- Email: [email protected]
Reply to B.E. Nieuwenhuys' comments on “anisotropy of the work function change in physical adsorption“
Surface Science 49 (1975) 681-685 0 North-Holland Publishing Company
REPLY TO B.E. NIEUWENHUYS’
COMMENTS ON “ANISOTROPY
OF
THE WORK FUNCTION CHANGE IN PHYSICAL ADSORPTION” J. MULLER * Received
10 December
1974; manuscript
received
in final form 24 February
1975
The most important property of the work function is its anisotropy. In ref. [l], I have used the physically well justified concept of Smoluchowski ]2] and extended it for physical adsorption of rare gases on metal surfaces. The result of this procedure was a finding that the work function change produced by physical adsorption (A@) is anisotropic, and two simple rules (one quantitative and the other qualitative) were proposed to be able to predict the A# anisotropy (given a rare gas on different crystal faces of the same metal). Nieuwenhuys has recently accumulated a considerable amount of experimental data [3-71 which are directly or indirectly relevant to the subject of ref. [ 11. In the previous letter [8] he selected some data, which, according to his opinion, show that the simple rules are not generally applicable. The confrontation of the simple rules with the bulk of the data available reveals that the rules bring systematics into the field, which is necessary both for obtaining further experimental data and for the development of the appropriate microscopic theory. I turn now to the specific points raised in Nieuwenhuys’ critical letter [8]. (i) In difference with Mignolet [9] (and Klemperer and Snaith [lo]). who used the jellium model [l l-141 of the metal surface, I have used a model in which the electron density variations are considered not only perpendicular to ?he surface but also along the surface of the real metal. In such a model the electrons tend to relax into positions between the surface metal atoms leaving the regions above the surface metal atoms with an excess positive charge. It can be shown that the Smoluchowski original model extended in the above direction is able to account for a number of the work function properties of bare [ 151 as well as physisorbate covered metals [16], including, for example, the remarkable difference between transition and alkali metals as concerns A$ data on physisorption. The jellium model (the uniform positive background model) quoted by Nieuwenhuys is quite irrelevant for the description of the surface property anisotropy just because it considers by definition only negative charges on the surface. (ii) The rule under consideration has been derived for macroscopic single crystals. * See Miiller
[ I] for the original paper, and Nieuwenhuys
[ 81 for the Comments.
682
J. Miiller/Anisotropy
of work function
change in adsorption
I have nevertheless suggested [ 1 ,171 that the known trend of -A$ to decrease with sintering of polycrystalline metal film (the bare film work function @u increases) seems to indicate its wider applicability. The above trends have been observed quite generally, including films of Ni, Fe, Cu and Ag. In his letter Nieuwenhuys admits Ni and Fe only, although in his original’paper [4] there is a complete list of relevant quotations. Previous authors listed the A# measured together with the temperature of deposition and/or annealing, the Go remained undetermined. For that reason I was unable to test the quantitative validity of the rule using the data for polycrystalline films sintered to different extents. Nieuwenhuys has measured both the A@(Xe) and the @n as s function of the annealing temperature (T,) for a number of metal films [4]. In this letter he has chosen just two examples (Ag and Rh) to argue that the rule is poorly obeyed. If we plot Nieuwenhuys’ data [4] in coordinates relevant to the simple rule (i.e. -A@ versus I$~) we obtain the situation shown in fig. 1. Although one would wish to have more points for some of the metals included (e.g. Pt, Ir), I think that the interpretation of the data in fig. 1 is not difficult. The data obtained within the interval of the annealing temperatures, 78 to 400 K, correlate according to the rule very well (the full straight lines in fig. 1). At approximately 400 K another effect comes into operation on top of the crystal face effect. This effect leads to a further decrease of the --A@ values (with the exception of Au) and for simplicity is approximated by the broken straigth lines (there is no reason why this new effect should give a linear correlation in coordinates of fig. 1). This new effect is most likely due to the contamination of the surface at elevated temperatures and came either from the glass substrate (alkali metals) or from the gas phase (residual gases). There remain three metals (Cu, Ag, Ru) studied in ref. [4] where the trend observed does not agree with my predictions. The first two cases hardly need any comments, as the data obtained do not agree with the other determinations including evaporation films [ 10,18,19] as well as macroscopic single crystals [ 191. The results obtained in Leiden with Ru films are known for some time [20,21] and could well indicate a negligible crystal face effect. This is not surprising and I have commented such a possibility in the original paper [ 1 ] already. There is little doubt that the application of the simple rule to Nieuwenhuys’ data helps to understand them better and reveals that the effects observed can be more complex than the crystal face effect alone. An additional confusion which prevented Nieuwenhuys to better understand his data obtained with polycrystalline films came from his additional data for Ir [4,6] obtained by the FEPH (field-emission probe-hole) technique. If it were true that atomically smooth faces such as (111) and (100) of Ir gave much larger A@(M) with Xe (1.8 and 1.6 eV respectively) than atomically rougher faces [-A$(hM) = - 1.O eV] , then why did the -A$ of Xe not increase with increasing r, of Ir films and remained for r, = 78&400 K close to just 1.O eV (cp. fig. 1):’ (iii) Nieuwenhuys points out that there are some A@(M) data which do not support the simple rule. A closer analysis reveals that they include: Xe, Kr/W (110) and Xe/Ir(l 1 l),(lOO). These data have the following common features: they all have
J. MiiNer/Anisotropy
4.5
of work function
5.0
change in adsorption
55
683
%b(eV)
Fig. 1. Maximum change of the work function due to Xe (A@max) on metal films annealed to different extents as a function of the bare film work function (@o). Films were evaporated at 78 K and annealed to Ta K given in brackets. (0) Ni (78, 293,406,570,589); (A) Pd (78, 293, 383, 483, 583); (m) Pt (78, 300, 373, 473, 573); (x) Au (78, 293, 373, 498, 573); (+) Rh (78, 293, 373,458); (0) Ir (78, 293, 373,473). The data are from ref. [4], Ni (570) and Pt (300) from ref. [7] ; Pt (78) is an estimate based on Q,, = 5.45 eV (ref. [3], chapter IV).
been obtained with microscopic single crystals under the presence of strong electric fields (FEPH), and also are in disagreement with the evaporated film data of these metals. The unusually large values of -A@ for Xe and Kr on the (110) region of W tip are known for some time 1221 and I have commented them recently [ 171, It is important that the reliability of just these values was seriously questioned by the authors themselves [22] on the basis of the unusually large pre-exponential term of the Fowler-Nordheim equation and a strong dependence of this term on the surface coverage. On the other hand, the A@(1 11) for Xe on W obtained in FEPH experiments [22] (-1.13 eV) has very recently been confirmed with the use of macroscopic single crystals [23] (-1 .l +O.l eV). A somewhat similar situation emerges from the analysis of Nieuwenhuys’ data for Ir and Pt obtained by the FEPH technique [4 -61 and the photoelectric method with films of Jr and Pt [4]. While both sets of data agree in principle for Pt, they disagree for Ir - particularly as concerns the (11 1) and (100) orientations. In this connection it is important to realize that in all cases studied so far the A@(hkZ) obtained with macroscopic single crystal faces were within the range of the corresponding A$ for films and vice versa (Xe/Cu, Ag, W, Pd). It seems to me very likely that just those FEPH data which violate the simple
684
J. MiillerJAnisotropy of work function change in adsorption
rule (and disagree with the data for films) should not be trusted. It is possible to come to the same conclusion by an independent manner [ 161. The reasons why some of the work function data obtained by the FEPH technique are under suspect are numerous. The Go (Ml) can be in error due to the application of the standard Fowler- Nordheim procedure (this includes the choice of 5 ~ the average value for the tip, and the negligence of the local variations of the field strength). The above difficulties should be diminished when we evaluate L#I (h/cl), the change in work function on adsorption. Here, however, another difficulty arises, as we do not know how the state of the adsorption is altered by the local variation of the field strength over the tip, or even over the particular region of the tip. This difficulty can be serious just for comparatively weak interactions, such as the rare gas physical adsorption. Another difficulty with the FEPH technique originates in the size of the single crystal face under investigation. It is far from obvious that the “single crystal” face of the diameter of some tens of A, which is surrounded by the stepped regions, should be representative of the macroscopic single crystal face. Until the above problems are solved preference should undoubtedly by given to the data obtained with macroscopic techniques. (iv) In his final comment Nieuwenhuys points out that all his experimental results can be interpreted within the CTNB (charge-transfer no-bond) model. The reasonings which led him to reject the polarization model and to prefer the CTNB model have already been published [6,7]. There are, unfortunately, serious inconsistencies in Nieuwenhuys’ analysis of the above two bonding models and this forces me to deal with the subject in more detail elsewhere [ 161. In conclusion 1 wish to say that the confrontation of the simple rules [l] with Nieuwenhuys’ data warrants interesting conclusions concerning their physical meaning (the data for polycrystalline films sintered to different extents) as well as their reliability (the FEPH data). Jaroslav
MLJLLER
Czechoslovak Academ.y of Sciences, J. Heyrovsk$ Institute of Physical Chemistry and Electrochemistry, 7 Mrichova, Prague 2, Czechoslovakia
References [ 1] [2] [ 31 [4] [S]
J. Miiller, Surface R. Smoluchowski, BE. Nieuwenhuys, B.E. Nieuwenhuys, B.E. Nieuwenhuys, 115. [6] B.E. Nieuwenhuys [7] B.E. Nieuwenhuys,
Sci. 42 (1974) 525. Phys. Rev. 60 (1941) 661. Ph.D. Thesis, Leiden (1974). R. Bouwman and W.M.H. Sachtler, Thin Solid Films 21 (1974) 51. D.T. Meijer and W.M.H. Sachtler, Phys. Status Solidi (a) 24 (1974) and W.M.H. Sachtler, Surface Sci. 45 (1974) 513. O.G. van Aardenne and W.M.H. Sachtler, Chem. Phys. 5 (1974)
418.
J. Miiller/Anisotropy [8] [9] [lo] [ll] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
of work function
change in adsorption
B.E. Nieuwenhuys, Surface Sci. 49 (1975) 363. J.C.P. Mignolet, Discussions Faraday Sot. 8 (1950) 105. D.F. Klemperer and J.C. Snaith, Surface Sci. 28 (1971) 209. J. Bardeen, Phys. Rev. 49 (1939) 653. J.R. Smith, Phys. Rev. 181 (1969) 522. N.D. Lang and W. Kohn, Phys. Rev. Bl (1970) 4555. N.D. Lang and W. Kohn, Phys. Rev. B3 (1971) 1215. .I. Mtiller, to be published. J. Mtiller, Chem. Phys., to be published. J. Mtiller, Surface Sci. 45 (1974) 314. A.G. Knapp and M.H.B. Stiddard, J.C.S. Faraday I68 (1972) 2139. M.A. Chesters, M. Hussain and J. Pritchard, Surface Sci. 35 (1973) 161. R. Bouwman and W.M.H. Sachtler, Ber. Bunsenges. Physik. Chem. 74 (1970) J. Mtiller, Chem. Commun. (1970) 1173. T. Engel and R. Gomer, J. Chem. Phys. 52 (1970) 5572. M.J. Dresser, T.E. Madey and J.T. Yates, Jr., Surface Sci. 42 (1974) 533.
685
1273.
REPLY TO B.E. NIEUWENHUYS’
COMMENTS ON “ANISOTROPY
OF
THE WORK FUNCTION CHANGE IN PHYSICAL ADSORPTION” J. MULLER * Received
10 December
1974; manuscript
received
in final form 24 February
1975
The most important property of the work function is its anisotropy. In ref. [l], I have used the physically well justified concept of Smoluchowski ]2] and extended it for physical adsorption of rare gases on metal surfaces. The result of this procedure was a finding that the work function change produced by physical adsorption (A@) is anisotropic, and two simple rules (one quantitative and the other qualitative) were proposed to be able to predict the A# anisotropy (given a rare gas on different crystal faces of the same metal). Nieuwenhuys has recently accumulated a considerable amount of experimental data [3-71 which are directly or indirectly relevant to the subject of ref. [ 11. In the previous letter [8] he selected some data, which, according to his opinion, show that the simple rules are not generally applicable. The confrontation of the simple rules with the bulk of the data available reveals that the rules bring systematics into the field, which is necessary both for obtaining further experimental data and for the development of the appropriate microscopic theory. I turn now to the specific points raised in Nieuwenhuys’ critical letter [8]. (i) In difference with Mignolet [9] (and Klemperer and Snaith [lo]). who used the jellium model [l l-141 of the metal surface, I have used a model in which the electron density variations are considered not only perpendicular to ?he surface but also along the surface of the real metal. In such a model the electrons tend to relax into positions between the surface metal atoms leaving the regions above the surface metal atoms with an excess positive charge. It can be shown that the Smoluchowski original model extended in the above direction is able to account for a number of the work function properties of bare [ 151 as well as physisorbate covered metals [16], including, for example, the remarkable difference between transition and alkali metals as concerns A$ data on physisorption. The jellium model (the uniform positive background model) quoted by Nieuwenhuys is quite irrelevant for the description of the surface property anisotropy just because it considers by definition only negative charges on the surface. (ii) The rule under consideration has been derived for macroscopic single crystals. * See Miiller
[ I] for the original paper, and Nieuwenhuys
[ 81 for the Comments.
682
J. Miiller/Anisotropy
of work function
change in adsorption
I have nevertheless suggested [ 1 ,171 that the known trend of -A$ to decrease with sintering of polycrystalline metal film (the bare film work function @u increases) seems to indicate its wider applicability. The above trends have been observed quite generally, including films of Ni, Fe, Cu and Ag. In his letter Nieuwenhuys admits Ni and Fe only, although in his original’paper [4] there is a complete list of relevant quotations. Previous authors listed the A# measured together with the temperature of deposition and/or annealing, the Go remained undetermined. For that reason I was unable to test the quantitative validity of the rule using the data for polycrystalline films sintered to different extents. Nieuwenhuys has measured both the A@(Xe) and the @n as s function of the annealing temperature (T,) for a number of metal films [4]. In this letter he has chosen just two examples (Ag and Rh) to argue that the rule is poorly obeyed. If we plot Nieuwenhuys’ data [4] in coordinates relevant to the simple rule (i.e. -A@ versus I$~) we obtain the situation shown in fig. 1. Although one would wish to have more points for some of the metals included (e.g. Pt, Ir), I think that the interpretation of the data in fig. 1 is not difficult. The data obtained within the interval of the annealing temperatures, 78 to 400 K, correlate according to the rule very well (the full straight lines in fig. 1). At approximately 400 K another effect comes into operation on top of the crystal face effect. This effect leads to a further decrease of the --A@ values (with the exception of Au) and for simplicity is approximated by the broken straigth lines (there is no reason why this new effect should give a linear correlation in coordinates of fig. 1). This new effect is most likely due to the contamination of the surface at elevated temperatures and came either from the glass substrate (alkali metals) or from the gas phase (residual gases). There remain three metals (Cu, Ag, Ru) studied in ref. [4] where the trend observed does not agree with my predictions. The first two cases hardly need any comments, as the data obtained do not agree with the other determinations including evaporation films [ 10,18,19] as well as macroscopic single crystals [ 191. The results obtained in Leiden with Ru films are known for some time [20,21] and could well indicate a negligible crystal face effect. This is not surprising and I have commented such a possibility in the original paper [ 1 ] already. There is little doubt that the application of the simple rule to Nieuwenhuys’ data helps to understand them better and reveals that the effects observed can be more complex than the crystal face effect alone. An additional confusion which prevented Nieuwenhuys to better understand his data obtained with polycrystalline films came from his additional data for Ir [4,6] obtained by the FEPH (field-emission probe-hole) technique. If it were true that atomically smooth faces such as (111) and (100) of Ir gave much larger A@(M) with Xe (1.8 and 1.6 eV respectively) than atomically rougher faces [-A$(hM) = - 1.O eV] , then why did the -A$ of Xe not increase with increasing r, of Ir films and remained for r, = 78&400 K close to just 1.O eV (cp. fig. 1):’ (iii) Nieuwenhuys points out that there are some A@(M) data which do not support the simple rule. A closer analysis reveals that they include: Xe, Kr/W (110) and Xe/Ir(l 1 l),(lOO). These data have the following common features: they all have
J. MiiNer/Anisotropy
4.5
of work function
5.0
change in adsorption
55
683
%b(eV)
Fig. 1. Maximum change of the work function due to Xe (A@max) on metal films annealed to different extents as a function of the bare film work function (@o). Films were evaporated at 78 K and annealed to Ta K given in brackets. (0) Ni (78, 293,406,570,589); (A) Pd (78, 293, 383, 483, 583); (m) Pt (78, 300, 373, 473, 573); (x) Au (78, 293, 373, 498, 573); (+) Rh (78, 293, 373,458); (0) Ir (78, 293, 373,473). The data are from ref. [4], Ni (570) and Pt (300) from ref. [7] ; Pt (78) is an estimate based on Q,, = 5.45 eV (ref. [3], chapter IV).
been obtained with microscopic single crystals under the presence of strong electric fields (FEPH), and also are in disagreement with the evaporated film data of these metals. The unusually large values of -A@ for Xe and Kr on the (110) region of W tip are known for some time 1221 and I have commented them recently [ 171, It is important that the reliability of just these values was seriously questioned by the authors themselves [22] on the basis of the unusually large pre-exponential term of the Fowler-Nordheim equation and a strong dependence of this term on the surface coverage. On the other hand, the A@(1 11) for Xe on W obtained in FEPH experiments [22] (-1.13 eV) has very recently been confirmed with the use of macroscopic single crystals [23] (-1 .l +O.l eV). A somewhat similar situation emerges from the analysis of Nieuwenhuys’ data for Ir and Pt obtained by the FEPH technique [4 -61 and the photoelectric method with films of Jr and Pt [4]. While both sets of data agree in principle for Pt, they disagree for Ir - particularly as concerns the (11 1) and (100) orientations. In this connection it is important to realize that in all cases studied so far the A@(hkZ) obtained with macroscopic single crystal faces were within the range of the corresponding A$ for films and vice versa (Xe/Cu, Ag, W, Pd). It seems to me very likely that just those FEPH data which violate the simple
684
J. MiillerJAnisotropy of work function change in adsorption
rule (and disagree with the data for films) should not be trusted. It is possible to come to the same conclusion by an independent manner [ 161. The reasons why some of the work function data obtained by the FEPH technique are under suspect are numerous. The Go (Ml) can be in error due to the application of the standard Fowler- Nordheim procedure (this includes the choice of 5 ~ the average value for the tip, and the negligence of the local variations of the field strength). The above difficulties should be diminished when we evaluate L#I (h/cl), the change in work function on adsorption. Here, however, another difficulty arises, as we do not know how the state of the adsorption is altered by the local variation of the field strength over the tip, or even over the particular region of the tip. This difficulty can be serious just for comparatively weak interactions, such as the rare gas physical adsorption. Another difficulty with the FEPH technique originates in the size of the single crystal face under investigation. It is far from obvious that the “single crystal” face of the diameter of some tens of A, which is surrounded by the stepped regions, should be representative of the macroscopic single crystal face. Until the above problems are solved preference should undoubtedly by given to the data obtained with macroscopic techniques. (iv) In his final comment Nieuwenhuys points out that all his experimental results can be interpreted within the CTNB (charge-transfer no-bond) model. The reasonings which led him to reject the polarization model and to prefer the CTNB model have already been published [6,7]. There are, unfortunately, serious inconsistencies in Nieuwenhuys’ analysis of the above two bonding models and this forces me to deal with the subject in more detail elsewhere [ 161. In conclusion 1 wish to say that the confrontation of the simple rules [l] with Nieuwenhuys’ data warrants interesting conclusions concerning their physical meaning (the data for polycrystalline films sintered to different extents) as well as their reliability (the FEPH data). Jaroslav
MLJLLER
Czechoslovak Academ.y of Sciences, J. Heyrovsk$ Institute of Physical Chemistry and Electrochemistry, 7 Mrichova, Prague 2, Czechoslovakia
References [ 1] [2] [ 31 [4] [S]
J. Miiller, Surface R. Smoluchowski, BE. Nieuwenhuys, B.E. Nieuwenhuys, B.E. Nieuwenhuys, 115. [6] B.E. Nieuwenhuys [7] B.E. Nieuwenhuys,
Sci. 42 (1974) 525. Phys. Rev. 60 (1941) 661. Ph.D. Thesis, Leiden (1974). R. Bouwman and W.M.H. Sachtler, Thin Solid Films 21 (1974) 51. D.T. Meijer and W.M.H. Sachtler, Phys. Status Solidi (a) 24 (1974) and W.M.H. Sachtler, Surface Sci. 45 (1974) 513. O.G. van Aardenne and W.M.H. Sachtler, Chem. Phys. 5 (1974)
418.
J. Miiller/Anisotropy [8] [9] [lo] [ll] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
of work function
change in adsorption
B.E. Nieuwenhuys, Surface Sci. 49 (1975) 363. J.C.P. Mignolet, Discussions Faraday Sot. 8 (1950) 105. D.F. Klemperer and J.C. Snaith, Surface Sci. 28 (1971) 209. J. Bardeen, Phys. Rev. 49 (1939) 653. J.R. Smith, Phys. Rev. 181 (1969) 522. N.D. Lang and W. Kohn, Phys. Rev. Bl (1970) 4555. N.D. Lang and W. Kohn, Phys. Rev. B3 (1971) 1215. .I. Mtiller, to be published. J. Mtiller, Chem. Phys., to be published. J. Mtiller, Surface Sci. 45 (1974) 314. A.G. Knapp and M.H.B. Stiddard, J.C.S. Faraday I68 (1972) 2139. M.A. Chesters, M. Hussain and J. Pritchard, Surface Sci. 35 (1973) 161. R. Bouwman and W.M.H. Sachtler, Ber. Bunsenges. Physik. Chem. 74 (1970) J. Mtiller, Chem. Commun. (1970) 1173. T. Engel and R. Gomer, J. Chem. Phys. 52 (1970) 5572. M.J. Dresser, T.E. Madey and J.T. Yates, Jr., Surface Sci. 42 (1974) 533.
685
1273.