method, Pt ion precursors and M 2nd ions would aggregate separately; hence, Pt clusters are expected to be segregated from M 2nd clusters with little interactions.
The goal of the research summarized here was to investigate the structure of Pt-Ru/C catalysts prepared by coimpregnation and successive impregnation methods by using in situ extended X-ray absorption fine structure (EXAFS) spectroscopy, which has been proven to be a powerful technique for the characterization of bimetallic catalysts [53,54].Since EXAFS contribution of metal-support interactions is crucial in determining the morphology of Ru-Pt clusters that is strongly coupled with metal oxide, M(Pt or M 2nd )-O contribution in situ catalyst pretreatment is necessary for preventing the oxidation of metal clusters.
The specific goals of this study were to investigate the characteristic peaks of bimetallic catalysts and determine the bimetallic interactions semiquantitatively by phase-andamplitude function corrected Fourier transform method.In this investigation, Pt/C, Ru/C, and reduced Pt-Ru black were prepared to study the structure for comparison.In developing a simple method to characterize the bimetallic interactions, the correlation between the ratio of Pt-Ru to Pt-Pt characteristic peak intensity ( Pt-Ru / Pt-Pt ) and the ratio of Pt-Ru to Pt-Pt coordination number ( Pt-Ru / Pt-Pt ) estimated from detailed EXAFS analysis was employed.This method can hopefully be extended to other important commercial bimetallic catalysts, specifically, for Pt-Pd catalysts.

Sample Preparation.
Ru-Pt catalyst samples were prepared by using coimpregnation and successive impregnation, respectively.For the coimpregnation sample (Pt-Ru co /C), 5 g of carbon black powder (Vulcan, XC-72R) was mixed with 33.5 mL of solution containing 0.660 g of H The details of the preparation procedure were shown in our previous paper [55].Pt-Re/-Al 2 O 3(d) catalyst was prepared by successive incipit wetness impregnation technique.The -Al 2 O 3 (Degussa, aluminum oxide C) was first brought in contact with a solution of Pt(NH 3 ) 4 (NO 3 ) 2 (Strem) in deionized water, followed by drying at 120 ∘ C overnight.The sample was then brought in contact with a solution of NH 4 (ReO 4 ) (Alfa), dried at 120 ∘ C for 12 h, and calcined at 500 ∘ C for 4 h [19].The catalyst sample contained 0.47 wt% Pt and 0.69 wt% Re (determined by inductively plasma optical emission spectroscopy using a Janell-Ash 1100 instrument) and was stored in a glove box before EXAFS measurement.

2.2.
Temperature-Programmed Reduction.The apparatus used for the temperature-programmed reduction (TPR) has been described by McNicol [56].The experiments were performed with H 2 : Ar mixture (10 : 90 molar) flowing at about 30 mL (NTP)/min at 1 atm.After 012 g of the catalyst sample was charged into a quartz TPR cell, gas stream was passed through the catalyst sample.When the thermal conductivity detector (TCD) signal became stable, the sample was heated at 10 ∘ C/min to 700 ∘ C by a temperatureprogrammed furnace.The reduction temperature was monitored with a K-type thermocouple and the water produced by reduction was trapped into a column of silica gel.The output polarity was set so that positive TCD signals indicate hydrogen consumption.

X-Ray Absorption Spectroscopy.
The EXAFS experiments with Pt-Ru/C and Pt-Pd/-Al 2 O 3 at Pt L III -edge were performed on X-ray beam line BL17C1 of National Synchrotron Radiation Research Center (NSRRC) in Hsinchu, Taiwan, while those samples at Ru and Pd K edge were measured on the superconducting wavelength shifter beam line 01C, with energy ranging from 5 to 33 keV.The Pt-Re/-Al 2 O 3(d) sample was measured on X-ray beam line X-11A at National Synchrotron Light Source at Brookhaven National Laboratory, Upton, Long Island, New York.Before EXAFS measurement, the samples were predried in flowing He for 1 h and then reduced in situ with flowing H 2 at 450 ∘ C for 4 h.
The electron storage ring of NSRRC was operated at an energy level of 1.5 GeV and a beam current of 360 mA.A Si(111) double-crystal monochromator was used for energy selection, and two Rh-coated mirrors (mainly for the purpose of collimating and focusing the synchrotron beam) are used to eliminate high energy photons.The transmission measurement geometry was arranged using gas-filled ionization chambers to monitor the intensities of the incident and transmitted X-rays and standard Pt, Pd foil, and Ru black were simultaneously measured as a reference so that energy calibration could be done between scans.
The Pt-Ru catalyst samples were pressed into a selfsupporting wafer and placed into a controlled-atmosphere EXAFS cell.In flowing 50 mL/min purified, dried grade 5 H 2 at atmospheric pressure, they were heated at a rate of 5 ∘ C/min to 300 ∘ C and held at that temperature for 1 h.After the cell was cooled to room temperature, the EXAFS measurements were made in energy from 200 eV below the platinum L III absorption edge (11564 eV) and ruthenium K edge (22117 eV), respectively, to 1200 eV above the edge.EXAFS measurements were also made with reference material, Pt-Ru black (Alfa Aesar).The Pt-Ru black sample was loaded into the EXAFS cell, reduced with H 2 300 ∘ C for 1 h; the sample was noted as Pt-Ru black .After the sample was cooled to room temperature, the measurement was performed in H 2 environment.The EXAFS measurement procedure for Pt-Pd/-Al 2 O 3 , Pt/-Al 2 O 3 , and Pt-Pd/-Al 2 O 3 is the same as that for Pt-Ru catalyst samples except that these samples (PtO 2 /-Al 2 O 3 , Pd(OAc) 2 /-Al 2 O 3 , and PtO 2 -Pd(OAc) 2 /-Al 2 O 3 ) were reduced ex situ at 450 ∘ C under the 580 psig of pure hydrogen for 6 h.
Data reduction and data analysis were performed with the XDAP code developed by Vaarkamp et al. [57].Standard procedures were followed to extract the EXAFS data from the measured absorption spectra.The preedge was approximated by a modified Victoreen curve [58] and the postedge background was subtracted using cubic spline routines [59,60].Normalization was performed by dividing the backgroundsubtracted spectrum by the height at 50 eV above the edge [57,61].Phase shifts and backscattering amplitudes functions of Pt-Ru, Ru-Pt, Pt-Pd, Pd-Pt, Pt-O, Pt-Cl, Ru-O, Ru-Cl, and Pd-O were generated by FEFF8 code [62] and used as reference files for EXAFS data analysis.Two separated TPR characteristic peaks being observed for Pt-Ru se /C would suggest that Pt and Ru clusters are segregated, with little or no bimetallic interaction [39,63].Since TPR is unable to determine the existence of bimetallic interaction conclusively, EXAFS was used as an alternative to examine the extent of the bimetallic interaction.

EXAFS Characteristic Peaks of Ru-Pt Bimetallic Interaction. A general expression for EXAFS function is formulated as follows:
The EXAFS function is a superposition of contributions from  different sine function coordination shell modulated by an amplitude function   ().In the argument of sine function,   is the average distance between the absorbing atom and the neighboring atoms in  coordination shell and   () is a phase shift function and is approximated by linear dependence on ; that is,   () =  0  +   * .In order to separate the contributions from different shells, Fourier transform is performed to the EXAFS function and is formulated as The resulting Fourier transformed function will give maxima at the radii that are related to the coordination distance   in (1).However, due to the presence of phase function, the Fourier transformed EXAFS function peaks at a distance of  + /2, instead of .Moreover, due to the influence of the nonlinear amplitude function, specifically, for an absorber-backscatter pair containing high-Z backscatter atom, more than one peak will be present for a single shell.These two factors complicate the study of bimetallic interactions by normal Fourier transform (2).In order to eliminate the complication, the Fourier transform is corrected by replacing () with () −() /() and the corrected Fourier transform is formulated as By this correction, only one peak will appear for an absorber-backscatter pair and the faulty coordination distance can also be corrected.Hence, for an absorberbackscattering pair X-Y, if right phase and amplitude function is applied, the imaginary part of the Fourier transformed EXAFS function will peak positively and center at the maximum of the magnitude; the correct coordination distance can thus be determined by the peak location [60,[64][65][66].With the help of phase and amplitude phase functions calculated from FEFF and the phase-and-amplitude corrected Fourier transform, different type of neighbors can be identified by different phase and/or amplitude correction.However, for bimetallic catalysts, the coexistence of two absorberbackscattering pairs complicates the phase-and-amplitude correction due to the difference in phase and amplitude function for different backscattering metals.For instance, when Pt-Pt phase-and-amplitude correction that was applied to EXAFS function involved both Pt-Ru and Pt-Pt pairs, the coordination distance of Pt-Pt will be correct, whereas the distance for Pt-Ru deviates from real distance of about ( Pt-Pt −  Pt-Ru )/2.
In order to identify the neighboring atom of Pt, a  3 -weighted Pt-Pt phase-and-amplitude corrected Fourier transform in the range of 3.0 <  < 18.0 Å−1 was performed to the EXAFS functions characterizing Pt-Ru black and Pt foil at Pt edge.As shown in Figure 2, a positive peak at about 2.75 Å appears in the imaginary part of the normalized Fourier transformed EXAFS for Pt-Ru black and Pt foil (Figure 2(a)), while a negative peak at about 2.5 Å was only observed for Pt-Ru black .For Pt-Ru black , the positive peak coincides with that for Pt foil and was assigned as the Pt-Pt characteristic peak.Thus the negative peak was assigned as Pt-Ru characteristic peak.Due to an approximate  radian difference between Pt-Pt and Pt-Ru phase function ( Pt-Pt 0 −  Pt-Ru 0 ), Pt-Ru contribution corrected by Pt-Pt phase function results in a negative peak.Another small peak appearing at about 3.2 Å was also contributed from Pt-Ru (Figure 2(a)); as mentioned in the preceding paragraphs, isolated Pt-Ru shell obtained from detailed EXAFS analysis (Figure 3(a)) confirmed the assignment.In order to confirm the assignments and to estimate the coordination distance of Pt-Ru, Pt-Ru phase-and-amplitude correction was applied to EXAFS function of Pt-Ru black .As expected, a positive and a negative peak appears at about 2.7 Å and 3.0 Å, respectively (Figure 2(b)).The positive peak is contributed from Pt-Ru.Again, due to the difference in the phase shift function between Pt-Pt and Pt-Ru, the characteristic peak of Pt-Pt shown in Figure 2(b) is negative and longer than that shown in Figure 2(a).
Similar to the Pt edge analysis, application of Ru-Ru phase and amplitude corrected to Fourier transformed EXAFS functions at Ru edge for Pt-Ru black and Ru black results in a positive peak for both samples and an additional negative peak for Pt-Ru black (Figure 2(c)).The negative peak was thus assigned as the characteristic peak of Ru-Pt.
For detailed analysis of EXAFS characterizing Pt-Ru black at Pt edge, 1st shell Pt-Pt and Pt-Ru contributions were isolated by Fourier filter (0.607 <  < 3.307 Å and 3.0 <  2 < 16.0 Å−1 ).At the beginning, the coarse structural parameters for Pt-Pt contribution were determined by adjusting the parameters to give the best agreement with the Pt-Pt characteristic peak of the Fourier-isolated EXAFS function (raw data).An EXAFS function calculated from the parameters for Pt-Pt contribution was then subtracted from the raw data.As expected, a symmetric peak appeared at about 2.7 Å after applying Pt-Ru phase-and-amplitude Fourier transform to the difference EXAFS data.The structural parameters for Pt-Ru contribution were then estimated by a single-shell fitting routine.More accurate parameters for Pt-Pt contribution were then estimated by fitting the difference data obtained from the subtraction of the calculated Pt-Ru contribution from the raw data.Further refinement of the analysis was done by the use of multishell fitting routine with the initial guesses obtained from single-shell fitting routine [67,68].
Detailed analysis of EXAFS characterizing Pt-Ru black at Ru edge is similar to that described for the Pt edge.The EXAFS data were Fourier-transformed over the useful range (3.0 <  2 < 16.0 Å−1 ) and the major contributions, Ru-Ru and Ru-Pt, were isolated by inverse Fourier transform in the range of 0.6 <  < 3.012 Å.The coarse structural parameters for Ru-Ru contribution were determined by fitting Ru-Ru characteristic peak, manually.The values of these parameters were fixed and then the difference file technique and singleshell fitting routine were applied to estimate the parameters for Ru-Pt.The contribution of Ru-Pt was subtracted from raw data and the improved parameters for Ru-Ru were then estimated by fitting the difference data.The recursive iterations were carried out until a satisfactory overall fit was reached.
As a criterion of the satisfactory overall fit, the distance and Debye-Waller factor characteristics of Ru-Pt interactions determined at Ru edge should be the same as those of the Pt-Ru interactions determined at Pt edge.The comparison of the parameters obtained at two edges is useful for determining the goodness of overall fit.Since the effects of  and Δ 2 and  and ΔE 0 on the calculated EXAFS are highly correlated and the contributions of Ru-Pt and Ru-Ru are strongly coupled, without any constraints in overall fit, the difference of the calculated Δ 2 between Pt-Ru and Ru-Pt contributions for Pt-Ru/C could be higher than 30%.Hence, Δ 2 value for Ru-Pt was tuned in the fitting process, until Δ 2 value for Pt-Ru is the same as that for the Ru-Pt (within standard deviation) reached in parameter estimation.The results of the detailed EXAFS analysis are shown in Table 1 and the illustration of Pt-Ru and Ru-Pt characteristic peaks was shown in Figure 3.

Effects of Preparation Methods on Pt-Ru Bimetallic Interactions.
A comparison of Pt-Pt phase-and-amplitude corrected Fourier transform of the EXAFS data at Pt edge representing Pt-Ru black , Pt-Ru co /C, Pt-Ru se /C, and Pt/C is shown in Figure 4(a).For all these samples except Pt/C, the appearance of the peak at 2.5 Å is the characteristic peak of Pt-Ru, suggesting the presence of Pt-Ru bimetallic interactions for all Pt-Ru samples.EXAFS evidence of the formation of bimetallic interaction for all Pt-Ru se /C indicates that a bimetallic catalyst with separated peaks in TPR spectrum still presents bimetallic interactions.
The higher intensity of the peak for Pt-Ru co /C with respect to that for Pt-Ru se /C demonstrates that Pt-Ru bimetallic interactions can be manipulated by impregnation method and the sample prepared by coimpregnation method leads to stronger bimetallic interactions.The statement is consistent with EXAFS data at Ru edge.As shown in Figure 4(b), a comparison of Ru-Ru phase-and-amplitude corrected Fourier transformed EXAFS data for Pt-Ru black , Pt-Ru co /C, and Pt-Ru se /C shows the presence of Ru-Pt characteristic peak and the peak intensity is in the order Pt- The EXAFS data for Pt-Ru co /C and Pt-Ru se /C were analyzed by a method slightly different from that described for Pt-Ru black to take care of the metal-carbon interactions.For EXAFS data at Pt edge, after isolating the major contributions of EXAFS data, structural parameters for Pt-Pt and Pt-Ru contributions were estimated by fitting the raw data in the high-k range (7.5 <  < 16 Å−1 ) by adopting the structural parameters for Pt-Ru black as the initial guesses; contributions from metal-carbon interactions (Pt-C) in this region are small.An EXAFS function calculated with the estimated parameters was then subtracted from the raw data (3.0 <  < 16 Å−1 ).The residual spectrum was expected to be contributed from Pt-C or Pt-Cl.For Pt-Ru se /C, as expected, after Pt-C phase corrected Fourier transform was performed  on the difference EXAFS function, a significant peak was detected at about 2.1 Å (Figure 5(a)).The first guess structural parameters for Pt-C were estimated by single-shell fitting routine and more accurate parameters for Pt-Pt, Pt-Ru, and Pt-C contributions were determined by fitting the raw data with 12 adjustable parameters; the initial guesses for Pt-Pt and Pt-Ru were estimated from raw data in the high-k range (7.5 <  < 16 Å−1 ).In contrast to Pt-Ru se /C, after Pt-C phase corrected Fourier transform was performed on the residual EXAFS, no significant peaks were observed in the range 1.0 <  < 3.0 Å for Pt-Ru co /C (Figure 5(a)).Pt-Ru co /C had Pt-Pt coordination number of 5.0 suggesting the formation of small Pt cluster size.Unless bilayer Pt-Ru clusters with Pt layer on the top [69] were formed on the carbon, Pt-C contribution should be observed in the residual spectrum for Pt-Ru co /C.In the model, oxophilic Ru in a low positive oxidation state interacts with carbon support, stabilizing the dispersion of the Pt clusters.EXAFS data at Ru edge were analyzed by a method similar to that described for the Pt edge.After the subtraction of Ru-Ru and Ru-Pt contributions from raw data of Pt-Ru co /C and Pt-Ru se /C, respectively, Ru-Cl phase corrected Fourier transform was applied to the resulting difference files.For Pt-Ru co /C, the appearance of a symmetrical peak at about 2.35 Å in the imaginary part suggests that the chloride ligands have not been removed completely, while the appearance of another peak at about 3.3 Å could be contributed from Ru-O * (where O * denotes the oxygen of oxygen containing functional groups on the carbon surface) (Figure 5(b)).The results suggest that Pt-Ru clusters could bond with the functional groups via chloride ligands.For Pt-Ru co /C, the appearance of an asymmetrical peak in the imaginary part suggests the formation of Ru-C contribution besides Ru-Cl, whereas the peak is too small to be analyzed by multishell fitting routine.
The structural parameters for the residual chloride ligands (Ru-Cl) and Ru-support interactions (Ru-C or Ru-O * ) [70,71] were estimated by multishell fitting routine.
Fourier transform Fourier transform More accurate parameters for Ru-Ru and Ru-Pt were then estimated by the subtraction of the calculated Ru-Cl and Ru-C (Ru-O * ) from raw data.The recursive iteration was carried out until a satisfactory overall fit was reached.The results of the EXAFS analysis at Pt and Ru edge were summarized in Tables 1 and 2, respectively, and the comparison of the data and the fit are shown in Figure 6.
As shown in Figure 7, the ratios of the Pt-Ru and Pt-Pt characteristic peak intensity ( Pt-Ru / Pt-Pt ) are correlated well with that of coordination number ( Pt-Ru / Pt-Pt ).Since  Pt-Ru represents the number of Ru atoms neighboring to a Pt atom, the higher the  Pt-Ru , the greater the Pt-Ru bimetallic interaction.By use of this correlation, the extent of bimetallic interactions can thus easily be estimated by the use of phase-and-amplitude corrected Fourier transform instead of tedious and time-consuming detailed EXAFS analysis.The extent of bimetallic interaction is an important consideration in bimetallic catalyst development.In developing commercial catalysts, many samples will be prepared.To save time and expensive catalytic performance tests, the method proposed here can be used for quick screening of promising catalyst samples for the performance tests.

TPR Characterizing Pt-Pd and Pt-Re Catalysts and EXAFS
Characteristic Peaks of Pt-Pd Bimetallic Interactions.In contrast to Pt/-Al 2 O 3 at 270 and Re/-Al 2 O 3 at 520 ∘ C, only one peak was observed in TPR characterizing Pt-Re/-Al 2 O 3 catalyst at 310 ∘ C, as shown in Figure 8(a).The results are consistent with TPR results reported in the literature for Pt-Re catalysts predried by He at low temperature (<300 ∘ C) before H 2 reduction [39].Inferred from TPR and EXAFS characterizing Pt-Ru co /C, the apparent reduction of a single peak indicates the formation of bimetallic interactions; besides the catalytic reduction of Re species by Pt as postulated in the literature [39,51,72,73], Pt-Re bimetallic interactions can also be formed from reaction of the mobile hydrated ReOx species and reduced Pt clusters.
The TPR profiles characterizing PtO 2 /-Al 2 O 3 , Pd(OAc) 2 /-Al 2 O 3 , and PtO 2 -Pd(OAc) 2 /-Al 2 O 3 are shown in Figure 8(b).It is apparent that the main reductive decomposition of acetate ligands in Pd(OAc) 2 on -Al 2 O 3 is at 470 ∘ C. In the presence of Pt, the reductive decomposition temperature decreased by about 50 ∘ C. The results combined with the long term stability tests [55] suggest the existence of a significant Pd-Pt bimetallic interaction.The exploration of  EXAFS characteristic peaks for Pt-Re and Pt-Pd interaction, respectively, is illustrated as follows.
A comparison of the Pt-Pt phase-and-amplitude corrected Fourier transform of the EXAFS spectra at Pt edge for Pt foil, Pt/-Al 2 O 3(d) , and Pt-Re/-Al 2 O 3(d) shows that major peak is located at a distance of 2.7-2.8Å, consistent with the metal-metal distance of bulk Pt and bulk Re (Figure 9 For Pt-Re/-Al 2 O 3, an additional peak appearing at about 3.4 Å cannot be assigned as the characteristic peak for Re-Pt because the phase functions of Re-Pt and Re-Re are very similar to each other; moreover, when experimentally determined Re-Re phase and amplitude function was applied, no significant peaks appear in  ranging from 3.0 to 3.5 Å. (Figure 9(c)).The discrepancy for the Fourier transforms corrected by theoretical and experimental phase and amplitude function could be caused by the side lobe generated from the finite range of the Fourier transform or the difference of amplitude functions obtained from theoretical and experimental methods.
Since the affinity of Pt-Re and Pt-Pd clusters to -Al 2 O 3 is higher than Pt-Ru to carbon, the contributions from low atomic weight scatterers, such as the oxygen atoms of the support, are more pronounced for Pt-Re/-Al 2 O 3(d) and Pt-Pd/-Al 2 O 3 than that for Pt-Ru/C.In exploring the characteristic peak of Pt-Re and Pt-Pd, to minimize the interference from low atomic weight scatters,  3 -weighted Pt-Pt phase-and-amplitude corrected Fourier transforms of the EXAFS functions were determined at the Pt edge for the -Al 2 O 3 supported catalyst samples.Similar to the Fourier transformed EXAFS for Pt-Ru black , for Pt-Pd/-Al 2 O 3 , a positive peak at about 2.7 Å combined with a shoulder (peak) at 2.45 Å was observed in imaginary part of the Fourier transformed EXAFS (Figure 10(a)).The peak at 2.45 Å was  in a shift of  to longer distance.However, we are unable to assign a characteristic peak for Pd-Pt because of the superposition of the peak with Pd-Pd characteristic peak; a small peak appearing at 3.1 Å could be contributed by Pd-Pt absorber-backscatter pair, whereas its intensity is only slightly higher than that of the side lobe of the main peak and is hard for the estimation of the extent of Pt-Pd bimetallic interactions.

Conclusions
Pt-Ru bimetallic interaction can easily be detected by Pt-Pt phase-and-amplitude corrected Fourier transformed EXAFS function and the Pt-Ru contributions were characterized by the peak appearing at 2.5 Á.Moreover, after detailed EXAFS analysis of Pt-Ru co /C, Pt-Ru se /C, and Pt-Ru black by combined difference file technique and multishell fitting routine, we found that  Pt-Ru / Pt-Pt are correlated well with  Pt-Ru / Pt-Pt .Since  Pt-Ru / Pt-Pt is the indication of the extent of bimetallic interaction, Pt-Ru bimetallic interaction can semiquantitatively be determined by simple Fourier transform technique using  Pt-Ru / Pt-Pt .
The method proposed in this study can be extended to characterize Pt-Pd/-Al 2 O 3 bimetallic interaction because the phase function of Pt-Pd is significantly different from that of Pt-Pt.In contrast, the method cannot be applied to Pt-Re catalysts because the phase functions of Pt-Pt and Pt-Re are similar and the Fourier transform ranges are limited by the appearance of the Re L III absorption edge.
Traditionally, TPR has been used to characterize bimetallic interactions whereas it is not a conclusive technique to quantify the interactions.By comparing the EXAFS and TPR results, we found that Pt-M 2nd catalysts characterized by TPR with a single peak do show bimetallic interaction, in agreement with the interpretation of most of the researchers.However, for bimetallic catalysts with separated peaks in TPR profile, the existence of bimetallic interactions remains the subject of discussion.However, in this study, EXAFS conclusively indicates that these kinds of catalysts can have bimetallic interactions.Furthermore, the correlation of  Pt-M / Pt-Pt with TPR results suggests that the extent of bimetallic interactions is in the following order: one TPR peak > two separated peaks with 2nd metal reduction peak shifting to lower temperature > two separated peaks without a significant peak shift.

Figure 3 :
Figure 3: Illustration of (a) Pt-Ru contribution for Pt-Ru black : imaginary part and magnitude of Fourier transform ( 2 -weighted, 3.0 <  < 16 Å−1 , Pt-Pt phase and amplitude corrected) for raw data at Pt L III edge minus calculated Pt-Pt (red line) and raw data (black line) and (b) Ru-Pt contribution for Pt-Ru black : imaginary part and magnitude of Fourier transform ( 2 -weighted, 3.0 <  < 16 Å−1 , Ru-Ru phase and amplitude corrected) for raw data at Ru K edge minus calculated Ru-Ru (red line) and raw data (black line).

Figure 5 :
Figure 5: Illustration of the EXAFS contributions characterizing the metal-carbon interactions.(a) Imaginary part and magnitude of Fourier transform ( 3 -weighted, 3.5 <  < 13 Å−1 , Pt-C phase corrected) of raw EXAFS data at Pt L III edge minus calculated metal-metal contributions (Pt-Pt + Pt-Ru) for Pt-Ru co /C (black line) and Pt-Ru se /C (red line); (b) imaginary part and magnitude of Fourier transform ( 1 -weighted, 3.0 <  < 14 Å−1 , Ru-Cl phase corrected) of raw EXAFS data at Ru K edge minus calculated metal-metal contributions (Ru-Ru + Ru-Pt) for Pt-Ru co /C (black line) and Pt-Ru se /C (red line).

Figure 6 :
Figure 6: Comparison of imaginary part and magnitude of Fourier transform ( 2 -weighted, 3.0 <  < 16 Å−1 , Pt-Pt phase and amplitude corrected) of raw EXAFS data at Pt L III edge (solid line) and sum of the calculated contributions (dashed line) for (a) Pt-Ru co /C and (b) Pt-Ru se /C.Comparison of imaginary part and magnitude of Fourier transform raw EXAFS data at Ru K edge ( 2 -weighted, 3.0 <  < 14 Å−1 , Ru-Ru phase and amplitude corrected) of experimental EXAFS data (solid line) at Ru K edge and sum of the calculated contributions (dashed line) for (c) Pt-Ru co /C and (d) Pt-Ru se /C.

Figure 7 :
Figure 7: Correlation between the ratio of Pt-Ru to Pt-Pt characteristic peak intensity ( Pt-Ru / Pt-Pt ) and the ratio of Pt-Ru to Pt-Pt coordination number ( Pt-Ru / Pt-Pt ).
(a)).The metal-metal distance in Pt foil and that in Pt/-Al 2 O 3(d) are about the same, whereas the distance characterizing Pt-Re/-Al 2 O 3(d) is significantly shorter (Figure 9(a)).A shortening of the metal-metal distance for Pt-Re/-Al 2 O 3(d) may suggest the formation of Pt-Re bimetallic interactions [19, 69].However, because the phase function for Pt-Pt is only slightly different from that for Pt-Re, no significant characteristic peak was observed in the Fourier transformed EXAFS.Similarly, a comparison of the Re-Re phase-andamplitude corrected Fourier transformed EXAFS data at Re edge shows that metal-metal distance in Pt-Re/-Al 2 O 3(d) is shorter than that in Re/-Al 2 O 3(d) , consistent with the formation of Pt-Re bimetallic interactions.

2
PtCl 6 ⋅6H 2 O and 0.625 g of RuCl 3 ⋅XH 2 O (Strem) and evacuated at 10 −2 -10 −3 torr and about 80 ∘ C for overnight.The catalyst sample contained 4.7 wt% Pt and 4.4 wt% Ru, as measured by inductively plasma optical emission measurement (ICP-OES, Perkin Elmer PE Optima 4300).For the successive impregnation sample (Pt-Ru se /C), PtO 2 /C was first prepared by impregnating carbon black powder with H 2 PtCl 6 , reducing in H 2 at 200 ∘ C for 2 h and then calcining in air at 200 ∘ C for 2 h.The resulting PtO 2 /C of 5.0 g was brought into contact with 0.625 g of RuCl 3 ⋅XH 2 O in 22.0 mL of doubly distilled deionized water, followed by evacuation at 10 −2 -10 −3 torr and about 80 ∘ C. The Pt-Ru se /C sample contained 4.4 wt% Pt and 4.1 wt% Ru.For TPR study, Pt/C and Ru/C catalysts were prepared by impregnation of carbon black with H 2 PtCl 6 and RuCl 3 , respectively, followed by drying at 80 ∘ C. Pt-Pd/-Al 2 O 3 catalyst was prepared by successive impregnation method using H 2 PtCl 6 and Pd(O 2 CCH 3 ) 2 as Pt and Pd precursors, respectively, and was denoted as Pt-Pd(OAc) 2 /-Al 2 O 3 .The -Al 2 O 3 support was prepared from Versal 250 and capital B pseudoboehmitic alumina powder.For comparison, Pt/-Al 2 O 3 was prepared by incorporation of H 2 PtCl 6 into -Al 2 O 3 and followed by air calcination at 450 ∘ C and Pd(OAc) 2 /-Al 2 O 3 was prepared by impregnating Pd(O 2 CCH 3 ) 2 on -Al 2 O 3 without calcination.

Table 1 :
Summary of EXAFS analysis results for Pt-Ru catalysts at Pt LIII edge.Δ 2 : the difference in Debye-Waller factors between sample and standard.Value in the parentheses is standard deviation (in the last given digit) calculated from the covariance matrix and the estimates of the noise level in the raw data.
a : the coordination number for the absorber-backscattering pair.b: the average absorber-backscattering distance.cd Δ 0 : the inner potential correction.* e

Table 2 :
Summary of EXAFS analysis results for Pt-Ru catalysts at Ru edge.Δ 2 : the difference in Debye-Waller factors between sample and standard.Value in the parentheses is standard deviation (in the last given digit) calculated from the covariance matrix and the estimates of the noise level in the raw data.Tuned parameter.
a : the coordination number for the absorber-backscattering pair.b: the average absorber-backscattering distance.cd Δ 0 : the inner potential correction.* e * f