Surface Reconstruction: Noble Metals

The paper reviews the most important achievements in the study of the surface-reconstruction phase transition on the low-index single crystals of Pt, Ir, and Au. A number of methods very sensitive to the surface layers are used to study the stable and metastable structures of the (100), (110), and (...

Ausführliche Beschreibung

Gespeichert in:

Bibliographische Detailangaben
Datum:	2001
Hauptverfasser:	Vasyl’yev, M.A., Bondarchuk, A.B, Tinkov, V.A.
Format:	Artikel
Sprache:	English
Veröffentlicht:	Інститут металофізики ім. Г.В. Курдюмова НАН України 2001
Schriftenreihe:	Успехи физики металлов
Online Zugang:	http://dspace.nbuv.gov.ua/handle/123456789/133378
Tags:	Tag hinzufügen Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:	Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:	Surface Reconstruction: Noble Metals / M.A. Vasyl’yev, A.B. Bondarchuk, V.A. Tinkov // Успехи физики металлов. — 2001. — Т. 2, № 1. — С. 85-108. — Бібліогр.: 46 назв. — англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine

id	irk-123456789-133378
record_format	dspace
spelling	irk-123456789-1333782018-05-25T03:03:22Z Surface Reconstruction: Noble Metals Vasyl’yev, M.A. Bondarchuk, A.B Tinkov, V.A. The paper reviews the most important achievements in the study of the surface-reconstruction phase transition on the low-index single crystals of Pt, Ir, and Au. A number of methods very sensitive to the surface layers are used to study the stable and metastable structures of the (100), (110), and (111) faces. Detailed consideration of the kinetics, critical parameters, activation energy, and geometrical models for 2D-surface phase transitions are presented. Main theoretical results based on the Ising 2D-model, calculations of the free energy with an embedded-atom method, and equilibrium Monte Carlo simulations are also given. В обзоре рассмотрены последние достижения в исследовании реконструкционных фазовых превращений в поверхностных слоях монокристаллов Pt, Ir и Au. Приведены результаты, полученные с помощью различных поверхностно-чувствительных экспериментальных методов. Наиболее подробно рассмотрены вопросы, относящиеся к кинетике, энергии активации, геометрическим моделям, критическим параметрам 2D-фазовых превращений. Представлены также результаты теоретических исследований, основанных на 2D-модели Изинга, методе внедренного атома и моделировании методом Монте-Карло. В обзорі розглянуто останні досягнення в дослідженні реконструкційніх фазових перетворень в поверхневих шарах монокристалів Pt, Ir та Au. Наведено результати, які отримані за допомогою різних повехнево-чутливих експериментальних методів. Найбільш детально розглянуто питання, що відносяться до кінетики, енергії активації, геометричних моделей, критичних параметрів 2D-фазових перетворень. Представлено також результати теоретичних досліджень, що базовані на 2D-моделі Ізінга, методі зануреного атому та моделюванні за методом Монте-Карло. 2001 Article Surface Reconstruction: Noble Metals / M.A. Vasyl’yev, A.B. Bondarchuk, V.A. Tinkov // Успехи физики металлов. — 2001. — Т. 2, № 1. — С. 85-108. — Бібліогр.: 46 назв. — англ. 1608-1021 PACS: 68.35.Bs, 68.35.Dv, 68.35.Gy, 68.35.Ja, 68.35.Rh, 75.70.Rf DOI: https://doi.org/10.15407/ufm.02.01.085 http://dspace.nbuv.gov.ua/handle/123456789/133378 en Успехи физики металлов Інститут металофізики ім. Г.В. Курдюмова НАН України
institution	Digital Library of Periodicals of National Academy of Sciences of Ukraine
collection	DSpace DC
language	English
description	The paper reviews the most important achievements in the study of the surface-reconstruction phase transition on the low-index single crystals of Pt, Ir, and Au. A number of methods very sensitive to the surface layers are used to study the stable and metastable structures of the (100), (110), and (111) faces. Detailed consideration of the kinetics, critical parameters, activation energy, and geometrical models for 2D-surface phase transitions are presented. Main theoretical results based on the Ising 2D-model, calculations of the free energy with an embedded-atom method, and equilibrium Monte Carlo simulations are also given.
format	Article
author	Vasyl’yev, M.A. Bondarchuk, A.B Tinkov, V.A.
spellingShingle	Vasyl’yev, M.A. Bondarchuk, A.B Tinkov, V.A. Surface Reconstruction: Noble Metals Успехи физики металлов
author_facet	Vasyl’yev, M.A. Bondarchuk, A.B Tinkov, V.A.
author_sort	Vasyl’yev, M.A.
title	Surface Reconstruction: Noble Metals
title_short	Surface Reconstruction: Noble Metals
title_full	Surface Reconstruction: Noble Metals
title_fullStr	Surface Reconstruction: Noble Metals
title_full_unstemmed	Surface Reconstruction: Noble Metals
title_sort	surface reconstruction: noble metals
publisher	Інститут металофізики ім. Г.В. Курдюмова НАН України
publishDate	2001
url	http://dspace.nbuv.gov.ua/handle/123456789/133378
citation_txt	Surface Reconstruction: Noble Metals / M.A. Vasyl’yev, A.B. Bondarchuk, V.A. Tinkov // Успехи физики металлов. — 2001. — Т. 2, № 1. — С. 85-108. — Бібліогр.: 46 назв. — англ.
series	Успехи физики металлов
work_keys_str_mv	AT vasylyevma surfacereconstructionnoblemetals AT bondarchukab surfacereconstructionnoblemetals AT tinkovva surfacereconstructionnoblemetals
first_indexed	2025-07-09T18:53:26Z
last_indexed	2025-07-09T18:53:26Z
_version_	1837196598691495936
fulltext	85 PACS numbers: 68.35.Bs, 68.35.Dv, 68.35.Gy, 68.35.Ja, 68.35.Rh, 75.70.Rf Surface Reconstruction: Noble Metals M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov G. V. Kurdyumov Institute for Metal Physics, N.A.S. of the Ukraine, 36 Academician Vernadsky Blvd., UA-03680 Kyyiv-142, Ukraine The paper reviews the most important achievements in the study of the surface- reconstruction phase transition on the low-index single crystals of Pt, Ir, and Au. A number of methods very sensitive to the surface layers are used to study the stable and metastable structures of the (100), (110), and (111) faces. Detailed considera- tion of the kinetics, critical parameters, activation energy, and geometrical models for 2D-surface phase transitions are presented. Main theoretical results based on the Ising 2D-model, calculations of the free energy with an embedded-atom method, and equilibrium Monte Carlo simulations are also given. В обзоре рассмотрены последние достижения в исследовании реконструк- ционных фазовых превращений в поверхностных слоях монокристаллов Pt, Ir и Au. Приведены результаты, полученные с помощью различных поверх- ностно-чувствительных экспериментальных методов. Наиболее подробно рассмотрены вопросы, относящиеся к кинетике, энергии активации, геомет- рическим моделям, критическим параметрам 2D-фазовых превращений. Представлены также результаты теоретических исследований, основанных на 2D-модели Изинга, методе внедренного атома и моделировании мето- дом Монте-Карло. В обзорі розглянуто останні досягнення в дослідженні реконструкційніх фа- зових перетворень в поверхневих шарах монокристалів Pt, Ir та Au. На- ведено результати, які отримані за допомогою різних повехнево-чутливих експериментальних методів. Найбільш детально розглянуто питання, що відносяться до кінетики, енергії активації, геометричних моделей, критичних параметрів 2D-фазових перетворень. Представлено також результати тео- ретичних досліджень, що базовані на 2D-моделі Ізінга, методі зануреного атому та моделюванні за методом Монте-Карло. Key words: surface phase transition, surface reconstruction, low-energy elec- tron diffraction, Ising model, 2D-symmetry. (Received October 27, 2000) Успехи физ. мет. / Usp. Fiz. Met. 2001, т. 2, сс. 85–108 Îòòèñêè äîñòóïíû íåïîñðåäñòâåííî îò èçäàòåëÿ Ôîòîêîïèðîâàíèå ðàçðåøåíî òîëüêî â ñîîòâåòñòâèè ñ ëèöåíçèåé 2001 ÈÌÔ (Èíñòèòóò ìåòàëëîôèçèêè èì. Ã. Â. Êóðäþìîâà ÍÀÍ Óêðàèíû) Íàïå÷àòàíî â Óêðàèíå. 86 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov 1. INTRODUCTION The determination of the atomic structure of metal surfaces, and ad- sorbed phase continues to be the central problem in surface science. Clearly, an understanding of the complex processes involved, for exam- ple, in chemisorption and catalysis requires a detailed knowledge of the positions of substrate and adsorbate atoms. It is well known that the atomic structure of the clean single-crystal metal surfaces may be differ- ent from that obtained by simply truncating the solid [1, 2]. There are, two kinds of such differences, namely the surface relaxation and surface re- construction, which have been studied extensively over the past few years. First is the rigid movement of one, or more, of the first surface lay- ers in the perpendicular direction without of atomic rearrangement within the layers. The last surface phenomenon is associated with some specific phase transition in pure metals: atomic rearrangements within the outer- most surface lattice with new 2D-symmetry. At present the clean metal surface known to reconstruct are the (100) faces of the single-crystal Pt, Ir, Au, Mo, W, V, Cr, the (110) faces of Pt, Ir, Au and the (111) face of Au [1]. It has been found that some surfaces reconstruct under the influence of adsorbates [3–6]. Reconstructed surface superstructures can also oc- cur when clean metal surface has been prepared, spontaneously with temperature. Such specific surface phenomena take place also in the case of ordering alloy systems, for example, Cu–Au, Cu–Pt, Pt–Fe, Co– Ni, Pt–Co [2, 7–9]. The surface reconstruction phase transformation may be reversible or irreversible. It is very important for understanding the sur- face reconstruction mechanism and the nature of the surface phase sta- bility to study the main characters phase transformation, the critical tem- perature, kinetics of surface reconstructions and the critical exponents. Reconstructed low-index surfaces of noble metals (Pt, Ir, Au) have been of great interest to surface scientists ever since their discovery (1965). The particularly striking feature of the (100) and (110) faces of these metals is that the reconstructed phase appears to be stable con- figuration of these surfaces under clean conditions. This behaviour is dif- ferent from that of other reconstructed metal surfaces, e.g. W(100), where a reversible structure transformation observed near 300 K [1]. The surface structure determination of catalytic active metals is of special interest. A detailed knowledge of the actual position of the cata- lyst surface atoms under certain pressure and temperature condition is of critical importance for the kinetics the reaction. Most of the pure metals and bimetallic catalysts used in the industry are based on noble metals [2, 7–9]. Last years the supported Pt-based catalysts are discussed [8– 10]. It should be stressed that we have focused our attention in the present review only on the surface structure of well-defined low-index single crys- tals Pt, Ir and Au, because the kinetics of the adsorption and chemical reaction on pure metals can be understand on the basis of the kinetics Surface Reconstruction: Noble Metals 87 parameters obtained from single-crystal studies. Many surface structure UHV analytical techniques have been used to investigate surface recon- struction. Basically, they can be categorized into reciprocal space tech- niques or diffraction techniques, as, e.g. low-energy electron diffraction (LEED), X-ray diffraction (XRD), and real-space techniques, as, e.g. low- energy ion scattering (LEIS), scanning tunneling microscopy (STM). Bel- low we present some important recent experimental and theoretical re- sults in this direction. 2. PHASE TRANSFORMATION OF THE (100) SURFACES 2.1. Pt(100)(11) and Pt(100)-hex-R Structures The clean Pt(100) surface is known to appear in two crystallographic phases, a metastable (11) structure and a reconstructed phase with a complicated LEED pattern [11]. It is commonly accepted that the latter is caused by a hexagonal close packing of the first atomic layer though the quantitative confirmation of this model by LEED structure determination is only at the beginning [11]. In most cases the superstructure was called (520) according to the size of the coincidence mesh. However this size is to some extent accidental and more reasonable characterization Pt(100)-hex or Pt(100)-hex-R were used. First detail observation of the kinetics structural transition Pt(100)(11) hex by LEED intensities was performed by Heinz et al. [12, 13]. They have shown that the structural transition of Pt(100) surface is irreversible. The (11) structure, which results by an ideal cut through the bulk recon- structs under temperature increase but is not restored by a following cor- responding decrease. So the metastable bulk-like structure cannot be established by a simple temperature treatment, but is prepared by ad- sorption of atoms or molecules which can be carefully desorbed after- wards without structural change of the substrate. The reorganization of surface atoms back to bulk structure during ad- sorption was interpreted by the fact that bulk layer chemical bonds are resorted causing the free energy of the surface to take its absolute mini- mum at the bulk (11) surface structure. This behaviour is demonstrated schematically in Fig. 1. The hexagonal superstructure is metastable at sufficiently high coverages and converts to the (11) structure by thermal activation. Careful desorption of the adlayer makes the surface return to the clean but metastable (11) phase. The clean metastable (11) phase was prepared by intermediate NO adsorption by such way: the clean Pt(100)-hex-R surface was exposed to NO for 5 min at about 3108 mbar and 350 K resulting in a streaked su- perstructure pattern. The streaks disappeared by heating the sample to about 420 K and (11) structure remains with, however, considerable background. This is believed to be due to disordered oxygen which was 88 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov presumably produced by thermal dissociation of NO, while nitrogen de- sorbs. The reduction of oxygen in a hydrogen atmosphere of about 7108 mbar for 5 min at 350 K has given a pure and background free (11) LEED pattern. In order to avoid residual OH molecules on the sur- face the sample was finally heated to 400 K but no further modification of the LEED pattern was observed. The time and temperature dependencies of the integral intensities of both, of decreasing integer order spots as well as growing superstructure spots, were recorded using computer controlled TV method because the change of intensities takes place within the order of minutes down to some ten seconds depending on the actual temperature. For example pure (11) diffraction pattern, (15) superstructure spots develop within the order of a minute when the crystal temperature was increased to about 410 K. The examples of intensity–temperature dependence and intensity–time dependence are presented in Fig. 2, 3. It appears that with increasing temperature, the intensity of the (10) spot decreases at first by simple diffuse scattering described by a Debye–Waller factor as demon- strated by the semi-logarithmic plot in Fig. 2b(1). With a heating rate of about 2 K/s, the transition starts at about 425 K with a steep decrease of intensity and ends above 500 K where again a Debye–Waller like de- pendence begins. In the region of the transition, the intensity of the su- perstructure rises from zero to a maximum, which, however, is influenced by thermal-diffuse scattering dominating the following temperature range (Fig. 2c). Up to about 470 K, the intermediate (15) structure shows up and only at higher temperatures the pattern splits into (520) or more precisely (525) or better Pt(100)-hex. This latter structure is also me- Figure 1. The changes of the free energy as a function of structural configuration and coverage (schematic) of the Pt(100) surface: h—hexagonal structure and q— quadratic structure of the first layer; d—disordered state [12]. Surface Reconstruction: Noble Metals 89 tastable and changes to Pt(100)-hex-R 0.7 at above 1100 K, a structure which corresponds to a 0.7-rotation of the surface layer. It should be clearly pointed out that the metastability of the starting (11) phase for- bids to give a precise transition temperature, which depends on the heat- ing rate. As demonstrated in Fig. 1, the transition (11)hex is believed to be an activated process. Heinz et al. [12] studied LEED intensities as a func- tion of time (Fig. 3) for different but constant elevated temperatures as parameter and estimated the activation energy of the transition, Å  1.1 eV. In the next work Heinz et al. [13] examined in more detail LEED beam profiles and gave a geometrical model of the reconstruction based on the shift of atomic rows. According to this model the structure transi- tion of the metastable (11) phase to the (525) structure takes place via a pure (15) structure with non-splitted but relatively broad 1/5 order spots. This is due to the nucleation of quasi-hexagonally ordered do- mains of the first layer whose diameters were found to be about 100 Å an average [12]. Quasi-hexagonally means, that the surface hexagon is in a way distorted to fit to the quadratic second layer, which mainly results in a unit mesh angle of about 59 instead of the ideal 60. When more and more such domains nucleate and grow during the process of the transi- tion their edges come to meet each other. At elevated temperatures they from larger coherent hexagonally ordered areas. This reduces the half width of a superstructure spots as observed. The beam splitting can be Figure 2. Intensity-temperature dependence at E  85 eV: a—time–temperature characteristic; b—intensity of integer order spot (10) (1—I, 2—lnI); c—intensity of the superstructure spot (06/5) [12]. 90 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov interpreted by a simultaneous reordering of the surface atoms to give a nearly ideal unit mesh angle accompanied by a small compression of lat- eral atomic distances of about 4%. 2.2. Ir(100)(11)  (15) Phase Transformation The reconstructed Ir(100) surface has a rather simple quasi-hexagonal superstructure (15) on the top layer. Several models have been pro- posed in order to explain the (15) LEED patterns such as the hexagonal model, the missing row hexagonal model, the shifted row model [11]. Among them, the buckled hexagonal model with bridge registry survives as the most realistic. The first study kinetics of (11)(15) surface re- construction on Ir(100) was presented by Heinz et al. [14]. Integrated LEED intensities and spot profiles were measured by means of a com- puter-controlled video system [12]. It is known that a clean Ir(100) surface has the reconstructed (15) phase which corresponds to thermodynamic equilibrium for the clean surface. In order to observe the transition (11)(15) the undisturbed metastable (11) configuration must be re- established. This was done by oxygen adsorption. On the reconstructed face adsorption of O2 at 5107 mbar was made at 475 K for about 2 min. After heating to about 750 K and exposing to hydrogen at 550 K (5107 mbar for about 1 min) the LEED pattern was only (11). Figure 3. Intensity–time dependence at 413 K: a—time–temperature change be- haviour; b—intensity of integer order; c—intensity of superstructure spots. E0  32 eV [12]. Surface Reconstruction: Noble Metals 91 After having prepared to clean (11) phase as described, the transition (11)(15) can be started by flashing to temperatures above 880 K. The system overcomes the corresponding activation energy and recon- structs into the hexagonal close packed (15) structure. However the transition from a metastable to a stable state suffers from the fact that the system is far from thermal equilibrium in the beginning and approaches the equilibrium state only for long times. So the transition process is tem- perature and time controlled. This means that no transition temperature can be defined, the transition can not be described by giving a tempera- ture dependence and it is irreversible. Dynamics of the thermal activation reconstruction process Ir(100)(11)(15) was investigated by meas- urement diffraction spot profiles as well integrated intensities as a func- tion of time for different constant temperatures (Fig. 4, 5). It was shown that the transition starts by the formation of intensity streaks along the (11) unit mesh edges and superstructure spots develop from these streaks at the cost of integer-order spots. Both spot widths and intensities change rapidly in the beginning of the transition but subsequently vary very slowly without approaching their equilibrium value. This happens Figure 4. Diffraction spot profiles at 84 eV for the (10) and (11) spots for several times after heating at 940 K: 1—flesh (1300 K); 2—5 min; 3—2 min; 4—1 min; 5—30 s; 6—20 s; 7—10 s; 8—5 s; 9—(11) structure [14]. 92 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov only after a flash to near or above 1300 K. The development of intensity streaks along the (11) unit mesh edges indicates that at least partial disorder appears in the beginning of the transition. The streaks are inter- preted to correspond to linear atomic rows, which have no phase correla- tion. The transition is assumed to start from the ordered phase via shift- ing of atomic rows forming a nearly hexagonal unit mesh. Of course the final quasi-hexagonally close-packed structure shows a higher density of surface atoms that the initial (11) phase. Therefore the formation of steps must be the consequence. The activation energy of the transition was determined from the intensity increase to be 0.88  0.03 eV. 2.3. Structures and Phases of the Au(100) Face Like the (100) faces of Pt and Ir, the Au(100) surface layer is believed to be hexagonal at room temperature in contrast to the square symmetry of the bulk lattice planes lying immediately beneath. Gibbs et al. [15] have presented the most detailed study of the struc- ture and phases of the clean Au(100) surface between room temperature and the bulk melting temperature of 1337 K. These experiments were performed in the glancing-incidence geometry, using an UHV apparatus for X-ray surface scattering. They identified three distinct structural phases exhibited by the Au(001) surface between T  300 K and T  1337 Figure 5. The change of the (06/5) and (6/50) spot profiles. E  99 eV; T  960 K [14]. Surface Reconstruction: Noble Metals 93 K. They are (1) disordered (1170  T  1320 K), (2) distorted-hexagonal (970  T  1170 K), and (3) rotated, distorted-hexagonal (300  T  970). The diffraction pattern observed in the high-temperature phase between T  1170 and 1337 K is shown in (Fig. 6). Solid squares are indexed in bulk-cubic reciprocal-lattice units (h, k, l). The area enclosed by the solid squares corresponds to the substrate unit cell. At elevated temperatures, the diffraction pattern has the (11) symme- try of the bulk, consistent with an unreconstructed or disordered surface layer. Below T  1170 K, there is a reversible transition to an incommen- surate, two-dimensional structure of hexagonal symmetry. Hexagonal re- ciprocal-lattice vectors (1,0)h and (0,1)h are indicated by open triangles in Fig. 7. Surrounding each are additional satellites along directions parallel to the bulk [110] direction; these are indicated by solid circles. The sepa- Figure 6. LEED pattern of the disordered phase [15]. Figure 7. LEED pattern of the disordered-hexagonal phase [15]. 94 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov ration between harmonics is the incommensurability   0.206 ( 0.001) 2 c (c   2ð/c  3.081 Å1 at room temperature). Below T  970 K, additional roads appear around each of the hexagonal rods. These occur at a fixed angle equal to  0.81 away from the [110] direction, and reveal the existence of rotated domains (51). The transition to the rotated, dis- torted-hexagonal phase is reversible. Gibbs et al. [15] also discussed the temperature dependence of the diffraction pattern obtained for the Au(100) surface. For example, the integrated intensities of the rotated and non-rotated components of the hexagonal peak at (1,0)h is plotted in Fig. 8. An abrupt increase in intensity for temperatures decreasing from T  1200 K is seen clearly. The width of the transition is about 10 K. It was found that both the incommensurability and the average rotational angle were weakly temperature dependent. 3. SURFACES WITH (110) ORIENTATION 3.1. Au(110)(11)  (12) Phase Transformation At room temperature the (110) surfaces of Au, Pt, and Ir reconstruct to a structure with (12) symmetry. It was shown above that a variety of ex- periments and total energy calculations clearly indicate that the (12) structure, of this metals is a missing-row (MR) geometry (Fig. 9). It is seen from this figure the ideal (110) surface represents a set of parallel atomic rows and every second row is missing in the (12) phase. Note Figure 8. Intensity–temperature dependence of the non-rotated (1) and rotated (2) components of the hexagonal peak [15]. Surface Reconstruction: Noble Metals 95 that the MR model is highly anisotropic. The attraction along the ]011[ rows is rather strong, and the net repulsion between adjacent rows is weaker by more that a factor of 10 [16]. At some temperature Tc a recon- structed (12) surface undergoes a reversible ‘deconstruction’ to a high temperature (11) structure which is no longer reconstructed. The nature of this phase transition has been studied with variety of experimental and theoretical approaches. The main results of this studies were performed for Au(110)(12)  (11) deconstruction transition. Wolf et al. [17] first investigated the dependence of the integrated in- tensity of the LEED (0, 1/2) reflex as a function of the temperature and have found that a continuous phase a transition from (12) superstruc- ture to the normal (11) structure take place at 713 K. Campuzano et al. [18, 19] have shown that this (12)(11) structural change is an order– disorder phase transition and that the (11) pattern seen in LEED be- longs to a disordered top layer on an ordered second layer in the lattice gas sense. These authors reported a transition temperature at Tc  650 K for the Au(110) surface. They used high resolution LEED systems to a detailed study the line shape and integrated intensity as a function of temperature and reported first measurements of the critical exponents of this continuing phase transition. The results show that the Au(110)(12)(11) transition is strictly two-dimensional, belonging to the Ising universality class with the exponents   0.13  (0.022),  1.75  0.03, and   1.02  0.02 (the predicted values are 0.125, 1.75 and 1, respectively). This was earlier predicted by Bak [20] on the basis of sym- metry considerations. It has been suggested by Campuzano et al. [19] that the high temperature (11) structure is disordered. The transition from ordered missing-row structure to disordered does not involve any mass diffusion, because on average only half monolayers of Au ‘ada- toms’ a top on Au(110) surface one lattice spacing. It is suggested that the missing-row structure is formed as atoms in terraces on the surface migrate the first time the crystal is heated. According to this view, the (12) regions on the surface would have finite sizes, the average size of the terrace widths. On the base of analysis the beam profiles of the (1, 1/2) spots they determined the average size of ordered regions to be  150 Å. Figure 9. Cross section of the missing row structure model. 96 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov Clark et al. [21] have extracted the specific-heat critical exponent  for the Au(110) (12) order–disorder phase transition using integrated LEED intensities. The resulting value   0.02  0.05 is consistent with the pre- dicted Ising universality class (  0) of this transition. They also esti- mated the critical exponent   1.1  0.1 from the measured increase in the FWHM of the (0, 1/2) beam in the range 0.02  t  0.07 (t  (T  Tc)/Tc is the reduced temperature). This result agrees with the Ising value   1. These values for  and  rule out the possibility that the tran- sition is first order concluded Clark et al. They also obtained the critical temperature Tc  695  3 K which is substantially larger than the 650 K [19]. Crark et al. concluded that this difference is consistent with that ex- pected from finite-size effects. The effective critical temperature Te is ex- pected to vary with the scale L of finite-size regions approximately as Tc  Te  T  aLTc, (1) where a is a constant and L is measured in lattice spacing [0.408 nm along the [ ]110 direction for Au(110)]. McRae [22] at latest of the studies, attributes this variability to trace impurities and find Tc  765 K. They investigated the segregation of Sn to the (110) surface of a nominally pure Au crystal by LEED, AES, and low- energy ion scattering. Sn coverages up to 0.2 monolayer were produced by annealing above 675 K. Surface Sn was shown to have a marked ef- fect on the (12)(11) phase transition, e.g. Sn coverages of 0.002 and 0.13 monolayer produced shifts to lower transition temperatures of 15 and 200 K, respectively. Derks et al. [23] reported first results of a near order analysis of the Au(110) surface using low energy ion (K, 300 eV) scattering and in the a computer based data analysis in the temperature range of 300–800 K. The results showed (i) that the top layer atoms at 800 K remain in their regular lattice sites, as suggested previously [18, 19], (ii) the atoms oc- cupy available lattice sites at random, and (iii) the random layer contains as many atoms as the (12) top-layer, i.e. 1/2 monolayer (Figs. 10a and 10b). The near order structure gives the necessary information for a mi- croscopic understanding of the phase transition and may answer the question whether large mass transport is occurring during the phase transition. Derks et al. [23] concluded that considering the LEED results [18, 19] the case of the ‘random’ 1/2 monolayer may be more likely, which requires that at most 1/4 of monolayer has to be moved by one 001 lattice distance during the phase transition. This interpretation of the ion scattering results establishes the hypothesis that the Au(110) sur- face has at all temperature an ‘absorbed’ layer of Au atoms correspond- ing to 1/2 of the full layer. At high temperature (800 K) the adatoms oc- cupy regular lattice sites randomly and form a partly ordered (12) struc- ture at low temperatures. Their it is implicitly assumed that the phase transition occurs between a full (11) surface to the (12) surface with an Surface Reconstruction: Noble Metals 97 occupancy of 1/2 monolayer. The authors agreed with the assumption of ‘lateral jumps’ from the [ ]111 rows into the ]011[ troughs. To produce Figs. 10a and 10b, correlated jumps, as proposed in Ref. [24], are not necessary. Pairs are formed accidentally, pairs and longer chains may be formed in addition due to the thermal diffusion along the rows. Dückers and Bonzel [25] have measured Tc using surface core-level shifts spectroscopy. The core-level binding energy separation between surface and bulk atoms, as well as the intensity ratio of surface-to-bulk contributions in a photoemission spectrum strongly depend on the struc- ture of surface. Many 4f1/2 spectra of the clean Au(110) face for temperature between Figure 10a. Schematic f.c.c. (110) surface—the ideal (12) reconstruction. Large circles—1 st layer, small circles—2 nd layer, black dots—3 rd layer [23]. 98 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov 100 K and 850 K were measured. As can be seen from Fig. 11 values for the core-level shift s,b are nearly constant for temperatures below 500 K and above 750 K but in region 500 K  T  750 K they show a clear de- crease. This behaviour in Fig. 11 indicates a change of the Au(110) sur- face structure between 500 K and 750 K. Dückers and Bonzel concluded that this structure change corresponds to the order–disorder transition of Au(110) from the (12) to the disordered (11) surface. From Fig. 11 they find a transition temperature of Tc  620  50 K. Recently van de Riet et al. [26] have reported ion scattering experi- ments from the Au(110) surface in the temperature range 300 K to 900 K Figure 10b. Schematic f.c.c. (110) surface—the (11) surface with a random 1/2 monolayer adsorbed. Large circles—1 st layer, small circles—2 nd layer, black dots—3 rd layer [23]. Surface Reconstruction: Noble Metals 99 and made some conclusion on the surface structure in the [ ]110 and [001] azimuth. They measured polar angular ion spectra of 4.6 keV neon ions scattered over 120 and compared them to calculated spectra. It was found that below 530 K (120 K below Tc  650 K) the chains in the [ ]110 direction have no vacancies, whereas above this temperature the relative number of vacancies was a strongly increasing function of the tempera- ture. Above the (12)–(11) transition temperature the order in the [ ]110 direction breaks down and spectra indicated the formation of steps above Tc. The spectra measured in the [001] direction indicate that the (12)– (11) transition is an order–disorder transition. The activation energy of vacancy formation was estimated to be 0.2 eV. Most recently, Keane et al. [27] have presented the results of a syn- chrotron X-ray scattering study of the thermal disordering of the Au (110)–(12) reconstructed surface. In this study (12) surface domains extended 550 Å in the [001] direction and 1630 Å in the [ ]110 direction, which are significantly large than those reported in previous studies [19]. Observing the temperature dependence of the superlattice and integral- Figure 11. Energy shift (a) and intensity ration (b) for the clean Au(110) surface as function of temperature [25]. 100 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov order bulk-forbidden (anti-Bragg) surface peaks, as well as the specular (110) reflectivity they found that at Tc  735 K the (12) surface under- goes a deconstruction transition characterized by the proliferation of compact anti-phase defects, with no measurable increase in the density of surface steps. It was found that this transition is described by critical exponents (  0.114,   1.45  0.15,   0.75  0.1) close to those charac- terizing a two-dimensional Ising transition. Keane et al. [27] also found that by 784 K there has been a significant increase in the densities of both anti-phase defects and surface steps of the type associated with surface roughening. This suggests that the Au(110) surface is disordered in a two-step process, deconstruction followed by roughening, with a dif- ference in the two transition temperatures of less that 50 K. 3.2. Surface Reconstruction of the Pt(110) Face The surface of Pt(110) has the same (12) missing row reconstruction as Au(110). The experimental observation of a similar transition for Pt(110) is not so clear as for Au(110) one. The fist attempt of Salmeron and So- morjai [44] to observe Pt(12)(110) transition with LEED was not suc- cessful. They reported that the (12) LEED spot intensity versus tem- perature showed features at 520, 720, 920, 990, and 1040 K. Above the last temperature, the (12) superspot disappeared entirely. However, the authors some concern about oxygen contamination. Recently Dückers and Bonzel [25] have performed an attempt to de- termine the transition temperature for Pt(110) using the temperature de- pendence of surface core-level shifts. Fig. 12 shows the variation of shift s,b and intensity ratio Is/Ib for the Pt(110) surface as a function of tem- perature. Thus all features observed for the order–disorder (12)(11) transition on Au(110) (Fig. 11) are reproduced also for Pt(110), except for a shift of the transition to higher temperature. Unfortunately the Pt sur- face did not stay clean at all temperatures. In the region 850 K  T  1050 K the authors observed potassium corresponding to a coverage of 0.02 on the surface. They explain the experimental results as indication for an order–disorder transition for the Pt(110) surface at 940  50 K. Only most recently two detailed experimental studies the clean Pt(110) recon- structed (12)(11) phase transition and its critical phenomena were reported [28, 29]. Robinson et al. [28] reported X-ray diffraction results for the Pt(110) phase transition, which they observed at Tc  1080  50 K. In many ways they found similar behaviour to Au(110) [19], with one impor- tant difference. Above Tc steps are created spontaneously and their den- sity diverges with temperature. This demonstrates conclusively that Pt(110) roughens above Tc. The simultaneous involvement of steps and reconstruction means that the ground state is at least fourfold degenerate and so the phase transition can no longer be classed as an Ising model. Yet the critical exponents   0.11 and   0.95 are consistent with the Surface Reconstruction: Noble Metals 101 Ising 2D-model, and agree well with Au(110) [18] (although those meas- urements were insensitive to steps). Thus the transition Pt(110) (12)(11) must be classified as a roughening transition and not a sim- ple two-state Ising 2D-transition. The temperature dependence peak height, peak shift and half-width are shown, for example in Fig. 13. Zuo et al. [29] have presented a high-resolution LEED study of the clean Pt(110) reconstructed (12)-to-(11) phase transition and its critical phenomena. The HRLEED (or spot-profile analysis LEED) has a resolu- tion of  6103 Å1 in k-space, about a factor of 10 better than the resolu- tion of conventional LEED. Such a resolution allows to accurately extract the (12) structure factor S(k, T)  I0(T)f(k)  (k, T) (2) from the half-order beams over a wide range of k, so that the critical ex- ponents which are related to S(k, T) can be accurately determined. For the Ising model in the critical scattering region Figure 12. Shift (a) and intensity ration (b) as a function of temperature for Pt(110) surface [25]. 102 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov I0(T)  t 2, (3) 0(T)  (t)  k 2 , (4) k(T)  (t), (5) where t  1  T/Tc is the reduced temperature and , , and  are the criti- cal exponents. Those critical exponents describe the decay of the long- range order, the fluctuation of the short-range order, and the decay of the correlation length of the order parameter, respectively. The critical expo- nents only depend on the symmetry of a system, but not on the detailed interaction energies among atoms. Therefore, various systems can be grouped into a few universality classes. Therefore, a measure of the structure factor S(k, T) from a superlattice beam near Tc allows to deter- mine the critical exponents , , and , as well as to test the scaling rela- tion   (2  ). The theoretical predictions for , ,  and  of the Ising 2D-model are 0.125, 1, 1.72, and 0.25, respectively. Temperature-dependent Bragg peak intensities I(T) of the (0, 1/2) beam are shown in Fig. 14 for E  85.69 and 40 eV. These I(T) curves were reversible in temperature and a rapid decay of the Bragg peak in- tensity occurs around Tc  960 K. The Bragg peak intensity of a superlat- tice beam versus temperature is a measure of the square of the order parameter at T  Tc and is a measure of the fluctuation of short-range or- der at T  Tc. The broadening of the FWHMs begins at almost the same temperature,  960 K, as the drop in peak intensity. The continuous drop Figure 13. Temperature-dependent peak height and its fit by (Tc  T) 2c (left), and peak shift (right) [28]. Surface Reconstruction: Noble Metals 103 of the (0 1/2) beam peak intensity, the continuous broadening of the FWHM, the absence of qualitative changes in the I(E) profiles, all as a function of temperature, clearly indicate that the (12) surface-layer dis- orders around 960 K and the transition is of second order. The anisot- ropic broadening in the [ ]110 and [001] directions Zuo et al. [29] ex- plained by the persistence of the short-range order in the form of short chains of atoms along the [ ]110 direction with little co-ordination among the rows. In the critical region the physical quantities I0(T), 0(T), k(T), which de- scribe the critical phenomena, the authors used an asymptotic power laws of the from Ant n (n  2, , ). The critical exponents (  0.125,   1.74,   0.99) extracted from the (0 1/2) superlattice beam were in excel- lent agreement with the prediction of the Ising 2D-class. However, an in- dication of increasing step density accompanying with the Ising transition was also observed. It was found also that the observed critical scattering maximizes around Tc and was limited by the finite size of the (12) or- dered region. At T  Tc, where only the short-range ordering exists, the structure factor of the superlattice, represented by (0 1/2) beam, showed scaling and isotropic behaviour even though the broadening along the [110] and the [001] directions was anisotropic. Figure 14. Temperature-dependent of the peak intensity of the (01/2) beam; E  85, 69, and 40 eV. The FWHMs is shown in the right hand scale. G(01/2) is the magnitude of the superlattice reciprocal lattice vector spanned from the (0 0) beam to the (0 1/2) beam [29]. 104 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov 3.3. Theoretical Studies Some theoretical studies were also devoted to the problems of the (110) surface f.c.c. crystals. Bak [20] fist has proposed a physical realization of Ising 2D-model to describe critical behaviour for (21)(11) transition. The temperature dependence of the order parameter, the intensities of the Bragg-satellites should go as I  (Tc  T) 2,   1/8. (6) The exponents , , and  can be studied by measuring the critical scat- tering near Tc. This idea was first supported by high-resolution LEED ex- periments [18]. Foiles and Daw [31] found in previous calculations with the embedded- atom method (EAM) that the missing-row structure is indeed lower in en- ergy than the (11) structure for both Pt and Au(110) surfaces. In the next work Daw and Foiles [32] have shown that Monte Carlo simulations using energies computed via the EAM show the occurrence of order– disorder transition on Au and Pt(110). They presented calculations de- termine the equilibrium structure as a function of temperature. Equilibrium Monte Carlo simulations were performed on (110) slab, seven layers thick (with two surfaces), and periodic along the [ ]110 and [001] directions. The geometry chosen for the current calculations represented an (88) surface unit cell. A mesh of ideal sites was laid out over each surface. The initial configurations were used an ideal missing-row structure, a Figure 15. The structure factor for the (12) diffraction spot as a function of tem- perature on Au(110) [33]. Surface Reconstruction: Noble Metals 105 (11) structure, and a random structure. In order to quantity the ordering the two-dimensional structure factor was computed. The structure factor was defined by S i Nj j ( ) exp( )k k R  2 . (7) The sum is over adatoms, k is the wave vector in reciprocal space, Rj is the position of j-th atoms, N is the number of adatoms, and the angular brackets denote an average over configurations. The resulting structure factor for Au evaluated at the (12) diffraction condition as a function of temperature is plotted in Fig. 15. The transition temperature (defined as the inflection point in the curve) occurs at around 570 K. The present cal- culations represent the first quantitatively realistic prediction of the order– disorder transformation of a surface reconstruction. The disorder to the surface was characterized in this calculations by the second moment of the structure factor (i.e. the spot width) in the vicinity of the wave vector for the (12) symmetry. The temperature dependence is very similar to that for the experimental LEED spot width, which is fairly constant below Tc and then increases rapidly with increasing temperature [18, 19]. The current theory showed, additionally, that the spot is not circular above Tc, but instead that the disorder perpendicular to the rows is stronger than that parallel. Daw and Foiles on the base of analysis the structure factor for (12) structure as a function of temperature concluded that below the critical temperature the rows for (12) structure are generally long and co- ordinated, with some defects present. Well above the Tc short-range or- der persists in the form of short chains of atoms along [ ]110 direction Figure 16. Snapshot of a randomly chosen structure after Monte Carlo simulation of the Au(110) surface [33]. 106 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov with little co-ordination among the rows. This was confirmed by snap- shots of individual configurations generated by the Monte Carlo simula- tions, as exemplified in Fig. 16 [33]. The snapshot of Fig. 16 shows that the atoms in the EAM–Monte Carlo simulations do indeed move off of ideal lattice sites. EAM–Monte Carlo simulations illustrates the role of re- laxation and vibrations since these effects are not included in the lattice- gas simulation within the framework of the Ising 2D-model, as examined in Ref. [18]. Bak [20] predicted an Ising model for deconstruction transition, how- ever he did not consider another possibility, namely, the deconstruction might be a roughening transition. Indeed, Wolf et al. [17] observed roughness on their Au(110) samples and they have found 0.23    0.37, arranger which is clearly above the Ising 2D-exponent, e.g.   1/8. How- ever they found roughness at all temperatures and this can only be an effect of metastability. Villain and Vilfan [34] first have presented the two-step disordering theory of the Au(110)–(12) reconstructed surface, including a two- dimensional Ising transition followed by surface roughening transition. From calculations of the free energy in terms of four energy parameters of meandering antiphase defects and surface steps (Fig. 17) Villain and Vilfan confirms the experimental observation that the deconstruction or- der–disorder transition of the (110) surface of Au and Pt belonging to the Ising universality class. Further, they predicted that at TR, approximately 100 K above Tc, deconstructed surface should undergo a Kosterlitz– Thoules roughening transition. The two-step disordering theory by Villain and Vilfan is consistent with most recently X-ray scattering study [27]. The more comprehensive data one can find in the recent review [2, 7, 8] Figure 17. Excitations on the (12)(110) surface of Au: a, b—Ising-like excita- tions; c–f—roughening transition [34]. Surface Reconstruction: Noble Metals 107 or in the appropriate references in the present paper [35–44]. The studies of the layer reconstruction in the clean metal surfaces has led to insights which motivated to investigate surface behaviour of single crystal metal alloys [2, 45, 46]. ACKNOWLEDGEMENTS This work was supported by INTAS-99 under Project #01882. REFERENCES 1. М. А. Васильев, Структура и динамика поверхности переходных металлов (Киев: Наукова думка: 1988). 2. M. A. Vasylyev, J. Phys. D: Appl. Phys., 30: No. 22: 3037 (1997). 3. F. Nieto, C. Uebing, and V. Pereyra, Surface Sci., 416: No. 1–2: 152 (1998). 4. N. Spiridis and J. Korecki, Appl. Surface Sci., 141: No. 3–4: 313 (1999). 5. D. Alfu and S. Gironcoli, Surface Sci., 437: No. 1–2: 18 (1999). 6. M. Lynch and P. Hu, Surface Sci., 458: No. 1–3: 1 (2000). 7. B. E. Nieuwenhuys, Surface Rev. and Lett., 3: No. 5–6: 1869 (1996). 8. B. E. Nieuwenhuys, Advances in Catalysis, 44: 259 (2000). 9. C. Becker, T. Pelster, M. Taneumura et al., Surface Sci., 427–428: No. 1–3: 403 (1999). 10. D. K. Captain and M. D. Amiridis, J. Catalysis, 184: 377 (1999). 11. M. A. Van Hove, R. J. Koester, P. S. Stair et al., Surface Sci., 103: No. 1: 189 (1981). 12. K. Heinz, E. Lang, K. Strayss et al., Appl. Surface Sci., 11–12: 611 (1982). 13. K. Heinz, E. Lang, K. Strayss et al., Surface Sci., 120: 401 (1982). 14. K. Heinz, G. Schmidi, L. Hammer et al., Phys. Rev. B, 32: No. 10: 6214 (1982). 15. D. Gibbs, B. M. Ocko, D. M. Zehner et al., Phys. Rev. B, 42: No. 12: 7330 (1990). 16. L. D. Roelofs, S. M. Foiles, M. S. Daw et al., Surface Sci., 234: 63 (1990). 17. D. Wolf and H. Jagondzinski, Surface Sci., 71: No. 1: 178 (1978). 18. J. C. Campuzano, M. S. Foster, G. Jennings et al., Phys. Rev. Lett., 54: No. 25: 2684 (1985). 19. J. C. Campuzano, G. Jennings, R. F. Willis et al., Surface Sci., 162: 484 (1985). 20. P. Bak, Solid State Communs, 32: 581 (1979). 21. D. E. Clark, W. P Unertl, and P. H. Kleban, Phys. Rev. B, 34: No. 6: 4379 (1986). 22. A. U. MacRae, Phys. Rev. B, 36: 2341 (1987). 23. H. Derks, J. Möller, and W. Heiland, Surface Sci., 188: 685 (1987). 24. T. T. Tsong and Q. Gao, Surface Sci., 182: 2257 (1987). 25. K. Dückers and H. P. Bonzel, Europhys. Lett., 7: No. 4: 371 (1988). 26. E. Riet van de, H. Derks, and W. Heiland, Surface Sci., 234: 53 (1990). 27. D. T. Keane, P. A. Bancel, J. L. Jordan-Sweet et al., Surface Sci., 250: 8 (1991). 28. J. Villain, Surface Sci., 199: 165 (1988). 29. J.-K. Zuo, Y.-L. He, and G.-C. Wang, J. Vac. Sci. Technol., 8: No. 3: 2474 (1990). 30. S. M. Foils and M. S. Daw, J. Vac. Sci. Technol. A, 3: 1565 (1985). 31. M. S. Daw and S. M. Foils, J. Vac. Sci. Technol. A, 4: 1412 (1986). 32. S. M. Foils, Surface Sci., 191: L779 (1987). 33. M. S. Daw, Surface Sci., 166: L161 (1986). 34. J. Villain and I. Vilfan, Surface Sci., 199: 165 (1988). 108 M. A. Vasyl’yev, A. B. Bondarchuk, and V. A. Tinkov 35. T. Ali, A. V. Walker, B. Klötzer et al., Surface Sci., 414: No. 1–2: 304 (1998). 36. A. G. Makeev and B. E. Nieuwenhuys, Surface Sci., 418: No. 2: 432 (1998). 37. Q. Ge, D. A. King, N. Marzari et al., Surface Sci., 418: No. 3: 529 (1998). 38. M. A. Gonzbler, J. de la Figuera, O. R. de la Fuente et al., Surface Sci., 429: No. 1–3: L486 (1999). 39. V. Jahns, D. M. Zehner, G. M. Watson et al., Surface Sci., 430: No. 1–3: 55 (1999). 40. I. Vilfan, F. Lanzon, and E. Adam, Surface Sci., 440: No. 1–2: 279 (1999). 41. M. S. Zei and G. Ertl, Surface Sci., 442: No. 1: 19 (1999). 42. Meng-Sheng Liao, C. R. Cabrera, and Y. Ichikawa, Surface Sci., 445: No. 2–3: 267 (2000). 43. D. Passerone, F. Ercolessi, and E. Tosatti, Surface Sci., 454–456: No. 1–3: 634: (2000). 44. M. Salmeron and G. A. Somorjai, Surface Sci., 91: No. 2, 3: 373 (1980). 45. U. Bardi, Prog. Phys., 57: 939 (1994). 46. M. A. Vasylyev, A. G. Blashchuk, O. P. Hryshchenko, and N. S. Mashovets, Phys. Low-Dim. Struct., 11, 12: 115 (1999).

Surface Reconstruction: Noble Metals

Institution

Ähnliche Einträge