626 Surface Science 143 (1984) 626638 North-Holland, Amsterdam MOLECULAR BEAM STUDIES OF THE DYNAMICS OF ACTIVATED ADSORPTION OF N, ON W(ll0): DISSOCIATION THRESHOLD AND NEW BINDING STATES J. LEE and R.J. MADIX Department of Chemicul Engineering, Stanfvrd Uniuersit_y, Stunford, California 94305, USA and J.E. SCHLAEGEL and D.J. AUERBACH IBM Research Laboratory, San Jose, California 95193, USA Received 26 July 1983; accepted for publication 17 February 1984 The activated adsorption of N, on W(110) was studied using molecular beams to achieve translational energies from 2 to 30 kcal/gmol. The beams were formed in nozzles with and without helium seeding. For the lowest beam kinetic energy the probability of dissociation upon collision was about 3 X lo-‘, increasing to 3 x 10-l at energies above 20 kcal/gmol. The dependence of the dissociation probability on nitrogen coverage suggested the process was direct in nature, not trapping dominated. The results show that the adsorption is translationally activated and that the reactive collision is not adequately described by a single one-dimensional barrier. As the beam energy was increased, the apparent saturation coverage by nitrogen increased, producing an unusual desorption state above a nitrogen atom coverage of 0.25. Nitrogen desorbed autocatalyti- tally from this state in a fashion observed previously for high concentration of oxygen from Pt(100). The emergence of this state at higher surface coverages indicates that the state of nitrogen adsorbed on tungsten and other metals at higher pressures may be quite different from the state normally observed in low pressure studies. 1. Introduction The effect of the surface structure on adsorption is well demonstrated by the dissociative chemisorption of N, molecule on various W single crystal planes. Early work function studies on field emission tips [1] and bulk single crystals [2] showed that whereas the work function of the (100) and (111) planes was changed by exposure to nitrogen, the (110) was unaffected. This result sug- gested that nitrogen dissociation did not occur on the (110) plane. Later studies by Adams and Germer [3] on the (310), (210) and (100) planes led to the conclusion that only sites with geometries characteristic of (100) planes were reactive for N, dissociation. More strikingly the initial sticking probability, Q, 0039-6028/84/$03.00 0 Elsevier Science Publishers B.V. (North-Holland Physics Publishing Division)
MOLECULAR BEAM STUDIES OF THE DYNAMICS OF ACTIVATED ADSORPTION OF N, ON W(ll0): DISSOCIATION THRESHOLD AND NEW BINDING STATES
J. LEE and R.J. MADIX
Department of Chemicul Engineering, Stanfvrd Uniuersit_y, Stunford, California 94305, USA
and
J.E. SCHLAEGEL and D.J. AUERBACH
IBM Research Laboratory, San Jose, California 95193, USA
Received 26 July 1983; accepted for publication 17 February 1984
The activated adsorption of N, on W(110) was studied using molecular beams to achieve
translational energies from 2 to 30 kcal/gmol. The beams were formed in nozzles with and without
helium seeding. For the lowest beam kinetic energy the probability of dissociation upon collision
was about 3 X lo-‘, increasing to 3 x 10-l at energies above 20 kcal/gmol. The dependence of the
dissociation probability on nitrogen coverage suggested the process was direct in nature, not
trapping dominated. The results show that the adsorption is translationally activated and that the
reactive collision is not adequately described by a single one-dimensional barrier. As the beam
energy was increased, the apparent saturation coverage by nitrogen increased, producing an
unusual desorption state above a nitrogen atom coverage of 0.25. Nitrogen desorbed autocatalyti-
tally from this state in a fashion observed previously for high concentration of oxygen from
Pt(100). The emergence of this state at higher surface coverages indicates that the state of nitrogen
adsorbed on tungsten and other metals at higher pressures may be quite different from the state
normally observed in low pressure studies.
1. Introduction
The effect of the surface structure on adsorption is well demonstrated by the dissociative chemisorption of N, molecule on various W single crystal planes. Early work function studies on field emission tips [1] and bulk single crystals [2] showed that whereas the work function of the (100) and (111) planes was changed by exposure to nitrogen, the (110) was unaffected. This result sug- gested that nitrogen dissociation did not occur on the (110) plane. Later studies by Adams and Germer [3] on the (310), (210) and (100) planes led to the conclusion that only sites with geometries characteristic of (100) planes were reactive for N, dissociation. More strikingly the initial sticking probability, Q,
J. Lee et al. / Actioated adsorption of N_, on W(110) 627
differs by two orders of magnitude of these planes. For example, s0 is 0.65 for W(100) [4-61, 0.08 for W(111) [7], and 3 f 1 X 10-j for W(110) [8-lo]. In an extensive study, Singh-Boparai et al. [ll] further concluded that N, dissocia- tion requires vacant pairs of 100 sites and that population of planes which are devoid of such sites proceeds migration of atomic nitrogen from such site pairs. An even lower value (5 x 10P4) of s0 was reported by Lin and Ehrlich [12] on W(110) from studies using field emission microscopy. Due to the strong effects of (100) type defects, variability in the measurement of s0 on W(110) surfaces may be expected.
The complete understanding of the extremely low reactivity of N, dissocia- tion exhibited by specific planes such as W(110) requires a detailed shape of the potential hyper-surface; for simplicity the reaction has been described in terms of a single activation barrier for adsorption. The simplest form of this barrier is one-dimensional (1D). Molecular beam studies by Balooch et al. nicely demonstrated the applicability of a 1D model for H, dissociation on Cu surfaces for which the normal component of translational energy was critical for passage over the barrier [13]. Recently Cosser et al. [14] observed that the
angular distribution of N, in associative desorption of P-nitrogen from W(110) closely fit a cos”@ (3 < n < 4) distribution; they estimated an activation barrier of 4.2 kcal/mol based on the 1D model. Motivated by these earlier observa- tions, we have studied the kinetic energy dependence of N, dissociation on W(110) by molecular beam techniques in order to determine if a threshold in translational energy could be found above which adsorption occurred.
2. Experimental apparatus and procedures
The experiment was performed with the molecular beam apparatus de- scribed elsewhere [15]. Briefly, a UHV chamber (volume of 2: 300 1) was pumped with a liquid N, trapped diffusion pump (S = 400 l/s) and a titanium
sublimation pump. A pressure of less than 5 X lo-” Torr was typically obtained after bake-out at 200 ‘C. The system was equipped with an Auger spectrometer, LEED and an Ar ion sputter gun for sample cleaning.
A tungsten sample (0.6 X 0.8 X 0.15 cm disc) was cut to expose the (110) surface, and both faces were polished by standard techniques. A careful X-ray examination, completed after the measurements reported here, showed the sample was misoriented by 1.39’. The sample was mounted on a rotatable manipulator and initially cleaned by repeated cycles of oxidation at 1800 K under an 0, background pressure of 1O-6 Torr followed by heating to 2500 K until no CO, partial pressure burst was observed with the mass spectrometer during the flash. No detectable amount of surface impurities was observed in Auger spectra.
The sample was heated by electron bombardment while biased at 5OOV. The temperature was monitored with W-5%Re/W-26%Re thermocouple spot-
welded at the edge of the sample. The temperature could be held constant or ramped linearly (in thermocouple vohage versus time) by a temperature controller.
A supersonic beam of N, or NJHe was directed at the W(110) surface (beam diameter = 0.25 cm), and the surface coverage was subsequently mea- sured by temperature programmed desorption (TPD) and Auger electron spectroscopy (AES). The kinetic energy of the N, beam was adjusted by varying the source temperature and/or the N,/He ratio. The relative beam strength was established with a quadrupole mass spectrometer (QMS) mounted on the beam axis operating in stagnation mode. The beam flux was estimated using the pumping speed and chamber pressure rise measured with an ion gauge. The flux was between 1.5 and 7.0 x 10’s molecules/cm~, depending on the beam energy, and was in all cases at least 100 times larger than he impingement rate of background N,. The mean velocity of the N, beam was
also measured by the time-of-flight method. For purposes of comparison to
earlier work which employed exposure from ambient gas, exposures will be stated in langmuirs (L). For reference we note 1 L of N, at 300 K corresponds to 3.83 X 1014 molecules/cm2.
The QMS in the apparatus was not in line of sight with the sample; therefore N, was detected which desorbed from the entire crystal surface. Consequently, it was necessary to perform comparative TPD measurements with the sample in and out of beam under the same conditions in order to correctly subtract the background from the total TPC signal. The background contribution thus measured typically amounted to 15%40% of the total TPD signal, depending on N, exposure and beam energy; it was most significant at large exposure and low beam energy. All the TPD data presented here were corrected for the background contribution. A cracking fraction of nitrogen (mass 14) was monitored to avoid interference from CO whose TPD spectra overlaps with that of N, at high coverage. Before each run the sample was flashed to 2500 K and cooled to 820 K for subsequent exposure to N, beam. The sample temperature was held at 820 K to ensure low coverages of hydrogen during the experiment. The dose time was accurately controlled by a beam flag interfaced to a computer. A sample heating rate of 38 K/s was used in TPD, and the TPD signal was digitized, read by a computer and integrated with simultaneous recording of the QMS analog output on an X-Y recorder.
3. Results
3.1. TPD spectra
TPD spectra taken with low (Ek = 2.04 kcal/mol) and high (E, = 24.9 kcal/mol) N, beam kinetic energy at 45’ incidence angle from the surface
J. Lee et al. / Activated adsorption of N, on W(110) 629
normal are shown in fig. 1 and fig. 2, respectively. For the low energy beam the TPD spectra showed a single desorption peak (p) which shifted from 1430 to 1310 K with increasing coverage. The spectra were somewhat symmetric about the peak, being about 180 K full width at half maximum (FWHM). Saturation coverage was reached at an N, exposure above about 400 L. These results are generally in good agreement with those obtained by Madey and Yates [8] and Tamm and Schmidt [9]. At high beam energy the TPD spectra (fig. 2) were very similar to those obtained with the low energy beam up to N, exposure near 5 L. However, with further N, exposure the coverage could be substan- tially increased beyond the saturation coverage obtainable with the low energy beam. At these higher coverages a second and sharp desorption peak centered at 1245 K appeared in TPD spectra. At saturation coverage a third peak appeared near 1180 K as a small shoulder on the second peak. These peaks are designated as p,, & and & nitrogen in order of increasing desorption tempera-
ture.
N,/W(IIO)
P EK = 2.04 kcol/mol
1000 1200 1400 1600 1800
Surface Temperature ( K)
p2 N2/W(Ii01
E, = 24.9 kcal /mol
1000 1200 1400 1600 1800
Surface Temperature ( K)
Fig. 1. TPD spectra of N,/W(llO) saturated with N, beam of E, = 2.04 kcal/mol at 0, = 45 o and r, = 820 K.
Fig. 2. TPD spectra of N,/W(llO) saturated with N, beam of E, = 24.9 kcal/mol at T, = 820 K
and(a)8,=45O and(b)8,=0°.
The TPD signal was not independently calibrated here. To provide a coverage calibration, we assumed that the onset of the & state corresponded to a coverage of 8 = 0.25 following the recent observation by Somerton and King [lo] that a well-defined p(2 x 2) LEED pattern formed on W(110) after
saturation with ambient Nz at room temperature. Nitrogen atom coverage for all other data points in these experiments was referred to this calibration point. With this reference 8, = 0.5 at the onset of ,8, state. Further, a saturation coverage of @N = 0.22 (3.1 x lOI atoms/cm’) was obtained at 820 K for an Nz beam generated from a room temperature nozzle with a flux-weighted average kinetic energy of 2.04 kcal/mol. This coverage is in excellent agreement with
the values reported by Somerton and King IlO] (3.5 X 1Ol4 atoms/cm”) and Besocke and Wagner [16](2.5 x 1014 atoms/cm’) near 900 K with background N,. Similar values were also reported by Madey and Yates [S] (3 X 10rJ atom/cm2) and Tamm and Schmidt (2 X lOi atoms/cm*) 191 at 300 K.
At intermediate beam energies, only the & or & and p3 states are observed in TPD. The half width of the j3, state was 50 K, and the peak temperatures did not shift with coverage, indicating first order desorption kinetics. However, the results of isothermal desorption below, showed that simple first order kinetics were not obeyed. No thermal conversion between & and & states was observed in a partial desorption of the & state followed by complete desorp- tion of the remaining portion after cooling.
The results of the isothermal desorption of the p, and & states from a saturated surface, in which the temperature was ramped to the leading edge of & state and held constant while monitoring the time evolution of the desorp- tion rate, are shown in fig. 3. It is clear that the desorption rate initially rose to a maximum and then gradually decayed with time, even though the tempera- ture was held constant. For any n th (n 2 0) order desorption kinetics, the rate
-5 0 5 IO 15 20 25
Time (set)
Fig. 3. Isothermal desorption spectra of 8, and & nitrogen saturated W(110).
J. Lee et 01. / Acfioaded adsorprrm of N_, on W(i10) 631
at nonzero time cannot exceed the initial value unless some of the kinetic
parameters change during desorption. Barteau et al. [17] observed similar behavior for oxygen desorption from Pt(100) (5 X 20). Based on LEED and isothermal desorption data, they interpreted their result in terms of a coverage dependent activation energy of desorption due to a strong attractive adatom
interaction, which provided a narrow peak in an autocatalytic desorption process. We believe that similar phenomena are involved for & nitrogen desorption from W( 110).
3.2. Coverage versus IV? exposure
The nitrogen surface coverage measured by TPD at varying N, beam exposures is shown in fig. 4 for a wide range of N, beam kinetic energies. Adsorption was made at 820 K with the N, beam at an incidence angle of 45O. On each curve the coverage which approximately corresponds to the ap-
pearance of p, and p, states in the TPD spectra shown in fig. 2 is marked with an arrow. In order to cross-check the results of fig. 4, experiments were done in which the surface concentration was measured by AES rather than TPD. The results were in complete agreement (see figs. 5 and 6). Furthermore, AES showed that the buildup of contaminants during the experiment was insignifi-
cant. Several significant features of beam energy effects are apparent from fig. 4.
First, the nitrogen atom coverage initially increased more rapidly with ex- posure as the beam energy was increased. In other words, the initial sticking
200 400 600 800
N, Exposure ( Equivalent Langmuir)
Fig. 4. N atom coverage versus N, exposure at various N, beam energy. T, = 820 K and Bi = 45 O.
The onset of & and & states are indicated by arrow marks on each curve.
N (380)
W(350)
250 300 350 4cO
Auger electron Energy kV)
Fig. 5. AES spectra of nitrogen saturated W(I 10) at T, = 820 K and Bi = 45 O. (a) 0, = 2.04
kcal/mol and (b) E, = 30,2 kcal/mol.
1 1 j ’ ’ l_.J-J ’ I 2 4 6 8 10
N, Exposure (102L)
Fig. 6. Auger peak-to-peak ratio y(N(38O}/W(3~~)) versus NZ exposure at vatious N, barn energy. Bi = 45 * and T, = 820 K.
J. Lee et al. / Activated adsarption of N, on W(IlO) 633
probability was dependent on beam energy in a manner indicative of an activated process. The sticking probability was obtained from the slope of the coverage versus exposure curves shown in fig. 4. The initial sticking probabil- ity, sO, is plotted against beam energy, E,, in fig. 7. The s0 values were obtained by a least-squares fit of the initial five to ten data point on the coverage-exposure curves, assuming second order Langmuir adsorption kinet- ics, namely s = s,, (1 - 8,)2. For higher beam energies (Ek 2 1 kcal/mol) measurements were made with a beam chopper with 1% duty cycle to expand the time necessary to measure the initial portion of the 8, versus exposure curves. At the lowest beam energy (E, = 2.04 kcal/mol), se was determined to be 3.5 x 10e3 which is in good agreement with values previously reported by other authors [&LO]. As shown in fig. 7, s0 increased slowly up to a kinetic energy of 10 kcal/mol and then rose rapidly before leveling off above 25
kcal/mol. Over the range of E, studied here, s0 increased by a factor of 80. At the highest beam energy (E, = 30.2 kcal/mol), s0 was 0.28, which is still
N, Beam energy ( kcal /gmol I 020~ ; ip ‘7 2; 2; 3p 3~
0.18
)r 0.16 .z
'5 ?i 0.14
ri: 0.12
ii? '5 0.10
.k= m G 0.08
.E 0.06
0.04
N2 Beam energy (@VI
Fig. 7. Initial sticking probability so versus N, energy at ffi = 45 O. Differentiated curve normalized to the peak is also shown by a dashed curve.
634 J. he et al / Actwated crdsorption of N, on W(ilO)
significantly smalfer than unity. The normal component of the beam energy at the inflection point E, COS*(@~) = 9.7 kcal/mol corresponds to the height of the activation barrier within a 1D model. The half width of the transition region (4 kcal/mol) is much wider than the typical spread of the N, energy (A E/E s 0.15). The results shown here deny a simple interpretation within a single 1D barrier model. This point will be discussed further below.
The sticking probability at nitrogen adatom coverages other than zero is also shown in fig. 8 for four different beam energies. The coverage dependence of the sticking probability was well represented by the function s = sg (1 - 8,)2 at low coverages, but it fell faster than the (1 - 8,)’ dependence predicted at higher coverages. This result implies that N, diss~iation occurred by a direct route and not via a precursor state. In contrast, N, adsorption on W(100) has
been shown to proceed via a weakly bound and highly mobile precursor intermediate, in which the sticking probability remains constant over a wide
range of coverage [4].
E,(kcal/gmol)
0 249
c. 14.0
0 70 0 2.04
* 2 ..= .- n
x g Id2 cn c ._
Y
.o G
z”
10-j
0. I 0.2 0.3 04 0.5
N atom fractional coverage (0!
Fig. 8. Sticking probability S versus N atom coverage at four different N, beam energy and
0,=450.
J. Lee et ul. / Actiuated udsorption of N_, on W(110) 635
Second, the “apparent” saturation coverage continuously increased with beam energy. For example, f?N.sa, varied from 0.22 at E, = 2.04 kcal/mol to
0.52 at E, = 24.9 kcal/mol. The “apparent” saturation coverage is not neces-
sarily the equilibrium coverage, and it is possible that the sicking probability is practically too small (s < 10p5) for further adsorption to occur. This unusual observation cannot be understood within a simple activation barrier model. It is possible that the activation barrier for adsorption is coverage-dependent, and the dissociation process becomes increasingly more activated with increas-
ing coverage due to the lateral interactions among adatoms. Lastly, the appearance of the & state in the TPD spectra occurs at
approximately the same coverage (8 = 0.25) independent of the beam energy; similarly, the p, state grows in at coverage OS. These results imply that the three states (&, & and p,) are sequentially populated, in agreement with the absence of thermal conversion between them.
4. Discussion
It is clear from the above results that dissociation of N, occurs on the W(110) plane with a sticking probability which is dependent on kinetic energy.
The population of the P-nitrogen state on this plane can thus occur directly, in addition to population via surface migration of N atoms from defect sites such as (100) sites. The latter process may be dominant, however, for adsorption carried out with ambient N, at low pressure and temperature and/or on a surface with a relatively high density of (100) sites.
From the energy dependence of the initial sticking probability, sO, on the kinetic energy of N, at 8, = 45 ‘, an activation barrier on one dimension, normal to the surface, of 9.7 kcal/gmol can be inferred. It is interesting to note that only approximately 10m6 of the molecules in a gas with a Boltzmann distribution at 300 K striking a surface have kinetic energy higher than this value. Cosser et al. have estimated a much smaller value of the barrier (4.2 kcal/gmol) from the angular distribution of N, desorption flux from W(110) based on a model in which they postulated that particles are initially ejected into a Lennard-Jones potential field with equal probability in all directions without energy exchange with the surface. It should be noted that the barrier to adsorption may in fact be larger than 9.7 kcal/gmol, as the validity of the one-dimensional mode has not been established. Since the value determined for the barrier height in the desorption experiments of Cosser et al. is model dependent, direct comparison of barrier heights with this work is not fruitful. Rather one should try to build a model which is consistent both with the adsorption and desorption measurements. In fact, the dependence of s0 on beam kinetic energy cannot be explained in terms of any single barrier model since this class of models predicts a step function change in s0 with incident
kinetic energy. Of course, the threshold behavior is expected to be broadened by the actual spread of the kinetic energy of the beam, AE,. At most AE,/E, = 0.20 in the present case. The energy range over which the two order of magnitude increase in s0 occurs is much too broad to be accounted for by this effect. The conclusion is that adsorption is not well represented by a single
one-dimensional potential energy surface. The effective barrier height may consist of a distribution of barrier heights instead of a single valued Euc,, as
suggested by Balooch et al. [13] for the dissociation on low index planes of Cu. In the actual collision the potential energy will depend on the exact location of impact due to the two-dimensional surface corrugation and on the orientation of the impinging molecule near the surface. It is also noteworthy that whereas Balooch et al. measured an increase in s0 at threshold of about a factor of four for I-I, dissociation on Cu, the increase in s0 for N, on W(ll0) was nearly two orders of magnitude. The relative increase in the (110) plane itself may be even more striking, as the effect of low concentrations of defects on the W(ll0) plane in the determination of s0 is not well established, and the value of s0 on a perfect (110) plane at low kinetic energy may be much lower.
From the results some tentative conclusions regarding the dynamics of activated adsorption of N, on W(110) can be drawn. The molecules in the beam are rotationally and vibrationally cold. The N, may be safely assumed to have been in the ground vibrational state, and rotational temperatures were
below 30 IS. Recent experiments in which NO was scattered from Ag(ll1) showed no measurable exchange of energy into the NO stretching vibration for
the scattered molecules. [Is]. Vibrational excitation of N, in the collision with W(ll0) surface is therefore not expected. On the other hand. appreciable rotational excitation of N, is expected by analogy with the NO scattering results [15]. The importance of variation of the rotational temperature of the beam was assessed to some degree by achieving the same beam translational energy of 8.0 kcal/gmol with and without helium seeding. Though in these two cases the r’otational temperature was low (- 10 and - 30 K, respectively), no difference was seen in sO, and it appears highly unlikely that the large increase in s0 was due to differing rotational energy in the beam. It is conceivable that rotationally excited molecules react “on the way out” after collision as they
tumble across the surface. This however seems unlikely. Measurements of the alignment of angular momentum, f, in diatom-surface scattering [15b] show a strong preferential alignment of J parallel to the surface; i.e. the molecules “tumble” end over end. Intuitively, this kind of motion seems unlikely to produce dissociation. Calculations by Tully et al. [18] suggest, furthermore, that increased rotational energy is not effective in producing dissociation. Also, trapping induced by translational-rotational energy exchange is not responsi-
ble for the increase in s0 with E,, as the rapid fall off in the sticking probability with nitrogen coverage indicates it is a direct process. In view of these considerations the reaction appears to be truly translationally activated.
J. Lee et al. / Activated adsorption of N2 on W(110) 631
The increase in translational energy apparently drives the molecule deeper into the repulsive barrier at the surface, facilitating W-N bond formation, leading to dissociation. The details of the dynamics of this process are being studied further.
The significant population of the NZ(&) state is unusual, but the origin of this state is not well established at this time. Clavenna and Schmidt [19] have reported formation of a similar sharp desorption state at 950 K on W(110)
formed by (1) decomposition of NH,, or (2) electron bombardment of the
molecular N, (7) state. The desorption peak temperature for this state did not change with coverage. Each of these methods of formation involves an in- creased driving force; in the case of the ammonia decomposition it is the free energy of the reaction. In the experiments reported here the state was accessed via increased translational energy which produced increased surface coverages. The p, and & states may thus be characteristic of nitrogen on tungsten at elevated temperatures or pressures at which higher adatom coverages are possible.
The autocatalytic evolution of N, from the & state may be due to the decomposition of a surface nitride which becomes less stable as nitrogen desorbs. Similar behavior was reported for oxygen on Pt(lOO) for which a series of complex surface structures are formed. Somerton and King recently concluded that &-nitrogen occupies a subsurface binding site and that the nitrogen atom is sandwiched between the two topmost layers giving rise to a periodic distortion of the first layer. The formation of &nitrogen upon saturation of this state appears to form a new surface phase.
5. Conclusions
The dissociative adsorption of N, on W(110) is translationally activated. The initial sticking probability increases with kinetic energy of impinging N, molecules. At higher kinetic energies nitrogen adatom coverages reach 0.5 monolayer, and a new surface phase is formed which exhibits autocatalytic decomposition kinetics.
Acknowledgments
We would like to thank David A. King for communicating to us the angle
resolved desorption measurements of N, from W(110) [14] prior to publication and for discussions of the experiments reported here in their formative stages. Discussions with John Tully and Herbert Pfnuer are also acknowledged with appreciation. Two of us (R.J.M. and J.L.) gratefully acknowledge the support of the DOE Office of Basic Energy Sciences, Grant DE-AT03-79ER10490.
638 J. Lee et al. / Actrvated dwrption of N, on W(1 IO)
References
[l] G. Ehrlich and F.G. Hudda. J. Chem. Phys. 35 (1961) 1421; 36 (1962) 3233.
[2] (a) T. Oguri, J. Phys. Sot. Japan 18 (1963) 9; 19 (1964) 83:
(b) T.A. Delchar and G. Ehrlich. J. Chem. Phys. 42 (1965) 2686.
