- Email: [email protected]
Oscillatory behaviour of the reduction of NO by H2 and NH3 over Rh studied by FEM
Applied Surface Science 67 (1993) 179-187 North-Holland
applied surface science
Oscillatory behaviour of the reduction of NO by studied by FEM
H 2
and NH 3 over Rh
M.F.H. van Tol, A. G i e l b e r t a n d B.E. N i e u w e n h u y s Department of Heterogeneous Catalysis, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA Leiden, Netherlands Received 10 August 1992; accepted for publication 21 August 1992
The NO reduction by H 2 and NH 3 on Rh has been studied by field emission microscopy (FEM). The N O - H 2 reaction shows oscillatory behaviour at 460 K and PNo = 1.5 × 10 -7 Torr and PH2 = 1 X 10 6 Torr. Unique features of FEM that can be used to study the oscillations in detail are the very high spatial resolution and the presence of, in principle, an indefinite number of different crystal planes. The oscillatory behaviour is reflected by periodic changes in the emission current and in the images observed. The communication between different surfaces present on the field emitter is displayed in real-time on a fluorescent screen. Diffusion and gas phase coupling seem to play a role. Many of the features reported earlier for the oscillatory behaviour of the N O - H 2 and N O - N H 3 reactions over Pt(100) are observed on Rh as well, including the surface explosion. The N O - N H 3 reaction does also show the surface explosion. Quite remarkable features are observed in the latter case. Single planes show oscillatory behaviour under conditions where others do not. The vacancy model proposed earlier for the oscillations over Pt(100), can be applied to the reactions described in this paper as well.
1. Introduction Oscillating reactions in heterogeneous catalysis have attracted much attention during the last years [1]. The rate oscillations observed during the CO oxidation by 0 2 over Pt have been studied most often. A lot of information on these phenomena has been obtained during study of this reaction. The N O - C O reaction over Pt was also studied in this respect and, more recently, the oscillatory behaviour of the N O - H 2 and N O - N H s reactions over Pt(100) has also been described. Most of the recent findings of the latter reactions have been presented and summarized in several papers by our group [2-5]. It was shown that a vacancy model involving periodic coverage changes of NO, oxygen and the hydrogen source was able to explain the oscillations observed during these DeNOx-reactions over Pt(100), and that it should also be applicable to Pt surfaces in general [4,5]. It was suggested that the surface reconstruction between the (1 x 1) and (5 x 20) phases of Pt(100) did take place 0169-4332/93/$06.00 © 1993
during oscillating behaviour, but that they were not the cause of the behaviour observed. Of course more information is needed on this subject at this stage. The latest research efforts have therefore been focused on the observation of oscillations on a variety of surfaces and on various metals, for example the N O - C O reaction on R h ( l l 0 ) [6] and the C O - O 2 reaction on Pd(ll0) [7]. By studying the rate oscillations over these metals more information can be obtained on the influence.of the reversible surface reconstructions on the observation of rate oscillations in general. A very recent study reported for two separated Pd(ll0) single crystals [7] gives elegant proof for the importance of gas phase coupling in the occurrence of oscillations. In this paper we use the high resolution power of field emission microscopy to study the N O - H 2 and N O - N H 3 reactions over the many crystal planes present on a Rh field emitter. This paper describes how to obtain the oscillations reported and summarizes the first results for the N O - H 2 and N O - N H 3 reactions. In the course of our
Elsevier Science Publishers B.V. All rights reserved
180
M.F.H. van Tol et al. / Oscillatory N O reduction by H 2 and NH~ ot,er Rh
experiments we b e c a m e aware of F E M observations of the dynamical behaviour of the C O - O z reaction over a Pt tip [8].
2. Experimental T h e experiments were performed using the digital field emission microscope (DigiFEM) on which we reported earlier [9,10], but now used mainly in the conventional mode. Images observed on a fluorescent screen were recorded in real-time using a C C D video camera. Special care was taken in cleaning the (100)oriented R h field emitters. The base pressure of the U H V system was better than 2 × 10- w Tort. High-purity gases were used because it was found that the use of high-purity gases is essential for long-lasting oscillatory behaviour. All gas pressures m e n t i o n e d are corrected for ion gauge sensitivity. For H 2 the correction factor was assumed to be 0.5. For N O and N H 3 the factor used was 1.3. During the preliminary experiments presented in this paper, the N O - H 2 ratio was varied and its influence on the oscillations was determined. Oscillations can be found easily for the N O - H 2 reaction over Rh, once some basic parameters are known. W e reported earlier the oscillations observed during the N O - H z and N O - N H 3 reactions on Pt(100) [2-5], and we derived a detailed reaction mechanism for these D e N O x - r e a c t i o n s over Pt, involving the important role of vacancies. F r o m the fact that we found that the surface reconstruction of Pt(100) is most likely not necessary for the observation of rate oscillations, the vacancy model should be applicable to many more reactions on many different metals and over a variety of crystallographic orientations. F r o m the conditions u n d e r which oscillations were observed on Pt, it should thus be possible to deduce the conditions u n d e r which the same reactions could show oscillatory behaviour on Rh, keeping in mind the differences in heats of adsorption of NO, Hz and N H 3 between R h and Pt, and also the high strength of the R h - N and R h - O bonds when c o m p a r e d to their Pt equivalent. The dissociation of N O on Rh is thus energetically favoured
in comparison with Pt. To remove the N- and O - a t o m s from Rh a relatively high partial pressure of the reducing agent, in this case H x or N H 3, will be required. The procedure to observe oscillating behaviour is rather easy: H 2 or NH~ is introduced into the system to the desired pressure at 300 K followed by the introduction of NO. The pressure is given time to stabilize. The UHV-system is used in the flow mode. Following pressure stabilization, an image is p r o d u c e d on the fluorescent screen. In our case an emission current of about 2 nA is used most of the time. The images are recorded by the C C D video c a m e r a and stored on tape. T h e tip is heated slowly with applied field from room t e m p e r a t u r e to the temperature at which a so-called surface" explosion occurs. This surface explosion results in a very bright image caused by an explosion in emission current due to N O desorption, dissociation and reaction, which lowers the work function of the system. The applied field was then lowered in order to prevent field effects and tip damage. The temperature is lowered very slowly until N O adsorption occurs, as judged from a lowering of the emission intcnsity at constant applied potential. At this point oscillations are observed. Both emission current and all processes displayed on the fluorescent screen are recorded.
3. Results For sake of clarity, in fig. 1 some of the crystal planes present on a Rh field emitter have been indicated. After introduction of a mixture of 1.5 )< 10 - 7 Torr N O and 1 × 10 -~' Torr H 2, the tip was heated slowly to 530 K, where a surface explosion was observed. The series of images obtained during this explosion is shown in fig. 2. Fig. 2a shows the image obtained just before the surface explosion occurs. The reaction bctwccn N O and H2 starts on the rough planes; on (321) and surroundings (fig. 2b). In fig. 2c the reaction is seen to spread into the direction of (311) and (320). In fig. 2d, almost the entire surface experiences a surface explosion. In fig. 2e the surface explosion has reached its peak. In figs. 2f and 2g
M.F.H. t:an Tol et al. / Oscillatory NO reduction by H 2 and N t t j ocer Rh
the surface becomes occupied with nitrogen atoms, as is judged from the characteristic Nimage. The entire surface explosion took about 0.5 s and was also characterised by a tremendous increase in emission intensity. All images shown in fig. 2 were recorded with a constant applied potential. After the observation of the surface explosion, the temperature was lowered slowly until oscillations were observed at 460 K. Just before the oscillations started a characteristic "fence-like" image was observed (fig. 3a). The appearance of this image is taken as t = 0. Fig. 3b was recorded at t = 1 s and was characterised by a brightening of part of the emission pattern. The oscillation starts around the (321) surface again. Fig. 3c at t = 2.9 s, or 1.9 s after the start of the oscillations, shows that the reaction spreads across the surface in the direction of (310). In figs. 3 d - 3 g it can be seen that the reaction fronts move symmetrically from two sides across the surface until, at t = 14 s, the reaction has covered the entire surface. Note that the (210) planes and areas around (211) do not seem to contribute to the oscillations under these conditions. This was also shown using the DigiFEM in the digital mode. Image 3h was recorded at t = 15 s, so 1 s after the maximum intensity was observed. The emis-
Fig. I. FEM-image of a Rh tip on which various crystal planes have been indicated.
181
sion intensity of the crystal planes decreases, starting on the smoother planes around (100) and between (100) and (111). After a while the image is again that of fig. 3a and the reaction front appears again. The oscillations are very stable for many hours because of the character of the reaction, the good base pressure of the UHV-system and the use of high-purity gases. The character of the reaction is important because in this case the reduction reactions do apparently not contain any component that inhibits the oscillations completely. As will be shown later in this paper for the N O - N H 3 reaction, reaction conditions exist under which parts of the surface are poisoned, but even then oscillations can be observed on several crystal planes. The oscillations are not influenced noticeably by the applied field: increasing and decreasing of the field strength during oscillations, or switching on and off of the applied field did not influence the oscillations to a measurable extent. This was also observed earlier during a surface explosion observed when heating in hydrogen of a Rh tip precovered with NO at 78 K [11]. Presently, more research is devoted to investigate N O - H 2 oscillatory behaviour under different N O / H z-ratios. The first results show that in the presence of a greater surplus of H 2 ( N O / H 2 = 1/13) the temperature at which oscillations are observed is lower than at N O / H 2 = 6.5, in agreement with data published for these reactions on Pt(100) [5]. It also showed that the period of the oscillations decreases when more H 2 is present. In addition to the reaction fronts where, as in fig. 3, the entire surface is involved, oscillations taking place on only one quarter of the surface, a "quadrant", may occur. Under different conditions the four equivalent quadrants seem to be able to act more or less individually in a way that the period between succeeding oscillations on a quadrant is not identical for all quadrants. So one quadrant can show more oscillations in a given time than another. In this case the oscillations do not spread over the entire surface. This is especially observed at lower temperatures, probably indicating that not only gas phase coupling but also diffusion rates can play an important role. The first results indicate that when less H 2 is
182
M.F.H. can Tol et al. / Oscilhtloo' NO n'duction hy tt, and NH~ m,er Rh
p r c s e n t , t h e (320) s u r f a c e s a r c c o v e r e d by N t o
oscillations whcn the NO-H,
s u c h a n e x t e n t t h a t it s e e m s as if t h e d i f f u s i o n o f t h e r e a c t i o n f r o n t s o v e r t h e s u r f a c e is b l o c k e d .
ied. T h e t e m p e r a t u r e
In fig. 4 t h e e m i s s i o n c u r r e n t is s h o w n as a function of time during the observation of rate
reaction was stud-
is 440 K, PNO = 1.5 × 10
7
T o r r a n d Pit+ = 1 × 10 + ~' T o r r . T h e f o u r q u a d r a n t s p r e s e n t o n t h e (lO0)-oricnted f i e l d e m i t t e r h a v e b e e n d e n o t e d by l e t t e r s
Fig. 2. Some imagcs that are observed during a surface explosion during the NO-H= reaction.
PN O =
1.5 × 10 7 Torr,
PH, -- 1 × 10 ~' Torr, T - 530 K. Tile entire surface explosion takes only 0.5 s. (a) The image observed just before the start of a
surface explosion. (b) The start of tile surface explosion. (c) (e) The spreading of the surface explosion across the surface. (F) and (G) The surface is covered by nitrogen atoms following the surface explosion.
M.F.H. ~,an Tol et al. / Oscillatoo, N O reduction by H 2 and N H 3 m'er Rh
183
(f)
Fig. 2. Continued. A - D . Initially a steep rise in the emission current is observed when an oscillation starts. Assuming the vacancy model can be applied to these reactions on Rh, the onset of the oscillations is autocatalytic and therefore very rapid. T h e decrease in emission intensity at the end of an oscillation is caused by the adsorption of reactants on the relatively empty quadrant. T h e period between
oscillations on the various quadrants ~s not the same. It can also be observed that sometimes the quadrants experience an oscillation at the same time or that a reaction front spreads over the surface and covers m o r e than one q u a d r a n t (for example B, C, D). Finally it can be observed that the second peak in fig. 4 (B, C) is higher than the first (A). This is caused by the fact that the
184
M.F.H. t,an Tol et at / Oscillator), NO reduction by H~ attd NH¢ oler Rh
reaction front sometimes one
quadrant.
period
between
is r a t h e r
Under
covers slightly more than constant
the oscillations
constant,
conditions, on one
but the period
the
quadrant
appears
not to
be the same on this matter Oscillations NO-NH
f o r all q u a d r a n t s . is p r e s e n t l y were
3 mixture
More
information
being collected.
also observed
containing
on Rh
3 × 10
7 Torr
for an NH s
Fig. 3. Some of the images recorded during an oscillation observed when studying the N O - H 2 reaction. (a) Characteristic image recorded just before the start of an oscillation. Because of its shape this image is called "'fence-like". PNO = 1.5 × 10 7 Tort, PH~ = 1 X 10 6 Tort, T = 460 K. This image is taken as t - 0. (b) Image recorded at t = 1 s. (c) Image recorded at t 2.9 s. (d)-(g) The reaction fronts move symmetrically across the surface from two sides. (g) is recorded at t = 14 s, at the point where the maximum in the emission current is observed. (h) t = 15 s. The emission intensity of the crystal planes decreases starting on the smoother planes around the (100) plane and between (100) and (11 l).
M.F.H. van Tol et al. / Oscillatory N O reduction by H 2 and N H 3 over R h
185
Fig. 3. Continued.
and 1.3 >( 10 - 7 Torr NO ( N O / N H 3 - r a t i o = 13/30). The surface explosion was completed at 610 K and oscillations started at 594 K under these conditions. Images showed extensive rhodium nitride formation on the surface. Under these conditions only the (511)-(711) areas showed oscillatory behaviour.
4. Discussion The N O - H 2 and N O - N H 3 reactions show oscillatory behaviour over Rh. These oscillations occur under strictly isothermal conditions due to the good heat-conductivity of the sample and filament in combination with the low reactant
186
M.F.H. can Tol et aL / Oscillator), NO reduction by H 2 atut NH~ orer Rh
l(nA)
I ~.0-q
a
2L
a,c,o
~8
72
96
120
Time ($) Fig. 4. Graph displaying the emission current as a function of time during the observation of oscillatory behaviour of the N O - H 2 reaction. T = 440 K, PNO = 1.5× 10 7 Torr, PH~ = 1 × 10 ~ Torr. The four different quadrants present on ~ the fourfold-symmetric (100)-oriented emitter are represented by the letters A - D .
pressures used. Field emission microscopy is an excellent tool to determine the effect of surface structure on the oscillatory behaviour and the surface explosions. The method to obtain the oscillations described is presented. It was shown that bright fronts move across the tip surface via (321) and (320) surfaces. We have interpreted this effect in terms of a moving reaction front. It may also be a diffusion front of a reaction product (most likely NHx). Future experiments are planned to distinguish between these possibilities. Especially (210) and (310), do not participate under these conditions. To our opinion this is caused by the fact that the R h - N bond strength on these planes is probably too strong and the reaction is inhibited by self-poisoning by R h - N . The conditions under which oscillating behaviour is or can be observed on Rh, can be predicted assuming the vacancy mechanism reported for these reactions on Pt(100) can be applied to Rh as well. The N O - N H 3 reaction over Rh does also show the surface explosion and oscillating behaviour. Somewhat different behaviour is observed. Especially interesting is the observation that under some conditions the oscillating behaviour is restricted to planes like (511) and (711). This happens when large patches of R h - N are present on the surface on the rougher planes like (210). Under these conditions there are, apparently, crystal planes where an optimal situation, concerning reactant ratio, total pressure and temperature, is reached for oscillations.
On the rough planes too much R h - N is present to allow oscillations to take place. The interaction between NO and N H 3 is hindered on these planes, and probably no sites are available for the dissociation of N H 3. On the smooth crystal planes the heats of adsorption of NO and N H 3 are lower, dissociation of the reactants is more difficult and the m e t a l - N and m e t a l - O bond strengths are weaker. Most likely oscillations on these planes need different reaction conditions like a different N O - N H 3 ratio a n d / o r another temperature [5]. Under the conditions described here, an optimal situation is obtained for those planes that have an intermediate terrace size, for example facilitating NH 3 adsorption, and steps that facilitate the dissociation of the reactants. On the other hand, these surfaces will not be poisoned by too high concentrations of nitrogen atoms because of the presence of lessreactive terrace sites with a lower R h - N bond strength.
5. Conclusions
The N O - H 2 and N O - N H 3 reactions show oscillatory behaviour over Rh field emission tips. The unique features of F E M make it an excellent tool for the study of oscillating reactions: the resolution is very high (of the order of 20 ,~) and all processes can be studied in real-time. The dynamical behaviour of oscillating reactions like reaction fronts and the surface explosion are displayed clearly on a fluorescent screen and can be recorded by a CCD camera coupled to a video recorder. If desired, the processes can be studied in minute detail using the DigiFEM in the digital mode. The N O - H 2 and N O - N H ~ reactions exhibited oscillatory behaviour under conveniently low pressures and low temperatures. Surface diffusion and gas phase communication seem to play a role. Their relative importance seems to depend on temperature as can be concluded from the first results obtained. Also the extent of surface nitride formation is important. From the results collected here it can be con-
M.F.H. t,an Tol et al. / Oscillatory N O reduction by H 2 and N t l 3 ocer R h
cluded that the need for a surface reconstruction to occur in order to be able to observe the oscillations reported seems rather unlikely. Many crystal planes participate in the surface explosion and the oscillatory behaviour, and for none of these planes reversible surface reconstructions, with a different reactivity of the surfaces towards one of the reactants, have been reported for the reactants used here. The vacancy model as was applied earlier to the N O - H 2 and N O - N H 3 reactions over Pt(100) and Pt in general seems also applicable to Rh. On the basis of this vacancy model the conditions under which oscillations should be observable could be predicted well. Some special p h e n o m e n a are observed for the N O - N H 3 reactions. Under the conditions used here, oscillations are observed at quite high temperatures. The conditions described seem to be the optimal conditions for oscillations on the (511)-(711) planes. On the rougher planes probably too much N is present thus inhibiting adsorption and dissociation of reactants. On the smoother planes the reaction conditions are most likely not right. Currently more research is in progress to study the oscillating reactions over Rh. F E M appears an ideal technique for the study of these intriguing phenomena. More details on the N O - N H 3 reaction will be published elsewhere. Special emphasis will be given to the
187
minimum number of atoms required to establish gas phase coupling [12].
References [1] G. Ertl. Adv. Catal. 37 (1990) 213. [2] J. Siera, P. Cobden, K.I. Tanaka and B.E. Nieuwenhuys, Catal. Lett. 10 (1991) 375. [3] P.D. Cobden, J. Siera and B.E. Nieuwenhuys, J. Vac. Sci. Technol. A, to be published. [4] M.F.H. van Tol, J. Siera and B.E. Nieuwenhuys, Proces Technol. 5 (1992) 22. [5] M.F.H. van Tol, J. Siera, P.D. Cobden and B.E. Nieuwenhuys, Surf. Sci. 274 (1992) 63. [6] V. Schmatloch and N. Kruse, Surf. Sci. 269/270 (1992) 488. [7] M. Ehsasi, M. Berdau, A. Karpowicz, K. Christmann and J.H. Block, 10th Int. Congr. on Catal., Budapest, 1992, Book of Abstracts, p. 33; M. Ehsasi and J.H. Block, Proc. Int. Conf. on Unsteady State Processes in in Catalysis, Ed. Yu.Sh. Matros, VSP Netherlands B.V. 1990, p. 47. [8] V. Gorodetskii and J.H. Block, private communication. [9] M.F.H. van Tol, F.A. Hondsmerk, J.W. Bakker and B.E. Nieuwenhuys, Surf. Sci. 266 (1992) 529. [10] M.F.H. van Tol, F.A. Hondsmerk, J.W. Bakker and B.E. Nieuwenhuys, Surf. Sci. 266 (1992) 214. [11] H.A.C.M. Hendrickx, A.M.E. Winkelman and B.E. Nieuwenhuys, Appl. Surf. Sci. 27 (1987) 458. [12] M.F.H. van Tol, A. de Maaijer-Gielbert and B.E. Nieuwenhuys, Chem. Phys. Lett., submitted.
applied surface science
Oscillatory behaviour of the reduction of NO by studied by FEM
H 2
and NH 3 over Rh
M.F.H. van Tol, A. G i e l b e r t a n d B.E. N i e u w e n h u y s Department of Heterogeneous Catalysis, Gorlaeus Laboratories, Leiden University, P.O. Box 9502, 2300 RA Leiden, Netherlands Received 10 August 1992; accepted for publication 21 August 1992
The NO reduction by H 2 and NH 3 on Rh has been studied by field emission microscopy (FEM). The N O - H 2 reaction shows oscillatory behaviour at 460 K and PNo = 1.5 × 10 -7 Torr and PH2 = 1 X 10 6 Torr. Unique features of FEM that can be used to study the oscillations in detail are the very high spatial resolution and the presence of, in principle, an indefinite number of different crystal planes. The oscillatory behaviour is reflected by periodic changes in the emission current and in the images observed. The communication between different surfaces present on the field emitter is displayed in real-time on a fluorescent screen. Diffusion and gas phase coupling seem to play a role. Many of the features reported earlier for the oscillatory behaviour of the N O - H 2 and N O - N H 3 reactions over Pt(100) are observed on Rh as well, including the surface explosion. The N O - N H 3 reaction does also show the surface explosion. Quite remarkable features are observed in the latter case. Single planes show oscillatory behaviour under conditions where others do not. The vacancy model proposed earlier for the oscillations over Pt(100), can be applied to the reactions described in this paper as well.
1. Introduction Oscillating reactions in heterogeneous catalysis have attracted much attention during the last years [1]. The rate oscillations observed during the CO oxidation by 0 2 over Pt have been studied most often. A lot of information on these phenomena has been obtained during study of this reaction. The N O - C O reaction over Pt was also studied in this respect and, more recently, the oscillatory behaviour of the N O - H 2 and N O - N H s reactions over Pt(100) has also been described. Most of the recent findings of the latter reactions have been presented and summarized in several papers by our group [2-5]. It was shown that a vacancy model involving periodic coverage changes of NO, oxygen and the hydrogen source was able to explain the oscillations observed during these DeNOx-reactions over Pt(100), and that it should also be applicable to Pt surfaces in general [4,5]. It was suggested that the surface reconstruction between the (1 x 1) and (5 x 20) phases of Pt(100) did take place 0169-4332/93/$06.00 © 1993
during oscillating behaviour, but that they were not the cause of the behaviour observed. Of course more information is needed on this subject at this stage. The latest research efforts have therefore been focused on the observation of oscillations on a variety of surfaces and on various metals, for example the N O - C O reaction on R h ( l l 0 ) [6] and the C O - O 2 reaction on Pd(ll0) [7]. By studying the rate oscillations over these metals more information can be obtained on the influence.of the reversible surface reconstructions on the observation of rate oscillations in general. A very recent study reported for two separated Pd(ll0) single crystals [7] gives elegant proof for the importance of gas phase coupling in the occurrence of oscillations. In this paper we use the high resolution power of field emission microscopy to study the N O - H 2 and N O - N H 3 reactions over the many crystal planes present on a Rh field emitter. This paper describes how to obtain the oscillations reported and summarizes the first results for the N O - H 2 and N O - N H 3 reactions. In the course of our
Elsevier Science Publishers B.V. All rights reserved
180
M.F.H. van Tol et al. / Oscillatory N O reduction by H 2 and NH~ ot,er Rh
experiments we b e c a m e aware of F E M observations of the dynamical behaviour of the C O - O z reaction over a Pt tip [8].
2. Experimental T h e experiments were performed using the digital field emission microscope (DigiFEM) on which we reported earlier [9,10], but now used mainly in the conventional mode. Images observed on a fluorescent screen were recorded in real-time using a C C D video camera. Special care was taken in cleaning the (100)oriented R h field emitters. The base pressure of the U H V system was better than 2 × 10- w Tort. High-purity gases were used because it was found that the use of high-purity gases is essential for long-lasting oscillatory behaviour. All gas pressures m e n t i o n e d are corrected for ion gauge sensitivity. For H 2 the correction factor was assumed to be 0.5. For N O and N H 3 the factor used was 1.3. During the preliminary experiments presented in this paper, the N O - H 2 ratio was varied and its influence on the oscillations was determined. Oscillations can be found easily for the N O - H 2 reaction over Rh, once some basic parameters are known. W e reported earlier the oscillations observed during the N O - H z and N O - N H 3 reactions on Pt(100) [2-5], and we derived a detailed reaction mechanism for these D e N O x - r e a c t i o n s over Pt, involving the important role of vacancies. F r o m the fact that we found that the surface reconstruction of Pt(100) is most likely not necessary for the observation of rate oscillations, the vacancy model should be applicable to many more reactions on many different metals and over a variety of crystallographic orientations. F r o m the conditions u n d e r which oscillations were observed on Pt, it should thus be possible to deduce the conditions u n d e r which the same reactions could show oscillatory behaviour on Rh, keeping in mind the differences in heats of adsorption of NO, Hz and N H 3 between R h and Pt, and also the high strength of the R h - N and R h - O bonds when c o m p a r e d to their Pt equivalent. The dissociation of N O on Rh is thus energetically favoured
in comparison with Pt. To remove the N- and O - a t o m s from Rh a relatively high partial pressure of the reducing agent, in this case H x or N H 3, will be required. The procedure to observe oscillating behaviour is rather easy: H 2 or NH~ is introduced into the system to the desired pressure at 300 K followed by the introduction of NO. The pressure is given time to stabilize. The UHV-system is used in the flow mode. Following pressure stabilization, an image is p r o d u c e d on the fluorescent screen. In our case an emission current of about 2 nA is used most of the time. The images are recorded by the C C D video c a m e r a and stored on tape. T h e tip is heated slowly with applied field from room t e m p e r a t u r e to the temperature at which a so-called surface" explosion occurs. This surface explosion results in a very bright image caused by an explosion in emission current due to N O desorption, dissociation and reaction, which lowers the work function of the system. The applied field was then lowered in order to prevent field effects and tip damage. The temperature is lowered very slowly until N O adsorption occurs, as judged from a lowering of the emission intcnsity at constant applied potential. At this point oscillations are observed. Both emission current and all processes displayed on the fluorescent screen are recorded.
3. Results For sake of clarity, in fig. 1 some of the crystal planes present on a Rh field emitter have been indicated. After introduction of a mixture of 1.5 )< 10 - 7 Torr N O and 1 × 10 -~' Torr H 2, the tip was heated slowly to 530 K, where a surface explosion was observed. The series of images obtained during this explosion is shown in fig. 2. Fig. 2a shows the image obtained just before the surface explosion occurs. The reaction bctwccn N O and H2 starts on the rough planes; on (321) and surroundings (fig. 2b). In fig. 2c the reaction is seen to spread into the direction of (311) and (320). In fig. 2d, almost the entire surface experiences a surface explosion. In fig. 2e the surface explosion has reached its peak. In figs. 2f and 2g
M.F.H. t:an Tol et al. / Oscillatory NO reduction by H 2 and N t t j ocer Rh
the surface becomes occupied with nitrogen atoms, as is judged from the characteristic Nimage. The entire surface explosion took about 0.5 s and was also characterised by a tremendous increase in emission intensity. All images shown in fig. 2 were recorded with a constant applied potential. After the observation of the surface explosion, the temperature was lowered slowly until oscillations were observed at 460 K. Just before the oscillations started a characteristic "fence-like" image was observed (fig. 3a). The appearance of this image is taken as t = 0. Fig. 3b was recorded at t = 1 s and was characterised by a brightening of part of the emission pattern. The oscillation starts around the (321) surface again. Fig. 3c at t = 2.9 s, or 1.9 s after the start of the oscillations, shows that the reaction spreads across the surface in the direction of (310). In figs. 3 d - 3 g it can be seen that the reaction fronts move symmetrically from two sides across the surface until, at t = 14 s, the reaction has covered the entire surface. Note that the (210) planes and areas around (211) do not seem to contribute to the oscillations under these conditions. This was also shown using the DigiFEM in the digital mode. Image 3h was recorded at t = 15 s, so 1 s after the maximum intensity was observed. The emis-
Fig. I. FEM-image of a Rh tip on which various crystal planes have been indicated.
181
sion intensity of the crystal planes decreases, starting on the smoother planes around (100) and between (100) and (111). After a while the image is again that of fig. 3a and the reaction front appears again. The oscillations are very stable for many hours because of the character of the reaction, the good base pressure of the UHV-system and the use of high-purity gases. The character of the reaction is important because in this case the reduction reactions do apparently not contain any component that inhibits the oscillations completely. As will be shown later in this paper for the N O - N H 3 reaction, reaction conditions exist under which parts of the surface are poisoned, but even then oscillations can be observed on several crystal planes. The oscillations are not influenced noticeably by the applied field: increasing and decreasing of the field strength during oscillations, or switching on and off of the applied field did not influence the oscillations to a measurable extent. This was also observed earlier during a surface explosion observed when heating in hydrogen of a Rh tip precovered with NO at 78 K [11]. Presently, more research is devoted to investigate N O - H 2 oscillatory behaviour under different N O / H z-ratios. The first results show that in the presence of a greater surplus of H 2 ( N O / H 2 = 1/13) the temperature at which oscillations are observed is lower than at N O / H 2 = 6.5, in agreement with data published for these reactions on Pt(100) [5]. It also showed that the period of the oscillations decreases when more H 2 is present. In addition to the reaction fronts where, as in fig. 3, the entire surface is involved, oscillations taking place on only one quarter of the surface, a "quadrant", may occur. Under different conditions the four equivalent quadrants seem to be able to act more or less individually in a way that the period between succeeding oscillations on a quadrant is not identical for all quadrants. So one quadrant can show more oscillations in a given time than another. In this case the oscillations do not spread over the entire surface. This is especially observed at lower temperatures, probably indicating that not only gas phase coupling but also diffusion rates can play an important role. The first results indicate that when less H 2 is
182
M.F.H. can Tol et al. / Oscilhtloo' NO n'duction hy tt, and NH~ m,er Rh
p r c s e n t , t h e (320) s u r f a c e s a r c c o v e r e d by N t o
oscillations whcn the NO-H,
s u c h a n e x t e n t t h a t it s e e m s as if t h e d i f f u s i o n o f t h e r e a c t i o n f r o n t s o v e r t h e s u r f a c e is b l o c k e d .
ied. T h e t e m p e r a t u r e
In fig. 4 t h e e m i s s i o n c u r r e n t is s h o w n as a function of time during the observation of rate
reaction was stud-
is 440 K, PNO = 1.5 × 10
7
T o r r a n d Pit+ = 1 × 10 + ~' T o r r . T h e f o u r q u a d r a n t s p r e s e n t o n t h e (lO0)-oricnted f i e l d e m i t t e r h a v e b e e n d e n o t e d by l e t t e r s
Fig. 2. Some imagcs that are observed during a surface explosion during the NO-H= reaction.
PN O =
1.5 × 10 7 Torr,
PH, -- 1 × 10 ~' Torr, T - 530 K. Tile entire surface explosion takes only 0.5 s. (a) The image observed just before the start of a
surface explosion. (b) The start of tile surface explosion. (c) (e) The spreading of the surface explosion across the surface. (F) and (G) The surface is covered by nitrogen atoms following the surface explosion.
M.F.H. ~,an Tol et al. / Oscillatoo, N O reduction by H 2 and N H 3 m'er Rh
183
(f)
Fig. 2. Continued. A - D . Initially a steep rise in the emission current is observed when an oscillation starts. Assuming the vacancy model can be applied to these reactions on Rh, the onset of the oscillations is autocatalytic and therefore very rapid. T h e decrease in emission intensity at the end of an oscillation is caused by the adsorption of reactants on the relatively empty quadrant. T h e period between
oscillations on the various quadrants ~s not the same. It can also be observed that sometimes the quadrants experience an oscillation at the same time or that a reaction front spreads over the surface and covers m o r e than one q u a d r a n t (for example B, C, D). Finally it can be observed that the second peak in fig. 4 (B, C) is higher than the first (A). This is caused by the fact that the
184
M.F.H. t,an Tol et at / Oscillator), NO reduction by H~ attd NH¢ oler Rh
reaction front sometimes one
quadrant.
period
between
is r a t h e r
Under
covers slightly more than constant
the oscillations
constant,
conditions, on one
but the period
the
quadrant
appears
not to
be the same on this matter Oscillations NO-NH
f o r all q u a d r a n t s . is p r e s e n t l y were
3 mixture
More
information
being collected.
also observed
containing
on Rh
3 × 10
7 Torr
for an NH s
Fig. 3. Some of the images recorded during an oscillation observed when studying the N O - H 2 reaction. (a) Characteristic image recorded just before the start of an oscillation. Because of its shape this image is called "'fence-like". PNO = 1.5 × 10 7 Tort, PH~ = 1 X 10 6 Tort, T = 460 K. This image is taken as t - 0. (b) Image recorded at t = 1 s. (c) Image recorded at t 2.9 s. (d)-(g) The reaction fronts move symmetrically across the surface from two sides. (g) is recorded at t = 14 s, at the point where the maximum in the emission current is observed. (h) t = 15 s. The emission intensity of the crystal planes decreases starting on the smoother planes around the (100) plane and between (100) and (11 l).
M.F.H. van Tol et al. / Oscillatory N O reduction by H 2 and N H 3 over R h
185
Fig. 3. Continued.
and 1.3 >( 10 - 7 Torr NO ( N O / N H 3 - r a t i o = 13/30). The surface explosion was completed at 610 K and oscillations started at 594 K under these conditions. Images showed extensive rhodium nitride formation on the surface. Under these conditions only the (511)-(711) areas showed oscillatory behaviour.
4. Discussion The N O - H 2 and N O - N H 3 reactions show oscillatory behaviour over Rh. These oscillations occur under strictly isothermal conditions due to the good heat-conductivity of the sample and filament in combination with the low reactant
186
M.F.H. can Tol et aL / Oscillator), NO reduction by H 2 atut NH~ orer Rh
l(nA)
I ~.0-q
a
2L
a,c,o
~8
72
96
120
Time ($) Fig. 4. Graph displaying the emission current as a function of time during the observation of oscillatory behaviour of the N O - H 2 reaction. T = 440 K, PNO = 1.5× 10 7 Torr, PH~ = 1 × 10 ~ Torr. The four different quadrants present on ~ the fourfold-symmetric (100)-oriented emitter are represented by the letters A - D .
pressures used. Field emission microscopy is an excellent tool to determine the effect of surface structure on the oscillatory behaviour and the surface explosions. The method to obtain the oscillations described is presented. It was shown that bright fronts move across the tip surface via (321) and (320) surfaces. We have interpreted this effect in terms of a moving reaction front. It may also be a diffusion front of a reaction product (most likely NHx). Future experiments are planned to distinguish between these possibilities. Especially (210) and (310), do not participate under these conditions. To our opinion this is caused by the fact that the R h - N bond strength on these planes is probably too strong and the reaction is inhibited by self-poisoning by R h - N . The conditions under which oscillating behaviour is or can be observed on Rh, can be predicted assuming the vacancy mechanism reported for these reactions on Pt(100) can be applied to Rh as well. The N O - N H 3 reaction over Rh does also show the surface explosion and oscillating behaviour. Somewhat different behaviour is observed. Especially interesting is the observation that under some conditions the oscillating behaviour is restricted to planes like (511) and (711). This happens when large patches of R h - N are present on the surface on the rougher planes like (210). Under these conditions there are, apparently, crystal planes where an optimal situation, concerning reactant ratio, total pressure and temperature, is reached for oscillations.
On the rough planes too much R h - N is present to allow oscillations to take place. The interaction between NO and N H 3 is hindered on these planes, and probably no sites are available for the dissociation of N H 3. On the smooth crystal planes the heats of adsorption of NO and N H 3 are lower, dissociation of the reactants is more difficult and the m e t a l - N and m e t a l - O bond strengths are weaker. Most likely oscillations on these planes need different reaction conditions like a different N O - N H 3 ratio a n d / o r another temperature [5]. Under the conditions described here, an optimal situation is obtained for those planes that have an intermediate terrace size, for example facilitating NH 3 adsorption, and steps that facilitate the dissociation of the reactants. On the other hand, these surfaces will not be poisoned by too high concentrations of nitrogen atoms because of the presence of lessreactive terrace sites with a lower R h - N bond strength.
5. Conclusions
The N O - H 2 and N O - N H 3 reactions show oscillatory behaviour over Rh field emission tips. The unique features of F E M make it an excellent tool for the study of oscillating reactions: the resolution is very high (of the order of 20 ,~) and all processes can be studied in real-time. The dynamical behaviour of oscillating reactions like reaction fronts and the surface explosion are displayed clearly on a fluorescent screen and can be recorded by a CCD camera coupled to a video recorder. If desired, the processes can be studied in minute detail using the DigiFEM in the digital mode. The N O - H 2 and N O - N H ~ reactions exhibited oscillatory behaviour under conveniently low pressures and low temperatures. Surface diffusion and gas phase communication seem to play a role. Their relative importance seems to depend on temperature as can be concluded from the first results obtained. Also the extent of surface nitride formation is important. From the results collected here it can be con-
M.F.H. t,an Tol et al. / Oscillatory N O reduction by H 2 and N t l 3 ocer R h
cluded that the need for a surface reconstruction to occur in order to be able to observe the oscillations reported seems rather unlikely. Many crystal planes participate in the surface explosion and the oscillatory behaviour, and for none of these planes reversible surface reconstructions, with a different reactivity of the surfaces towards one of the reactants, have been reported for the reactants used here. The vacancy model as was applied earlier to the N O - H 2 and N O - N H 3 reactions over Pt(100) and Pt in general seems also applicable to Rh. On the basis of this vacancy model the conditions under which oscillations should be observable could be predicted well. Some special p h e n o m e n a are observed for the N O - N H 3 reactions. Under the conditions used here, oscillations are observed at quite high temperatures. The conditions described seem to be the optimal conditions for oscillations on the (511)-(711) planes. On the rougher planes probably too much N is present thus inhibiting adsorption and dissociation of reactants. On the smoother planes the reaction conditions are most likely not right. Currently more research is in progress to study the oscillating reactions over Rh. F E M appears an ideal technique for the study of these intriguing phenomena. More details on the N O - N H 3 reaction will be published elsewhere. Special emphasis will be given to the
187
minimum number of atoms required to establish gas phase coupling [12].
References [1] G. Ertl. Adv. Catal. 37 (1990) 213. [2] J. Siera, P. Cobden, K.I. Tanaka and B.E. Nieuwenhuys, Catal. Lett. 10 (1991) 375. [3] P.D. Cobden, J. Siera and B.E. Nieuwenhuys, J. Vac. Sci. Technol. A, to be published. [4] M.F.H. van Tol, J. Siera and B.E. Nieuwenhuys, Proces Technol. 5 (1992) 22. [5] M.F.H. van Tol, J. Siera, P.D. Cobden and B.E. Nieuwenhuys, Surf. Sci. 274 (1992) 63. [6] V. Schmatloch and N. Kruse, Surf. Sci. 269/270 (1992) 488. [7] M. Ehsasi, M. Berdau, A. Karpowicz, K. Christmann and J.H. Block, 10th Int. Congr. on Catal., Budapest, 1992, Book of Abstracts, p. 33; M. Ehsasi and J.H. Block, Proc. Int. Conf. on Unsteady State Processes in in Catalysis, Ed. Yu.Sh. Matros, VSP Netherlands B.V. 1990, p. 47. [8] V. Gorodetskii and J.H. Block, private communication. [9] M.F.H. van Tol, F.A. Hondsmerk, J.W. Bakker and B.E. Nieuwenhuys, Surf. Sci. 266 (1992) 529. [10] M.F.H. van Tol, F.A. Hondsmerk, J.W. Bakker and B.E. Nieuwenhuys, Surf. Sci. 266 (1992) 214. [11] H.A.C.M. Hendrickx, A.M.E. Winkelman and B.E. Nieuwenhuys, Appl. Surf. Sci. 27 (1987) 458. [12] M.F.H. van Tol, A. de Maaijer-Gielbert and B.E. Nieuwenhuys, Chem. Phys. Lett., submitted.