Chemisorption of molecular nitrogen on palladium surfaces at and

Aug 6, 1984 - technology, inparticular from nitrogen fixation catalysis.1-3. It is well-known ... even at high temperature.4,5 These general trends fo...
0 downloads 0 Views 393KB Size
264

Langmuir 1985, 1, 264-266

study as well as the fact that the exchanges were run in an excess of deuterium while in the present work only a limited amount of deuterium was available on the catalyst.

Acknowledgment. The SX200 Mass Spectrometer was

purchased with funds provided by Grant PRM-8112908 from the National Science Foundation. This research was supported by a grant from the Procter and Gamble Co. Registry No. Pt, 7440-06-4;cyclopentane,287-92-3.

Chemisorption of Molecular Nitrogen on Palladium Surfaces at and above Room Temperature Eizo Miyazaki,* Isao Kojima, and Sumio Kojima Department of Chemistry, Tokyo Institute of Technology, Meguro-ku, Tokyo 152, Japan Received August 6, 1984. In Final Form: November 27, 1984 The adsorption of nitrogen on polycrystalline Pd surfaces under ultrahigh vacuum (uhv) conditions has been studied at and above room temperature by means of TPTD-MS (temperature programmed thermal desorption mass spectrometry), isotopic exchange reaction, and theoretical calculations of cluster models. It is found that a nitrogen molecule in the ground state easily adsorbs at and above room temperature on the Pd surface if the surface was in advance exposed to the atomic nitrogen produced by a hot tungsten filament, and it is suggested that an atomic nitrogen, which is strongly Pound to the surface layer and negatively charged, may create an active site required for chemisorption of a molecular nitrogen. The interaction of a nitrogen molecule with metal surfaces is of considerable interest for both science and technology, in particular from nitrogen fixation It is well-known that this molecule chemisorbs dissociatively at and above room temperature on the transition d metals such as Ti, Zr,Nb, Ta, Cr, Mo, and W, which are situated in the left region of the periodic table, and that on these metals the chemisorbed nitrogen atoms diffuse into bulk a t higher temperature to form the metal nitrides."" However, the adsorption of a nitrogen molecule does not occur at and above room temperature on noble metals such as Pd and Pt either dissociatively or molecularly and the corresponding metal nitride is not formed even at high temperature.4v6 These general trends for the chemisorption of a nitrogen molecule on various transition d metals are understandable from the potential energy curves calculated by one of the present authorss in which the activation energy for dissociation from a nitrogen molecule is considerably higher on the noble metals. The weak molecularly adsorbed states with desorption energy of 10-60 kJ/mol have been observed at lower temperature on various metals, but when the surfaces were allowed to warm up toward room temperature, the adsorbates simply desorb.49'271321

On the other hand, nitrogen is known to adsorb on Pt, Ir, Rh, Pd after the gas was atomized by a high-frequency di~charge,'~J~ hot W filament,16J7or Pd surface by electron bombardment for more than 1h.l* However, the details of the adsorbate characteristics have not been elucidated. We first report here that the chemisorption of a molecular nitrogen occurs with a desorption energy of 123 kJ/mol at and above room temperature on the Pd surface if the surface was exposed in advanced to the atomic nitrogen produced by a hot W filament and that this chemisorptive site for molecular nitrogen is induced by the presence of the most strongly adsorbed or absorbed nitrogen. A polycrystalline Pd plate (18 X 15 X 0.1 mm, 99.99%) obtained from Johnson-Matthey Ltd.,a tungsten filament for producing the nitrogen atoms in the gas phase, and a mass spectrometer were set in a uhv chamber. The base torr (1torr = 133.3 Pa). The appressure was 3 X paratus used is similar to the one reported by Wilf and Dawson.17 In the TPDS-MS measurements, the temperature of the Pd plate was raised at constant rate of 15 K/s by a IR lamp and monitored by a Pt-Pt/Rh thermocouple which was spot welded to the Pd sample. The sample plate was heated at 1073 K in vacuo and then cleaned by alternate treatment of exposure to oxygen and hydrogen.lg

(1)Leigh, G. J. "The Chemistry and Biochemistry of Nitrogen Fixation"; Postgate, J. R., Ed.; Plenum: London, 1971. (2)Hardy, R. W. A ' Treatise on Dinitrogen Fixation"; Wiley: New York. 1979. - _._ , -(3)Ozaki, A.; Aika, K. Catal.: Sci. Technol. 1981,1, 87. Ogata, Y.; Aika, K.; Onishi, T. Surf. Sci. 1984,140,L285. (4)Broden, G.; Rhodin, T. N.; Brucker, C.; Benbow, R.; Hurrych, Z. Surf. Sci. 1976, 59,593. (5)Miyazaki, E. J. Catal. 1980,65,84 and references therein. (6) Brearley, W.; Surplice, N. A. Surf. Sci. 1977,62,93. (7)Yates, J. T.,Jr.; Madey, T. E. J. Chem. Phys. 1966,43, 1055. (8)Bozso, F.; Ertl, G.; Weies, M. J. Catal. 1977,50,519. (9)Housley, H.; King, D. A. Surf. Sci. 1977,62,93. (IO) Foord, J. S.; Goddard, P. J.; Lambert, k.M. Surf. Sci. 1980,94, 339. (11)Kishi, K.;Roberta, M. W. Surf. Sci. 1977,62,252.

(12)Shigeishi, P. A.; King, D. A. Surf. Sci. 1977,62,379. (13)Hendrickx, H. A. C. M.; Hoek, A.; Nieuwenhuys, B. E. Surf. Sci. 1983,81,135 and references therein. (14)Schwaha, K.;Bechtold, E. Surf. Sci. 1977,66,343. (15)Kiss, J.;Berko, A,; Solymosi, F. Roc. Znt. Conf. Solid Surf. 4th 1980,1, 521. (16)Mimeat, V. J.; Hansen, R. S., J. Phys. Chem. 1966, 70, 3001. (17)Wilf, M.; Dawson, P. T. Surf. Sci. 1976,60,561. (18)Kunimori, K.; Kawai, T.; Kondow, T.; Onishi, T.; Tamaru, K. Surf. Sci. 1976,54, 525. (19)Kojima, I.; Miyazaki, E.; Yasumori, I. J. Chem. Soc., Faraday Trans. 1 1982,78,1423. (20)Obuchi, A.;Naito, S.; Onishi, T.; Tamaru, K. Surf. Sci. 1982,122, 235. (21)Ibbotaon, D. E.; Wittrig, T. S.; Weinberg, W. H. Surf. Sci. 1981, 110. 313.

0743-746318512401-0264$01.50/0 0 1985 American Chemical Societv

Letters

Langmuir, Vol. 1, No. 2, 1985 265

;?!AL ,

,

,

8

iC

,

m Y

5 = c

C

2

4

6

NUYBCR OF N A T W

XlC13 cm-’

Figure 2. Relationship between the number of molecular adsorbates I and that of atomic adsorbates 111.

5Lw

700

900

1100

TEMPERATURE / K

Figure 1. Thermal desorption spectra of nitrogen from polycrystalline P d (a) s ctrum of lsNzfrom Pd after exposed at 373 K to lsNzat 2 X 10B”torr for 30 min under a hot W filament; (b) spectrum of I6N2from Pd after exposed to I6N2under the same conditions as those in (a) and then heated at 473 K for 5 min in vacuo; (c) spectrum of laNzfrom Pd after exposed to lsN2under the same conditions as those in (a) and then heated at 773 K for 5 min in vacuo; (d) spectrum of I6N2from Pd after treatment under the same conditions as those in (c) and further exposed to lsN2at room temperature without using the hot W filament. After these procedures, no impurities were detected on the surface by XPS and UPS spectra. However, in order to eliminate the contribution of CO (28 amu), the TD measurements were made by using labeled nitrogen, 15N2. First it was confirmed that no TD spectrum was observed through the clean surface to which nitrogen molecules had been exposed. However, once the W filament was heated in nitrogen gas, the TD spectrum was obtained: Figure l a shows the spectrum obtained for the Pd sample that had been exposed at 373 K to 16N2gas molecules at 2 X 10” torr for 30 min under the heated W filament at about 2000 K. Three peaks, designated by I, 11, and 111, appeared with the temperature maxima (T-) at 493,633, and 973 K, respectively. In order to obtain more detailed information of the absorbatea correspondingto these peaks (designated by adsorbates I, 11, and 111, respectively), the Pd sample thus covered with the different kinds of adsorbed states of nitrogen was heated at 473 K for 5 min in vacuo and then the TD measurement was carried out from room temperature to about 1100 K. As expected, peak I had disappeared (Figure lb) from the original spectrum, Figure la. Similarly the covered surface with the adsorbates 1-111 was heated at 773 K for 5 min. The spectrum obtained in this case was shown in Figure IC where the peaks I and I1 have disappeared, showing that the adsorbates I and 11, among the three kinds of adsorbates, have been completely desorbed from the surface by this treatment. The sample thus heated at 773 K was cooled to room temperature so as to keep the only remaining adsorbates, 111, on the surface, and usual ground-state 15Nzmolecules at 2 X lo-’ torr were exposed for a few minutes to this sample without heating the W filament. The TD spectrum for this sample is shown in Figure Id: A new peak (shadowed peak) has appeared at the same temperature as that for the peak I in Figure l a with similar shape too,indicating that the adsorbed states corresponding to the peak I and the new peak are same.

For the sample that was desorbed at 1073 K, followed by exposure to the ground-state nitrogen gas molecules at room temperature without heating the W filament, no peaks have been observed. These results indicate an important fact: that the active sites for the chemisorption of nitrogen molecules are created when the adsorbates 111 are present in advance on the surface layer. In order to make the nature of these adsorbates clearer, the respective peak was analyzed by using many TD spectra obtained by varying the exposure of nitrogen gas or the adsorption temperature of the Pd sample: T- for peak I is found to be independent of the concentration of the adsorbates I, suggesting that the desorption occurs obeying first-order kinetics. When an equimolar mixture of 15N2and 14N2was exposed to the surface covered with the adsorbates I11 only, the TD peak corresponding to 29 amu was not observed, indicating that the exchange reaction, 15N2+ 14N2= 2l5NI4N, does not occur, therefore the adsorbed state of the. adsorbates I is molecular, in agreement with the first-order desorption rate. Assuming the exponential factor to be 1 X 1013/sfor the desorption rate, activation energy for the desorption was evaluated to be 123 kJ/mol for the adsorbates I. However, isotope mixing was observed for the peaks I1 and 111. The desorption peak position of the adsorbates I1 is identical with that of atomically adsorbed nitrogen produced by dissociation of adsorbed NO on a Pd surface.% In addition, when the adsorption temperature was raised to 473 or 523 K, the intensity of peak I1 obtained by the TD experiment was found to be decreased, whereas, to the contrary, that of peak I11 was increased. These results could suggest that the adsorbates I1 are atomically adsorbed on the surface active sites and that they penetrate with increasing temperatures below some layers of the Pd surface to become the adsorbates I11 due to requirement of activation energy for the penetration. Further, Figure 2 shows how the number of the adsorbates I varies with that of the adsorbates I11 Interestingly, it is seen that the number of saturated molecular adsorbates I k proportional to that of atomic adsorbates 111 with a slope of 45O. One possible interpretation of this result along with the above results is that the chemisorptive sites for the molecular adsorbates I are induced by atomic adsorbates 111, which form a monoatomic underlayer. Formation of a monoatomic underlayer of nitrogen has been observed on a Ti single crystal by Marcus et aLZ2 This possibility concerning the chemisorptive sites for molecular nitrogen was theoretically considered by using two different cluster models by the DV-XaMO (discrete (22) Shih, H.D.; Jepsen, P.W.; Marcus, P.M. Phy. Rev. Lett. 1976, 36,798

Book Reviews

266 Langmuir, Vol. I , No. 2, 1985

0 r

0

N(3) 2nd

lcyev

Ocr Pdt2I

Figure 3. Schematic diagram of model clusters (Pd4-N2 or NPd,-N,) used for calculations. Table I. Comparison of Calculated Atomic Charges and Overlap Populations for Two Model Clusters, NF'd3-N2 and Pdr-Nz

N(1) -0.116 N(2) +1.373 -0.010 Pd(1) -0.101 +0.169

-0,124

f0.165

+1.375 -0.023 -0.096 +0.139 iO.009

variational-Xa molecular orbital) method.23 As shown previously," the DV-Xa cluster method is powerful to treat surface and chemisorption systems. The main results calculated are briefly described below. The details will be published elsewhere. The model clusters used here to represent the Pd(ll1) surface can be depicted schematically in Figure 3: The (23) Miyazaki, E.; Tsukada, T.; Adachi, H. Surf. Sci. 1983,131, L390. (24) Tsukada, M.; Miyazaki, E.; Adachi, H. J. Phys. SOC.J p . 1981, 50, 3032.

first layer consists of three Pd atoms and the %fold site has a P d or N atom in the second layer.23 The computational details of the method have been described elsewhere.24 A representative result at h = 2.501 au is given in Table I: The overlayers of N atoms are negatively charged with the similar values for both clusters of NPd3-N2 and Pd4-N2, while the surface Pd atoms are positively charged with significantly different values for the two clusters, i.e., a larger value (+0.165) for the NPd3-N2 cluster than that (+0.009) for the Pd4-N2 cluster, suggesting that an electrical Coulomb force between the adsorbed N2 molecule and the surface Pd atoms is operative to form the surface dipole. Further, the overlap populations between the N(2) atom and Pd(1) atom are larger by about 20% for the NPd3-N2 cluster (+0.169) than for the Pd4-N2 cluster (+0.139). Both results support that an underlayer of N atoms enhances the attractive interaction of N2 molecules with the surface P d atoms, which is qualitatively in agreement with the experimental results described above. Finally, similar experiments to those described above have been carried out on Pt surfaces; two desorption peaks, a small peak at 450 K and a large peak at 520 K, have been observed only when the W filament was heated. The results are very similar to those obtained by Schwaha and Bechtold14who used a high-frequency discharge tube for producing an activated nitrogen. This is another evidence that the chemisorption sites for molecular nitrogen on the Pd surface are not produced by tungsten deposition on the surface by the hot W filament.

Acknowledgment. The numerical calculations were performed by a M-200H system at the IMS Computer Center. Registry No. Na, 1127-37-9;Pd, 7440-05-3.

Book Reviews Modern Methods of Particle Size Analysis. Edited by H. G. Barth. Wiley, New York. 1984. 309 pages. $55.00.

This book is Volume 73 of a series of monographs on analytical chemistry and ita applicationsand reviews the latest developments in instrumentation for the analysis of particle size and distribution in dispersions in liquid media. Seven of the nine chapters are based on papers presented at the 1981meeting of the Federation of Analytical Chemistry and Spectroscopy Societies in Philadelphia. Surprisingly, two of these are concerned with the characterization of polymers in solution using photon correlation spectroscopy and hydrodynamic chromatography. Both chapters admirably discuss new developments in the application of the techniques but, as some would argue, do not really match the title of the book. The first two chapters provide general reviews of instrumentation. Chapter 1 (Commercial Instrumentation for Particle Size Analysis, by L. G. Bunville) includes the Coulter principle, sedimentation techniques, hydrodynamic chromatography, and nonimaging optical methods (pulse sensor, light scattering, and photon correlation spectroscopy). Chapter 2 by M. J. Groves concentrates on the characterizationof submicron dispersions and emulsions, using image analysis, hydrodynamic chromatography, light scattering, photon correlation spectroscopy, laser doppler

anemometry, and centrifugal and electrical zone sensing devices (Coulter principle). Both accounts include some limited theory and principles of operation and cover (together) most of the commonly used commercial systems, with some repetition. The authors' analysis of advantages and limitations is useful as a preliminary guide to instrument selection and a list of the manufacturers is provided at the end of the book. The remaining chapters specialize. B. B. Weiner (Brookhaven Instruments) writes on photon correlation spectroscopy and Fraunhofer diffraction, covering in a very adequate manner both the theory and data analysis. P. E. Plantz (Leeds and Northrup/Microtrac) discusses light scattering and Fraunhofer diffraction as involved in the development of Microtrac Particle Size Analyzers. K. D. Caldwell reviews the work of his laboratory a t the University of Utah on the development of field-flow fractionation for particle size analysis, and A. Hamielec (McMaster University) reports on detection systems in particle chromatography. The title of the book might suggest t,hat it contains a comprehensive review of particle sizing techniques, but it does not. However, it does contain critical analyses of the state of the art of a number of methods that are still being developed and allows the reader to make a judgement on what would best suit his current requirements and whether there are prospects for improvements over the horizon. G . D. Parfitt, Carnegie-Mellon University