31.12.2013 Views

Jump processes in surface diffusion - Bilkent University - Faculty of ...

Jump processes in surface diffusion - Bilkent University - Faculty of ...

Jump processes in surface diffusion - Bilkent University - Faculty of ...

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Surface Science Reports 62 (2007) 39–61<br />

www.elsevier.com/locate/surfrep<br />

<strong>Jump</strong> <strong>processes</strong> <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong><br />

Grazyna Antczak a,b,∗ , Gert Ehrlich a<br />

a Department <strong>of</strong> Materials Science and Eng<strong>in</strong>eer<strong>in</strong>g, <strong>University</strong> <strong>of</strong> Ill<strong>in</strong>ois at Urbana-Champaign, Urbana, IL 61801, USA<br />

b Institute <strong>of</strong> Experimental Physics, <strong>University</strong> <strong>of</strong> Wroclaw, Wroclaw, Poland<br />

Received 2 December 2006; accepted 4 December 2006<br />

Abstract<br />

The traditional view <strong>of</strong> the <strong>surface</strong> <strong>diffusion</strong> <strong>of</strong> metal atoms on metal <strong>surface</strong>s was that atoms carry on a random walk between nearest-neighbor<br />

<strong>surface</strong> sites. Through field ion microscopic observations and molecular dynamics simulations this picture has been changed completely. Diffusion<br />

by an adatom exchang<strong>in</strong>g with an atom <strong>of</strong> the substrate has been identified on fcc(110), and subsequently also on fcc(100) planes. At elevated<br />

temperatures, multiple events have been found by simulations <strong>in</strong> which an atom enters the lattice, and a lattice atom at some distance from the<br />

entry po<strong>in</strong>t pops out. Much at the same time the contribution <strong>of</strong> long jumps, spann<strong>in</strong>g more than a nearest-neighbour distance, has been exam<strong>in</strong>ed;<br />

their rates have been measured, and such transitions have been found to contribute significantly, at least on tungsten <strong>surface</strong>s. As higher <strong>diffusion</strong><br />

temperatures become accessible, additional jump <strong>processes</strong> can be expected to be revealed.<br />

c○ 2007 Elsevier B.V. All rights reserved.<br />

Keywords: Atom jumps; Surface <strong>diffusion</strong>; Field ion microscopy; Molecular dynamics<br />

Contents<br />

1. Introduction............................................................................................................................................................................ 39<br />

1.1. Surface diffusivities....................................................................................................................................................... 40<br />

2. Atom exchange ....................................................................................................................................................................... 40<br />

2.1. On fcc(110) planes ........................................................................................................................................................ 40<br />

2.2. On (100) planes ............................................................................................................................................................ 46<br />

2.3. Via multiple events........................................................................................................................................................ 48<br />

3. Long and rebound jumps.......................................................................................................................................................... 50<br />

3.1. Theoretical work........................................................................................................................................................... 50<br />

3.2. Experimental studies ..................................................................................................................................................... 51<br />

4. Summary ............................................................................................................................................................................... 58<br />

Acknowledgements ................................................................................................................................................................. 59<br />

References ............................................................................................................................................................................. 59<br />

1. Introduction<br />

Surface <strong>diffusion</strong> plays a significant role <strong>in</strong> crystal and film<br />

growth, <strong>in</strong> evaporation and condensation, <strong>in</strong> <strong>surface</strong> chemical<br />

∗ Correspond<strong>in</strong>g author at: Department <strong>of</strong> Materials Science and Eng<strong>in</strong>eer<strong>in</strong>g,<br />

<strong>University</strong> <strong>of</strong> Ill<strong>in</strong>ois at Urbana-Champaign, Urbana, IL 61801, USA. Tel.:<br />

+1 2173336752.<br />

E-mail address: antczak@mrl.uiuc.edu (G. Antczak).<br />

reactions and catalysis, <strong>in</strong> s<strong>in</strong>ter<strong>in</strong>g as well as <strong>in</strong> other <strong>surface</strong><br />

<strong>processes</strong>. As such there has been considerable <strong>in</strong>terest <strong>in</strong> how<br />

<strong>diffusion</strong> occurs; dur<strong>in</strong>g the last few decades much novel and<br />

significant material has been discovered. Here we will briefly<br />

review what has been learned about how <strong>diffusion</strong> <strong>of</strong> s<strong>in</strong>gle<br />

metal atoms takes place on metal <strong>surface</strong>s, what sort <strong>of</strong> atomic<br />

jumps occur and how <strong>in</strong>formation about these <strong>processes</strong> has<br />

been obta<strong>in</strong>ed.<br />

0167-5729/$ - see front matter c○ 2007 Elsevier B.V. All rights reserved.<br />

doi:10.1016/j.surfrep.2006.12.001


40 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 1. Hard-sphere models <strong>of</strong> bcc (a) (110), (b) (211), and (c) (321) planes.<br />

1.1. Surface diffusivities<br />

Diffusion is characterized by the material parameter D, the<br />

diffusivity <strong>in</strong> the (one-dimensional) <strong>diffusion</strong> equation:<br />

J = −D ∂c<br />

∂x<br />

connect<strong>in</strong>g the adatom flux J to the concentration gradient<br />

∂c/∂x. The diffusivity is given by the usual Arrhenius relation<br />

(<br />

D = D o exp − E )<br />

D<br />

, (2)<br />

kT<br />

where the prefactor D o , usually considered a constant, and<br />

the activation energy E D are obta<strong>in</strong>ed from the temperature<br />

dependence. Langmuir [1], <strong>in</strong> 1933, already viewed the atoms<br />

<strong>in</strong> <strong>surface</strong> <strong>diffusion</strong> as hopp<strong>in</strong>g between elementary spaces at a<br />

separation a, and was able to show that for an <strong>in</strong>dividual adatom<br />

the diffusivity was given <strong>in</strong> terms <strong>of</strong> the lifetime τ at a particular<br />

site by<br />

D = a 2 /2τ. (3)<br />

In transition-state theory the one-dimensional diffusivity is<br />

written more elaborately as:<br />

( ) (<br />

D = υl 2 SD<br />

exp exp − E )<br />

D<br />

, (4)<br />

k kT<br />

where υ gives the effective vibrational frequency <strong>of</strong> the adatom,<br />

l its jump length, and S D as well as E D the entropy and<br />

activation energy for <strong>diffusion</strong>. We can, however, <strong>in</strong>troduce the<br />

abbreviation:<br />

( )<br />

SD<br />

υ o = υ exp , (5)<br />

k<br />

so that the diffusivity appears as <strong>in</strong> Eq. (2), but with<br />

D o = υ o l 2 . (6)<br />

On <strong>surface</strong>s it would be difficult to measure atom fluxes<br />

and gradients. The mass flux <strong>of</strong> material over a <strong>surface</strong> can<br />

<strong>of</strong> course be detected, but it is not clear how to disentangle<br />

<strong>in</strong>teractions between the atoms. The diffusivity is therefore<br />

obta<strong>in</strong>ed alternatively from the E<strong>in</strong>ste<strong>in</strong> relation, which ties the<br />

diffusivity to the mean-square displacement 〈x 2 〉,<br />

〈x 2 〉 = 2Dt, (7)<br />

(1)<br />

Fig. 2. Models <strong>of</strong> fcc (a) (111) and (b) (100) planes.<br />

where t is the time <strong>in</strong>terval <strong>of</strong> the measurements.<br />

The elementary formalism is now <strong>in</strong> hand, and us<strong>in</strong>g it much<br />

data has been gathered about <strong>diffusion</strong> k<strong>in</strong>etics [2]. But what<br />

are the jump <strong>processes</strong> participat<strong>in</strong>g <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>? If<br />

atoms always jump between nearest-neighbour sites, then s<strong>in</strong>ce<br />

a typical vibrational frequency at the <strong>surface</strong> is ∼10 12 s −1 , and<br />

the spac<strong>in</strong>g is ∼3 Å, the prefactor D o should, if we neglect<br />

entropy contributions, be <strong>of</strong> magnitude ∼10 −3 cm 2 /s. It turns<br />

out that this is close to the value <strong>of</strong> the prefactor for many<br />

<strong>diffusion</strong> systems [2], start<strong>in</strong>g with one <strong>of</strong> the first studied, W<br />

on W(110), for which a value <strong>of</strong> 2.6×10 −3 cm 2 /s was found [3,<br />

4]. This agreement suggested that this simple view <strong>of</strong> <strong>diffusion</strong><br />

<strong>processes</strong> had considerable merit. However, the story has turned<br />

out to be much more <strong>in</strong>terest<strong>in</strong>g than that.<br />

2. Atom exchange<br />

2.1. On fcc(110) planes<br />

Early studies <strong>of</strong> <strong>in</strong>dividual metal atoms diffus<strong>in</strong>g on metals<br />

were with tungsten atoms on (110), (211), and (321) planes<br />

<strong>of</strong> bcc tungsten, models <strong>of</strong> which are shown <strong>in</strong> Fig. 1. Also<br />

studied somewhat later were rhodium atoms on rhodium, an fcc<br />

metal [4]. Exam<strong>in</strong>ed were (111) and (100), as well as (110),<br />

(311) and (331) planes, shown <strong>in</strong> Figs. 2 and 3. What is clear<br />

is that the (321) and (211) planes on tungsten and (311), (331)<br />

and (110) planes on rhodium are channelled, and that an adatom<br />

mov<strong>in</strong>g along one <strong>of</strong> these channels might be expected to stay <strong>in</strong><br />

such a channel and execute strictly one-dimensional <strong>diffusion</strong>.<br />

That was <strong>in</strong> fact found to be the case as shown <strong>in</strong> Fig. 4<br />

for W(211), mak<strong>in</strong>g predictions <strong>of</strong> the direction <strong>in</strong> <strong>diffusion</strong><br />

seem<strong>in</strong>gly straightforward.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 41<br />

Fig. 3. Models <strong>of</strong> fcc (a) (110), (b) (311), and (c) (331) planes.<br />

Fig. 5. Mechanism for cross-channel <strong>diffusion</strong> <strong>of</strong> adatom on Pt(110) proposed<br />

by Bassett and Webber [5]. (a) Adatom <strong>in</strong> channel. (b) Vacancy forms <strong>in</strong><br />

channel wall. (c) Adatom has moved <strong>in</strong>to vacancy, lattice atom is now <strong>in</strong><br />

channel site.<br />

Fig. 4. Location <strong>of</strong> sites on W(211) at which Rh adatom has been detected after<br />

<strong>diffusion</strong> at 197 K. Motion is strictly one-dimensional. Shown at the bottom is<br />

a count <strong>of</strong> the number sighted at each site.<br />

This simple picture was suddenly destroyed, however, by the<br />

work <strong>of</strong> Bassett and Webber [5] <strong>in</strong> 1978, who studied another<br />

fcc metal, plat<strong>in</strong>um. On (331) and (311) planes <strong>diffusion</strong><br />

occurred, as expected, along the channels <strong>of</strong> the <strong>surface</strong>s. On<br />

the (110) plane, however, <strong>diffusion</strong> was two-dimensional! The<br />

activation energy for <strong>in</strong>-channel motion was 0.84 ± 0.1 eV, for<br />

cross-channel movement it was smaller, only 0.78 ± 0.1 eV.<br />

Similar results were found with the <strong>diffusion</strong> <strong>of</strong> iridium atoms<br />

on these <strong>surface</strong>s. What had been established here was that<br />

<strong>diffusion</strong> on channelled <strong>surface</strong>s could occur <strong>in</strong> two dimensions,<br />

depend<strong>in</strong>g on the particular system. Predict<strong>in</strong>g the direction <strong>in</strong><br />

which <strong>diffusion</strong> occurred was not just a matter <strong>of</strong> look<strong>in</strong>g at<br />

<strong>surface</strong> geometry.<br />

The question still rema<strong>in</strong>ed how cross-channel motion took<br />

place. Bassett and Webber [5] had two ideas. One was that<br />

large fluctuations occurred <strong>in</strong> the rows <strong>of</strong> atoms constitut<strong>in</strong>g<br />

the channel walls, creat<strong>in</strong>g holes through which adatoms could<br />

easily jump. A more likely possibility was that the movement<br />

<strong>of</strong> a channel atom created a hole that could be filled either<br />

by the adatom or a <strong>surface</strong> atom, as shown <strong>in</strong> Fig. 5. They<br />

concluded that “Further experimental and theoretical studies are<br />

required to clarify the nature <strong>of</strong> <strong>in</strong>ter-channel <strong>diffusion</strong>”. Even<br />

without an understand<strong>in</strong>g <strong>of</strong> the mechanism <strong>of</strong> cross-channel<br />

motion, the phenomenon <strong>of</strong> cross-channel <strong>diffusion</strong> had been<br />

established.<br />

What actually happened <strong>in</strong> cross-channel atom movement<br />

was soon uncovered <strong>in</strong> experiments by Wrigley and Ehrlich [6].<br />

They took advantage <strong>of</strong> the atom probe [7], to measure the<br />

chemical identity <strong>of</strong> atoms on Ir(110)-(1 × 1). In this work<br />

the system was calibrated by first deposit<strong>in</strong>g an iridium and<br />

separately a tungsten atom on the <strong>surface</strong> kept at ∼50 K. The<br />

mass <strong>of</strong> the <strong>in</strong>dividual atoms was then obta<strong>in</strong>ed by measur<strong>in</strong>g<br />

the time <strong>of</strong> flight to a detector ∼1 m away by field desorb<strong>in</strong>g<br />

the atoms; the results are shown <strong>in</strong> Fig. 6. With this <strong>in</strong>formation<br />

<strong>in</strong> hand, a tungsten atom was placed on the clean <strong>surface</strong>,<br />

which was heated until cross-channel <strong>diffusion</strong> took place. The<br />

atom observed by field ion microscopy after diffus<strong>in</strong>g <strong>in</strong>to<br />

the adjacent channel was then field evaporated, and as also<br />

shown <strong>in</strong> Fig. 6 proved to be iridium, an atom from the lattice.<br />

Thereafter the top layer <strong>of</strong> the <strong>surface</strong> was field evaporated and<br />

analyzed, as <strong>in</strong> Fig. 7, and usually a tungsten atom was detected<br />

<strong>in</strong> the <strong>surface</strong> layer, as expected if atom exchange had occurred.<br />

This was not the case when the <strong>surface</strong> was probed after crosschannel<br />

motion had not been seen. These experiments were the<br />

first direct pro<strong>of</strong> <strong>of</strong> an <strong>in</strong>terchange between a substrate and an<br />

adsorbed atom dur<strong>in</strong>g cross-channel <strong>diffusion</strong>.<br />

A number <strong>of</strong> theoretical <strong>in</strong>vestigations were stimulated<br />

by the results on Pt(110). Halicioglu [8,9] carried out<br />

calculations for <strong>diffusion</strong> on Pt(110) with Lennard-Jones twobody<br />

potentials and found that <strong>in</strong> cross-channel <strong>diffusion</strong> an<br />

adatom <strong>in</strong>teracts with an atom <strong>in</strong> the channel wall to form<br />

a dumbbell, as <strong>in</strong> Fig. 8(b). This could decompose by the<br />

adatom <strong>in</strong>corporat<strong>in</strong>g <strong>in</strong>to the <strong>surface</strong> and the <strong>surface</strong> atom<br />

go<strong>in</strong>g <strong>in</strong>to the adjacent channel, or else by return<strong>in</strong>g to the<br />

orig<strong>in</strong>al channel, as <strong>in</strong>dicated <strong>in</strong> Fig. 8(c).<br />

Just slightly later, DeLorenzi et al. [10,11] did molecular<br />

dynamics calculations, aga<strong>in</strong> us<strong>in</strong>g Lennard-Jones potentials,<br />

on fcc <strong>surface</strong>s. Most <strong>in</strong>terest<strong>in</strong>g were the results for (110)<br />

planes, shown <strong>in</strong> Fig. 9, on which both <strong>in</strong>-channel and


42 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 6. Distribution <strong>of</strong> atomic weights <strong>of</strong> <strong>in</strong>dividual atoms field evaporated<br />

from Ir(110) <strong>surface</strong> [6]. Top: after deposition <strong>of</strong> Ir atom. Bottom: after<br />

deposition <strong>of</strong> W atom. Center: After deposition <strong>of</strong> W atom, followed by heat<strong>in</strong>g<br />

and observed cross-channel <strong>diffusion</strong> event; material desorbed is iridium.<br />

Fig. 7. Atomic weight distribution <strong>of</strong> material field evaporated from first<br />

Ir(110) layer [6]. Left: after cross-channel <strong>diffusion</strong> has occurred, a W atom<br />

is detected <strong>in</strong> the first substrate layer. Right: when no cross-channel motion was<br />

detected, iridium is desorbed.<br />

cross-channel motion was discovered, the latter with a<br />

lower activation energy. In cross-channel <strong>diffusion</strong> they aga<strong>in</strong><br />

envisioned a dumbbell <strong>in</strong>termediate, as shown <strong>in</strong> Fig. 8(b),<br />

<strong>in</strong> which a pair <strong>of</strong> atoms sat across the ridgel<strong>in</strong>e. When<br />

this decomposes, the adatom can move back to its orig<strong>in</strong>al<br />

channel, or else it can move <strong>in</strong>to the channel wall, plac<strong>in</strong>g<br />

a lattice atom <strong>in</strong> the adjacent channel. Additional simulations<br />

<strong>of</strong> <strong>diffusion</strong> across channels on a Lennard-Jones fcc crystal<br />

were done by Mruzik and Pound [12]. On the (110) plane,<br />

cross-channel motion aga<strong>in</strong> occurred by exchange with a lattice<br />

atom, but movement on the (113) plane was along the channels.<br />

Gar<strong>of</strong>al<strong>in</strong>i and Halicioglu [13] did similar estimates on the<br />

Pt(110) plane at both low and high temperatures. At lower<br />

temperatures both iridium and gold atoms diffused along the<br />

channels, but at higher temperatures exchange took place for<br />

iridium, and a plat<strong>in</strong>um lattice atom appeared <strong>in</strong> the next<br />

channel. Gold at this temperature also formed a dumbbell<br />

with an atom from the channel wall, but returned to cont<strong>in</strong>ue<br />

<strong>diffusion</strong> <strong>in</strong> its orig<strong>in</strong>al channel, so <strong>diffusion</strong> was really onedimensional.<br />

More experimental work was reported <strong>in</strong> 1982 by<br />

Wrigley [14], who exam<strong>in</strong>ed the temperature dependence <strong>of</strong><br />

cross-channel motion, as shown <strong>in</strong> Fig. 10, with an activation<br />

energy <strong>of</strong> 0.74 ± 0.09 eV and a prefactor 1.4(×16 ±1 ) ×<br />

10 −6 cm 2 /s. Data were obta<strong>in</strong>ed at only five temperatures,<br />

and keep<strong>in</strong>g <strong>in</strong> m<strong>in</strong>d the low prefactor, must be viewed with<br />

some doubt. What was not clear at this po<strong>in</strong>t was the reason<br />

for the magnitude <strong>of</strong> the prefactor. Was it associated with the<br />

complicated mechanism, or were the measurements not detailed<br />

enough?<br />

At essentially the same time, Tung and Graham [15] looked<br />

at self-<strong>diffusion</strong> on various <strong>surface</strong>s <strong>of</strong> nickel. The behaviour<br />

<strong>of</strong> the (110) plane varied depend<strong>in</strong>g on whether it had been<br />

cleaned thermally, or had been subjected to field evaporation<br />

<strong>in</strong> hydrogen gas. Diffusion measurements were made after the<br />

hydrogen treatment and some field evaporation, as well as after<br />

thermal clean<strong>in</strong>g only, and gave the results shown <strong>in</strong> Fig. 11. For<br />

both treatments self-<strong>diffusion</strong> was two-dimensional on Ni(110);<br />

however, <strong>diffusion</strong> after thermal treatment led to very low<br />

<strong>diffusion</strong> prefactors ∼10 −7 –10 −9 cm 2 /s, with an activation<br />

energy <strong>of</strong> 0.23 ± 0.04 eV for <strong>in</strong>-channel movement and 0.32 ±<br />

0.05 eV across the channels. After hydrogen treatment, the<br />

characteristics for <strong>in</strong>-channel motion were a <strong>diffusion</strong> barrier<br />

<strong>of</strong> 0.30 ± 0.06 eV and a prefactor <strong>of</strong> 10 1 cm 2 /s; for crosschannel<br />

<strong>diffusion</strong> the barrier was lower, 0.25 ± 0.06 eV, with<br />

a prefactor <strong>of</strong> 10 −1 cm 2 /s. In the temperature range 80–90<br />

K, cross-channel motion predom<strong>in</strong>ated. On Ni(331) and (113),<br />

however, movement was always one-dimensional. Tung [16]<br />

also briefly studied self-<strong>diffusion</strong> on Al(110), determ<strong>in</strong><strong>in</strong>g<br />

the temperature for the beg<strong>in</strong>n<strong>in</strong>g <strong>of</strong> atom movement. Twodimensional<br />

<strong>diffusion</strong> was observed on the (110) plane, at a<br />

temperature <strong>of</strong> 154 K for both directions, lead<strong>in</strong>g to an estimate<br />

<strong>of</strong> 0.43 eV for the activation energy. It should be noted that<br />

the <strong>diffusion</strong> characteristics, both for nickel and alum<strong>in</strong>um<br />

(110) are uncerta<strong>in</strong>, and the data have been reanalyzed [17].<br />

Nevertheless, these observations firmly established crosschannel<br />

motion on these <strong>surface</strong>s.<br />

Another five years later, <strong>in</strong> 1986, Kellogg [18] measured<br />

the rates <strong>of</strong> <strong>in</strong>- and cross-channel self-<strong>diffusion</strong> on Pt(110)<br />

over a range <strong>of</strong> temperatures. From an Arrhenius plot, shown<br />

<strong>in</strong> Fig. 12, he found an activation <strong>of</strong> 0.72 ± 0.07 eV with a<br />

prefactor <strong>of</strong> 6 × 10 −4 cm 2 /s for <strong>diffusion</strong> along the channels,


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 43<br />

Fig. 8. Schematic <strong>of</strong> atom exchange process <strong>in</strong> self-<strong>diffusion</strong> on fcc(110) <strong>surface</strong>: (a) Atom <strong>in</strong> the equilibrium position. (b) At the saddle po<strong>in</strong>t, atom pair sits as a<br />

dumbbell across [1¯10] row <strong>of</strong> substrate atoms. (c 1 )–(c 4 ) After <strong>diffusion</strong>, atoms distributed uniformly over allowed sites.<br />

and 0.69 ± 0.07 eV with a prefactor <strong>of</strong> 3 × 10 −4 cm 2 /s for<br />

motion across the channels. It appeared that the prefactors for<br />

<strong>in</strong>- and cross-channel motion were similar and normal.<br />

Chen and Tsong [19] aga<strong>in</strong> looked at self-<strong>diffusion</strong> on<br />

Ir(110) <strong>in</strong> 1991, and as shown <strong>in</strong> Fig. 13, observed crosschannel<br />

<strong>diffusion</strong> with an activation energy <strong>of</strong> 0.71 ± 0.02 eV<br />

and a prefactor 6 × 10 −2.0±1.8 cm 2 /s; the values for <strong>in</strong>channel<br />

motion were 0.80 ± 0.04 eV and a prefactor <strong>of</strong><br />

4 × 10 −3.0±0.8 cm 2 /s, so cross-channel jumps occurred more<br />

rapidly. Earlier work by Wrigley [14] was reasonably close <strong>in</strong><br />

the <strong>diffusion</strong> barrier, but the present prefactor value is clearly<br />

preferable. Above and beyond the usual measurements, Chen<br />

and Tsong also looked at the distribution <strong>of</strong> displacements to<br />

get additional <strong>in</strong>formation about the <strong>diffusion</strong> process. These<br />

observations are shown <strong>in</strong> Fig. 14, with x <strong>in</strong>dicat<strong>in</strong>g <strong>in</strong>-channel<br />

<strong>diffusion</strong> and y cross-channel jump<strong>in</strong>g. What is especially<br />

<strong>in</strong>terest<strong>in</strong>g here is the fact that 80% <strong>of</strong> the jumps were <strong>in</strong> the<br />

〈112〉 direction, shown <strong>in</strong> Fig. 8, with the rema<strong>in</strong>der along<br />

〈100〉. Accord<strong>in</strong>g to all the simulations <strong>of</strong> such self-<strong>diffusion</strong><br />

done at that time, a cross-channel transition should have the<br />

same probability <strong>in</strong> four directions, whereas <strong>in</strong> the experiments,<br />

transitions <strong>in</strong> the direction <strong>of</strong> the mov<strong>in</strong>g adatom were favoured.<br />

These results were not accidental, however.<br />

Kellogg [20] <strong>in</strong> the same year, looked at the <strong>diffusion</strong> <strong>of</strong><br />

plat<strong>in</strong>um on the (110) plane <strong>of</strong> nickel. For these measurements<br />

nickel was not cleaned <strong>in</strong> hydrogen; <strong>in</strong>stead a comb<strong>in</strong>ation<br />

<strong>of</strong> thermal clean<strong>in</strong>g, sputter<strong>in</strong>g and evaporation was found to<br />

be essential for a good image <strong>of</strong> the (110) plane. Although


44 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 11. Arrhenius plot for <strong>in</strong>- and cross-channel <strong>diffusion</strong> <strong>of</strong> Ni on Ni(110)<br />

plane [15]. Left: Substrate after thermal treatment. Right: Nickel substrate had<br />

been hydrogen fired and field evaporated before experiments.<br />

Fig. 9. Adatom trajectories <strong>in</strong> <strong>diffusion</strong> on fcc(110) plane at 0.4T m [11].<br />

Transitions across the [1¯10] channels are apparent.<br />

Fig. 12. Self-<strong>diffusion</strong> <strong>of</strong> Pt on Pt(110) <strong>surface</strong> [18]. In-channel <strong>diffusion</strong>:<br />

E D = 0.72 ± 0.07 eV, D o = 6 × 10 −4 cm 2 /s. Cross-channel <strong>diffusion</strong>:<br />

E D = 0.69 ± 0.07 eV, D o = 3 × 10 −4 cm 2 /s.<br />

Fig. 10. Early Arrhenius plot for the cross-channel <strong>diffusion</strong> <strong>of</strong> iridium atoms<br />

on Ir(110) [14].<br />

Kellogg did not rely on an atom probe, he was able to<br />

discrim<strong>in</strong>ate between Ni and Pt atoms through differences <strong>in</strong> the<br />

voltages necessary for field evaporation—the plat<strong>in</strong>um atoms<br />

were removed at significantly higher voltages; they also had<br />

a much larger image spot. When a Pt atom was deposited<br />

on the cold <strong>surface</strong>, the image looked as <strong>in</strong> Fig. 15(a). After<br />

heat<strong>in</strong>g to 112 K, the image <strong>in</strong> Fig. 15(b) yielded a much<br />

smaller spot size, <strong>in</strong>dicative <strong>of</strong> Ni. Upon remov<strong>in</strong>g the adatom,<br />

<strong>in</strong> Fig. 15(c), and field evaporat<strong>in</strong>g the topmost layer, as is<br />

shown <strong>in</strong> Fig. 15(d), a plat<strong>in</strong>um atom was retrieved from<br />

the <strong>surface</strong> layer. Clearly exchange had occurred between a<br />

Fig. 13. Self-<strong>diffusion</strong> <strong>of</strong> Ir on Ir(110) plane [19]. Both <strong>in</strong>-channel and crosschannel<br />

motion is observed.<br />

plat<strong>in</strong>um adatom and a nickel atom from the substrate, and the<br />

atom enter<strong>in</strong>g the lattice was plat<strong>in</strong>um not nickel. Assum<strong>in</strong>g<br />

a prefactor <strong>of</strong> 1 × 10 −3 cm 2 /s, a barrier <strong>of</strong> 0.28 eV was<br />

determ<strong>in</strong>ed for the displacement <strong>of</strong> plat<strong>in</strong>um. The distribution<br />

<strong>of</strong> displacements was also observed <strong>in</strong> these experiments and it<br />

was found that <strong>in</strong> 62.5% the replacement atom appeared <strong>in</strong> the<br />

〈112〉 direction, never along 〈001〉; the rema<strong>in</strong><strong>in</strong>g experiments


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 45<br />

Fig. 14. Distribution <strong>of</strong> displacements <strong>in</strong> <strong>diffusion</strong> <strong>of</strong> Ir on Ir(110) at different<br />

temperatures [19]. Transitions <strong>in</strong> direction <strong>of</strong> mov<strong>in</strong>g atom are favoured,<br />

contrary to theoretical predictions.<br />

Fig. 16. Exchange between Re adatom and Ir(110) substrate illustrated <strong>in</strong> field<br />

ion images [21]. In (b), Re atom has been deposited on the clean Ir(110)<br />

<strong>surface</strong> <strong>in</strong> (a), giv<strong>in</strong>g a round image spot, a schematic <strong>of</strong> which is <strong>in</strong> (e). After<br />

one m<strong>in</strong>ute at 255 K, oblong atom, Ir, is found <strong>in</strong> (c), also illustrated <strong>in</strong> (f).<br />

Substituted Re atom is revealed on partial field evaporation, <strong>in</strong> (d).<br />

Fig. 15. Neon field ion images show<strong>in</strong>g exchange <strong>of</strong> Pt adatom with Ni from<br />

Ni(110) <strong>surface</strong> [20]. (a) S<strong>in</strong>gle Pt adatom on Ni(110) plane. (b) After one<br />

m<strong>in</strong>ute at 112 K, adatom has changed to Ni, judged from lower desorption<br />

field. (c) Adatom has been removed by field evaporation. (d) After removal <strong>of</strong><br />

one layer <strong>of</strong> nickel, Pt adatom is aga<strong>in</strong> apparent.<br />

ended with an atom <strong>in</strong> the orig<strong>in</strong>al row. Kellogg speculated that<br />

this was due to the atom from the channel wall ma<strong>in</strong>ta<strong>in</strong><strong>in</strong>g its<br />

cohesion with the adatom, but this conclusion was not reached<br />

<strong>in</strong> previous molecular dynamics simulations. The mechanism<br />

<strong>of</strong> cross-channel movement is still wreathed <strong>in</strong> uncerta<strong>in</strong>ty.<br />

A year later, Chen et al. [21] carried out field ion microscopic<br />

studies <strong>of</strong> rhenium atom <strong>diffusion</strong> on the Ir(110) <strong>surface</strong>. They<br />

were able to dist<strong>in</strong>guish between Re and Ir atoms, as rhenium<br />

gave a circular image spot and iridium yielded an elongated<br />

image. A rhenium atom was deposited on the <strong>surface</strong> and then<br />

heated to ∼256 K. This moved the atom spot one space over<br />

and elongated the image, suggest<strong>in</strong>g that an atom replacement<br />

had occurred. The top layer <strong>of</strong> the substrate was then field<br />

evaporated, as shown <strong>in</strong> Fig. 16, reveal<strong>in</strong>g a bright rhenium<br />

atom, and demonstrat<strong>in</strong>g that exchange with a lattice atom had<br />

<strong>in</strong>deed taken place. F<strong>in</strong>ally, it should be noted that <strong>in</strong> 2002,<br />

Pedemonte et al. [22] were able to detect some cross-channel<br />

movement <strong>in</strong> helium scatter<strong>in</strong>g experiments on Ag(110) at<br />

temperatures above 750 K, but did not derive energetics for<br />

such movement.<br />

At the beg<strong>in</strong>n<strong>in</strong>g <strong>of</strong> the n<strong>in</strong>eties there started quite a<br />

large number <strong>of</strong> efforts to calculate diffusivities <strong>of</strong> metal<br />

atoms on metals. Such estimates, usually us<strong>in</strong>g semi-empirical<br />

<strong>in</strong>teractions, were done for the (110) planes <strong>of</strong> Al [17,23–<br />

25], Ni [17,26–28], Cu [17,27–33], Pd [17,27,28,34], Ag [17,<br />

28,30,32,34,35], Ir [28,36,37], Pt [17,27,28,38,39], Au [17,27,<br />

28,32], and Pb [40]. In all <strong>of</strong> these the barrier to <strong>diffusion</strong><br />

along the channels was lower than the barrier for cross-channel<br />

transitions, despite the fact that for Al, Ni, Ir and Pt the opposite<br />

had been demonstrated <strong>in</strong> experiments. In short, <strong>diffusion</strong> by<br />

atom exchange on fcc(110) is well established, if not yet well<br />

understood. What is still not clear is why a particular system<br />

undergoes exchange while another one does not, and why a<br />

particular direction is favoured <strong>in</strong> <strong>diffusion</strong>.


46 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 19. Map <strong>of</strong> sites on Pt(100) at which Pt adatom was found after <strong>diffusion</strong> at<br />

175 K [46]. A c(2 × 2) net is formed, <strong>in</strong>dicative <strong>of</strong> jumps by atom replacement.<br />

Fig. 17. Atom migration on bcc (100) <strong>surface</strong>, shown <strong>in</strong> molecular dynamics<br />

simulations modelled by Price potential [42,43]. Straight arrows j <strong>in</strong>dicate<br />

nearest-neighbour jumps, curved arrows e show adatom exchange with<br />

substrate [41].<br />

2.2. On (100) planes<br />

While all this work study<strong>in</strong>g atom exchange on fcc(110)<br />

planes was go<strong>in</strong>g on, DeLorenzi and Jacucci [41] had been<br />

cont<strong>in</strong>u<strong>in</strong>g their efforts to exam<strong>in</strong>e atomic jumps <strong>in</strong> <strong>surface</strong><br />

<strong>diffusion</strong>. Their attention turned to bcc materials, which meant<br />

they had to resort to a different type <strong>of</strong> <strong>in</strong>teraction for<br />

simulations. They relied on a metallic potential devised by<br />

Price [42,43] to describe sodium, and used this <strong>in</strong> exam<strong>in</strong><strong>in</strong>g<br />

<strong>diffusion</strong> on various planes <strong>of</strong> a bcc lattice. Of specific <strong>in</strong>terest<br />

here is what was observed on the (100) plane. In the usual<br />

course <strong>of</strong> events, we expect <strong>diffusion</strong> to take place by an atom<br />

jump<strong>in</strong>g from its normal b<strong>in</strong>d<strong>in</strong>g site, at the centre <strong>of</strong> four<br />

<strong>surface</strong> atoms, to a neighbour<strong>in</strong>g b<strong>in</strong>d<strong>in</strong>g site <strong>in</strong> a straight l<strong>in</strong>e<br />

at right angles to the border <strong>of</strong> the unit cell. What frequently<br />

happened <strong>in</strong> the simulations, however, was different and is<br />

shown <strong>in</strong> Fig. 17, where short straight arrows <strong>in</strong>dicate normal<br />

jumps, between nearest neighbours, and curved ones reveal an<br />

atom-exchange process.<br />

In this exchange an adatom moves <strong>in</strong>to the outer lattice<br />

layer, dislodg<strong>in</strong>g a lattice atom which ends up on the <strong>surface</strong><br />

<strong>in</strong> a position diagonal to the start<strong>in</strong>g po<strong>in</strong>t, as suggested <strong>in</strong><br />

the schematic <strong>in</strong> Fig. 18. In their own words, DeLorenzi<br />

and Jacucci found that “In addition to conventional nearest<br />

neighbour jumps between <strong>surface</strong> sites, the adatom undergoes<br />

migration events rem<strong>in</strong>iscent <strong>of</strong> exchange <strong>processes</strong> <strong>of</strong> the<br />

<strong>in</strong>tersticialcy <strong>in</strong> the bulk. In these events, atom A belong<strong>in</strong>g to<br />

the <strong>surface</strong> layer is replaced by atom B orig<strong>in</strong>ally constitut<strong>in</strong>g<br />

the adatom. As a result, atom A ends up as an adatom located<br />

at a site displaced from the one orig<strong>in</strong>ally occupied by the<br />

po<strong>in</strong>t defect”. “This is the first observation, <strong>in</strong> simulations or<br />

real experiments, <strong>of</strong> the occurrence <strong>of</strong> exchange events between<br />

adatom and a substitutional atom as a quantitatively important<br />

process contribut<strong>in</strong>g to atomic <strong>diffusion</strong> on isotropic crystal<br />

<strong>surface</strong>s.” A new exchange event had been discovered; one<br />

question rema<strong>in</strong><strong>in</strong>g was the generality <strong>of</strong> this process. Would<br />

it occur <strong>in</strong> practical systems?<br />

Regrettably, noth<strong>in</strong>g was done to test the work <strong>of</strong> DeLorenzi<br />

and Jacucci, until five years later two studies appeared: Kellogg<br />

and Feibelman [44] studied Pt(100) and Chen and Tsong [45]<br />

simultaneously looked at Ir(100). They both exam<strong>in</strong>ed the<br />

displacements carried out on the <strong>surface</strong>, and found that atoms<br />

moved diagonally and not just to nearest neighbors, creat<strong>in</strong>g<br />

a c(2 × 2) map <strong>of</strong> b<strong>in</strong>d<strong>in</strong>g sites, as shown <strong>in</strong> Fig. 19. From<br />

the mean-square displacement at 175 K, the activation energy<br />

for <strong>diffusion</strong> on Pt(100) was estimated as 0.47 eV, assum<strong>in</strong>g<br />

the usual prefactor <strong>of</strong> 10 −3 cm 2 /s. Chen and Tsong did better<br />

<strong>in</strong> study<strong>in</strong>g iridium. From an Arrhenius plot <strong>of</strong> the meansquare<br />

displacement per unit time, shown <strong>in</strong> Fig. 20, they found<br />

a barrier <strong>of</strong> 0.84 ± 0.05 eV and a diffusivity prefactor <strong>of</strong><br />

6.3(×11 ±1 ) × 10 −2 cm 2 /s. Just like Kellogg and Feibelman,<br />

they observed diagonal atom movement for iridium. On these<br />

two <strong>surface</strong>s, <strong>diffusion</strong> occurred by atom exchange, as had been<br />

found earlier by DeLorenzi and Jacucci [41].<br />

Fig. 18. Schematic show<strong>in</strong>g exchange between adatom and substrate atom on (100) plane: (a) Adatom at equilibrium site. (b) Adatom and lattice atom <strong>in</strong> transition<br />

state. (c) Adatom <strong>in</strong>corporated <strong>in</strong>to lattice; atom from lattice has turned <strong>in</strong>to adatom.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 47<br />

Fig. 20. Arrhenius plot for <strong>diffusion</strong> <strong>of</strong> Ir adatom on Ir(100) observed <strong>in</strong> field<br />

ion microscope [45].<br />

Fig. 23. Interaction <strong>of</strong> Re adatom with Ir(100) <strong>surface</strong> [48]. Re atom is placed<br />

on (100) plane <strong>in</strong> (b) and is then heated to ∼230 K, yield<strong>in</strong>g an oblong image<br />

spot <strong>in</strong> (c), which rotates on heat<strong>in</strong>g to ∼262 K. This sequence is shown<br />

schematically <strong>in</strong> (e)–(g).<br />

Fig. 21. Dependence <strong>of</strong> Pt atom diffusivity on Pt(100) upon 1/T [46].<br />

Fig. 22. B<strong>in</strong>d<strong>in</strong>g sites for Pd on Pt(100) after <strong>diffusion</strong> at 265 K, <strong>in</strong>dicat<strong>in</strong>g<br />

ord<strong>in</strong>ary hopp<strong>in</strong>g [47].<br />

Further studies <strong>of</strong> self-<strong>diffusion</strong> on Pt(100) were carried out<br />

<strong>in</strong> 1991 by Kellogg [46], who aga<strong>in</strong> found a diagonal map<br />

<strong>of</strong> displacements for plat<strong>in</strong>um atoms, <strong>in</strong>dicat<strong>in</strong>g an exchange<br />

process. The <strong>diffusion</strong> barrier obta<strong>in</strong>ed from the Arrhenius plot<br />

<strong>in</strong> Fig. 21 was 0.47 ± 0.01 eV, with a prefactor 1.3(×10 ±1 ) ×<br />

10 −3 cm 2 /s. Subsequent work [47] showed that palladium<br />

atoms carried out the usual atomic jumps, as <strong>in</strong>dicated <strong>in</strong><br />

Fig. 22, while nickel and plat<strong>in</strong>um underwent atomic exchange<br />

reactions <strong>in</strong> <strong>diffusion</strong>.<br />

In the next year, Tsong and Chen [48] put a rhenium atom<br />

on Ir(100). When heated to a temperature <strong>of</strong> ∼230 K, the<br />

Re displaced an Ir atom from the <strong>surface</strong> layer to form a<br />

dimer above the vacancy, as <strong>in</strong> Fig. 23. Above 280 K the<br />

dimer decomposed and the rhenium entered the lattice, with an<br />

iridium atom left to cont<strong>in</strong>ue <strong>surface</strong> <strong>diffusion</strong>, another example<br />

<strong>of</strong> atom exchange.<br />

Kellogg [49] also studied the behavior <strong>of</strong> plat<strong>in</strong>um atoms on<br />

Ni(100). He was able to dist<strong>in</strong>guish between the two <strong>in</strong> terms<br />

<strong>of</strong> the higher voltage required to field evaporate plat<strong>in</strong>um than<br />

nickel atoms. After deposit<strong>in</strong>g plat<strong>in</strong>um on the <strong>surface</strong> at 77<br />

K, and then heat<strong>in</strong>g to 250 K, the plat<strong>in</strong>um disappeared. On<br />

field evaporat<strong>in</strong>g the <strong>surface</strong> layer plat<strong>in</strong>um appeared aga<strong>in</strong>, as<br />

<strong>in</strong> Fig. 24, reveal<strong>in</strong>g that an exchange process had taken place.<br />

Additional confirmation for iridium <strong>diffusion</strong> on Ir(100) by<br />

atom exchange was provided by the work <strong>of</strong> Friedl et al. [50]<br />

<strong>in</strong> 1992. They plotted the sites visited <strong>in</strong> <strong>diffusion</strong> and found<br />

a c(2 × 2) net, as expected if exchange took place between<br />

an adatom and a lattice atom. However, they proposed an<br />

alternative explanation: the <strong>surface</strong> could reconstruct and create<br />

a c(2 × 2) sub-lattice for diffus<strong>in</strong>g atoms. This explanation<br />

did not survive the test <strong>of</strong> time, however, s<strong>in</strong>ce reconstruction<br />

should occur <strong>in</strong>dependent <strong>of</strong> the type <strong>of</strong> diffus<strong>in</strong>g atom. Some<br />

years later, Fu and Tsong [51] aga<strong>in</strong> looked at self-<strong>diffusion</strong> on<br />

Ir(100) and aga<strong>in</strong> observed a c(2 × 2) net.<br />

In the meantime, quite a number <strong>of</strong> calculations were<br />

made for <strong>surface</strong> <strong>diffusion</strong> on a variety <strong>of</strong> fcc(100) planes,<br />

with a mixture <strong>of</strong> results. More then ten estimates have been<br />

carried out for Al(100). Most <strong>of</strong> them considered exchange


48 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 24. Replacement <strong>of</strong> Pt adatom on Ni(100) by Ni atom [49]. (a) Pt atom<br />

deposited on Ni(100) at 77 K. (b) After heat<strong>in</strong>g to 250 K for one m<strong>in</strong>ute, Pt<br />

adatom disappears. (c) After partial field evaporation. (d) On complete field<br />

evaporation <strong>of</strong> one layer <strong>of</strong> nickel, Pt atom reappears.<br />

as a possible mechanism [17,23,25,52–59], and roughly half<br />

favored atom exchange dur<strong>in</strong>g <strong>diffusion</strong>. For Ni(100) [17,59–<br />

71], out <strong>of</strong> a total <strong>of</strong> n<strong>in</strong>eteen estimates, only eight considered<br />

exchange as an option, and fewer than half <strong>of</strong> these <strong>in</strong>dicated<br />

an exchange <strong>of</strong> atoms as primary <strong>in</strong> the <strong>diffusion</strong> process.<br />

Here it is important to note the observation <strong>of</strong> Perk<strong>in</strong>s and<br />

DePristo [65] as well as <strong>of</strong> Chang and Wei [60] that the<br />

activation energy for exchange depends strongly on the size <strong>of</strong><br />

the cell used <strong>in</strong> the calculations, while the hopp<strong>in</strong>g energy is<br />

not sensitive to this factor. Much more work has been done to<br />

understand <strong>diffusion</strong> on Cu(100) [17,29,53,58–68,70–90,110–<br />

112]. Half the <strong>in</strong>vestigations considered exchange as an option,<br />

but only three studies gave an <strong>in</strong>dication <strong>of</strong> atom exchange<br />

tak<strong>in</strong>g place. For Pd(100) [17,27,59–62,64,66,69–71,92,93],<br />

three estimates favored hopp<strong>in</strong>g as the pr<strong>in</strong>cipal mechanism,<br />

but after <strong>in</strong>creas<strong>in</strong>g the size <strong>of</strong> the cell <strong>in</strong> the calculations<br />

three favoured exchange [17,60,65]. Only one out <strong>of</strong> six studies<br />

considered exchange <strong>in</strong> <strong>diffusion</strong> on Ag(100) [17,27,59–62,64–<br />

71,76,91,94–97,109–112] possibly <strong>in</strong>dicat<strong>in</strong>g a rate <strong>of</strong> atom<br />

exchange comparable to atom hopp<strong>in</strong>g; the rema<strong>in</strong>der gave<br />

hopp<strong>in</strong>g as the mode <strong>of</strong> <strong>diffusion</strong>. Not too much calculational<br />

work has been done to evaluate <strong>diffusion</strong> characteristics on<br />

Ir(100) [37,92], Pt(100) [17,92,98], and Au(100) [17,66,99],<br />

but all <strong>of</strong> them <strong>in</strong>dicate that the preferred path for <strong>diffusion</strong> was<br />

by atom exchange.<br />

Although the outcome <strong>of</strong> some <strong>of</strong> the theoretical estimates<br />

is not that certa<strong>in</strong>, the evidence is firm that exchange between<br />

an adatom and a lattice atom occurs <strong>in</strong> <strong>diffusion</strong> on (100)<br />

planes <strong>of</strong> Ir [45], Pt [44], and Ni [49], and probably also on<br />

Au [17,66,99]. There have been attempts to rationalize the<br />

conditions under which exchange will dom<strong>in</strong>ate <strong>in</strong> <strong>diffusion</strong>.<br />

For Al(100), Feibelman [100] po<strong>in</strong>ted out that the transition<br />

state for atom exchange was stabilized by the covalent nature<br />

<strong>of</strong> alum<strong>in</strong>um. Kellogg et al. [47] correlated exchange <strong>diffusion</strong><br />

with the relaxation <strong>of</strong> <strong>surface</strong> atoms around the b<strong>in</strong>d<strong>in</strong>g site <strong>of</strong><br />

the adatom. Yu and Scheffler [94] argued that “tensile <strong>surface</strong><br />

stress plays the key role for the exchange <strong>diffusion</strong> on fcc(100)<br />

<strong>surface</strong>s”, and is important not only for Au(100), but also for<br />

Fig. 25. Multiple atom exchange process <strong>in</strong> molecular dynamics simulation <strong>of</strong><br />

adatom on Cu(100) at 900 K [75]. Adatom (black) moves <strong>in</strong>to substrate, caus<strong>in</strong>g<br />

eventual emergence <strong>of</strong> a substrate atom at some distance from the orig<strong>in</strong>al entry.<br />

Al(100) and 5d metals. However, Feibelman and Stumpf [92]<br />

have done detailed density functional calculations for the (100)<br />

<strong>surface</strong>s <strong>of</strong> Rh, Ir, Pd, and Pt, and found no clear relation<br />

between <strong>surface</strong> stress and the <strong>diffusion</strong> barrier. Instead, they<br />

concluded that exchange <strong>diffusion</strong> was favored, as proposed<br />

by Kellogg et al. [47], when the relaxation <strong>of</strong> the substrate<br />

around an adatom was largest. Right now, however, it appears<br />

that predict<strong>in</strong>g from experimental <strong>in</strong>formation which systems<br />

will undergo atom exchange <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong> is an uncerta<strong>in</strong><br />

matter.<br />

2.3. Via multiple events<br />

The results for exchange <strong>processes</strong> <strong>in</strong> <strong>diffusion</strong> described<br />

so far have been obta<strong>in</strong>ed at reasonably low temperatures,<br />

and have revealed a s<strong>in</strong>gle event. Experiments at higher<br />

temperatures, to explore the possibility <strong>of</strong> multiple <strong>processes</strong>,<br />

are difficult and have not yet been explored. However, these<br />

conditions are accessible to molecular dynamics simulations,<br />

and have been probed start<strong>in</strong>g <strong>in</strong> 1993 with the work <strong>of</strong> Black<br />

and Tian [75], who studied copper on Cu(100) rely<strong>in</strong>g on<br />

embedded atom potentials [101]. At 900 K, a high temperature<br />

for copper, they found that an atom adsorbed on the <strong>surface</strong><br />

entered <strong>in</strong>to the <strong>surface</strong> layer, stra<strong>in</strong><strong>in</strong>g the adjacent <strong>surface</strong><br />

atoms as <strong>in</strong>dicated <strong>in</strong> Fig. 25. The stra<strong>in</strong> caused a <strong>surface</strong> atom<br />

to leave, popp<strong>in</strong>g out, not adjacent to the orig<strong>in</strong>al entry po<strong>in</strong>t,<br />

but farther away.<br />

This peculiar event was aga<strong>in</strong> found <strong>in</strong> the work <strong>of</strong><br />

Cohen [76], who did similar simulations on Ag, Al, Au, Cu, Pd,<br />

Pt, and Ni. On the (100) plane <strong>of</strong> alum<strong>in</strong>um she found that an<br />

atom entered the <strong>surface</strong> and travelled two sites and then over<br />

one before emerg<strong>in</strong>g aga<strong>in</strong>. The barrier for this novel <strong>diffusion</strong><br />

was much higher than for ord<strong>in</strong>ary hopp<strong>in</strong>g, and for all the<br />

metals above except for nickel would only become important<br />

above half the melt<strong>in</strong>g po<strong>in</strong>t. For nickel, the temperature for this<br />

<strong>diffusion</strong> process was expected to be even higher. An Arrhenius<br />

plot <strong>of</strong> the different <strong>diffusion</strong> <strong>processes</strong> is given <strong>in</strong> Fig. 26,<br />

and shows clearly that at elevated temperatures the new type<br />

<strong>of</strong> <strong>diffusion</strong> event makes significant contributions.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 49<br />

Fig. 26. Arrhenius plot for three different mechanisms <strong>of</strong> <strong>diffusion</strong> for Ag<br />

on Ag(100) derived <strong>in</strong> molecular dynamics simulations [76]. New refers to<br />

multiple displacement exchange process, which has a higher activation energy<br />

and therefore only contributes at elevated temperatures.<br />

Fig. 28. Arrhenius plot for frequency <strong>of</strong> jumps <strong>in</strong> self-<strong>diffusion</strong> on<br />

Cu(100) [82]. Filled squares—s<strong>in</strong>gle jumps; open squares—double jumps.<br />

Filled circles—simple exchange; filled and open triangles—double and triple<br />

exchange. Rhombic squares—quadruple exchange.<br />

Table 1<br />

<strong>Jump</strong> characteristics <strong>of</strong> Cu on Cu(100) [82]<br />

<strong>Jump</strong> Migration barrier (eV) D o (cm 2 /s)<br />

S<strong>in</strong>gle 0.43 ± 0.02 3.4×10 −3±0.2<br />

Double 0.71 ± 0.05 38 × 10 −3±0.5<br />

Simple exchange 0.70 ± 0.04 42 × 10 −3±0.3<br />

Double exchange 0.70 ± 0.06 45 × 10 −3±0.6<br />

Triple exchange 0.82 ± 0.08 104 ×<br />

10 −3±1.0<br />

Quadruple exchange 0.75 ± 0.12 86 × 10 −3±0.9<br />

Fig. 27. Molecular dynamics simulation <strong>of</strong> self-<strong>diffusion</strong> on Cu(100) at 950<br />

K [82]. Adatom squeezes <strong>in</strong>to the <strong>surface</strong> layer, and eventually a <strong>surface</strong> atom<br />

pops out at a considerable distance from the first entry.<br />

A more detailed exam<strong>in</strong>ation <strong>of</strong> high temperature <strong>surface</strong><br />

<strong>processes</strong> was carried out by Evangelakis and Papanicolaou<br />

[82] for copper atoms on the Cu(100) plane. They used<br />

the quite reliable RGL [102,103] potential to carry out their<br />

molecular dynamics simulations, which were done at 700 K and<br />

higher. At more elevated temperatures they observed double<br />

jumps, as well as exchange <strong>processes</strong>. Most <strong>in</strong>terest<strong>in</strong>g, however,<br />

were events such as shown <strong>in</strong> Fig. 27: an adatom enters<br />

the <strong>surface</strong> layer, disturb<strong>in</strong>g a row <strong>of</strong> <strong>surface</strong> atoms, the last<br />

<strong>of</strong> which exits the <strong>surface</strong> to become an adatom. From observations<br />

<strong>of</strong> the jump frequency at different temperatures they<br />

were able to construct the Arrhenius plot <strong>in</strong> Fig. 28, yield<strong>in</strong>g<br />

the <strong>diffusion</strong> characteristics <strong>of</strong> the different <strong>processes</strong>, shown<br />

<strong>in</strong> Table 1. It is important to recognize that the atom exchange<br />

events all have much the same k<strong>in</strong>etics, <strong>in</strong>dependent <strong>of</strong> the<br />

length <strong>of</strong> the process, which may be related to the extent <strong>of</strong><br />

Fig. 29. Schematics show<strong>in</strong>g atom movement <strong>in</strong> correlated jump-exchange<br />

(je), exchange-jump (ej), and f<strong>in</strong>ally jump-exchange-jump (jej) [104]. Dark<br />

grey <strong>in</strong>dicates atom <strong>in</strong>itially placed <strong>in</strong> channel; light grey shows atom from<br />

<strong>surface</strong> row tak<strong>in</strong>g part <strong>in</strong> exchange.<br />

the stra<strong>in</strong> field; these transitions also occur over barriers much<br />

higher than for ord<strong>in</strong>ary jumps.<br />

Also to be noted is the work <strong>of</strong> Ferrando on Ag(110) [104],<br />

who at temperatures above 600 K observed not only exchange<br />

but also correlated exchange-jump movements, as sketched <strong>in</strong><br />

Fig. 29. He also observed a drastic change <strong>in</strong> the frequency <strong>of</strong><br />

events when correlation between jumps and exchange occurred.<br />

Correlated jump and exchange <strong>processes</strong> were also observed on<br />

Cu(110) [32].


50 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 30. Temperature dependence <strong>of</strong> length <strong>of</strong> correlated jumps <strong>in</strong> <strong>diffusion</strong> <strong>of</strong><br />

adatom on Lennard-Jones fcc(100) <strong>surface</strong> [107].<br />

Fig. 32. Adatom trajectories on bcc(110) plane modelled with Price<br />

potential [42,43] dur<strong>in</strong>g 200 ps at ∼0.4T m [41]. (a) S<strong>in</strong>gle jumps. (b) Double<br />

jumps. (c) Complicated trajectories.<br />

3. Long and rebound jumps<br />

3.1. Theoretical work<br />

Fig. 31. Adatom trajectories on Lennard-Jones (100) <strong>surface</strong> at T = 0.34T m ,<br />

reveal<strong>in</strong>g non-nearest neighbour transitions [11].<br />

The study by Evangelakis and Papanicolaou [82] has been<br />

the most detailed work on multiple exchange events, but<br />

these <strong>processes</strong> were rediscovered several years later by Xiao<br />

et al. [105,106]. Us<strong>in</strong>g EAM potentials they found multiple<br />

atom exchange events for stra<strong>in</strong>ed Cu(100) and Pt(100), which<br />

were designated as crowdions.<br />

Although as yet there are no experiments to demonstrate<br />

such large scale exchange <strong>processes</strong>, there is little doubt that<br />

they will be found at elevated temperatures. However, detailed<br />

theoretical studies <strong>of</strong> such complicated <strong>processes</strong> will have<br />

to wait until procedures for <strong>in</strong>vestigat<strong>in</strong>g simple exchanges<br />

become reliable. One th<strong>in</strong>g is certa<strong>in</strong>: much more extended<br />

cell sizes will be needed for such calculations. From the<br />

experimental po<strong>in</strong>t <strong>of</strong> view, STM should be the most suitable<br />

tool to uncover such transitions.<br />

The view that <strong>in</strong> <strong>diffusion</strong> over a <strong>surface</strong>, atoms jump at<br />

random between nearest-neighbor sites, rema<strong>in</strong>ed widespread<br />

until the end <strong>of</strong> the seventies. At that time a number <strong>of</strong><br />

simulations appeared which suggested a more complicated<br />

picture <strong>of</strong> atomic events. Tully et al. [107] <strong>in</strong> 1979 carried out<br />

ghost particle simulations for a (100) <strong>surface</strong> <strong>of</strong> a Lennard-<br />

Jones crystal at different temperatures below the melt<strong>in</strong>g po<strong>in</strong>t<br />

T m . They discovered that, as shown <strong>in</strong> Fig. 30, the average<br />

jump length more than doubled as the temperature <strong>in</strong>creased<br />

from 0.2T m to 0.6T m . More extensive molecular dynamics were<br />

carried out at much the same time by DeLorenzi et al. [10,11],<br />

aga<strong>in</strong> on an fcc Lennard-Jones crystal. Most <strong>in</strong>terest<strong>in</strong>g was<br />

the fact that on the (100) plane at ∼0.3T m long jumps were<br />

observed, between sites as much as three to four spac<strong>in</strong>gs apart.<br />

This is clear from the atom trajectories <strong>in</strong> Fig. 31.<br />

Corrections to <strong>diffusion</strong> rates predicted by transition-state<br />

theory for Rh(100) stemm<strong>in</strong>g from transitions to sites farther<br />

away than a nearest-neighbour distance were done by Voter<br />

and Doll [108], aga<strong>in</strong> with Lennard-Jones <strong>in</strong>teractions. Only<br />

<strong>in</strong> the vic<strong>in</strong>ity <strong>of</strong> 1000 K were jumps to other than nearestneighbour<br />

positions found. In the same year, 1985, DeLorenzi<br />

and Jacucci [41] published molecular dynamics simulation<br />

on various bcc <strong>surface</strong>s modelled with a metallic potential<br />

developed by Price [42,43] for sodium. On the most densely<br />

packed <strong>surface</strong>, the (110), at a temperature <strong>of</strong> 0.4T m , they<br />

found not only jumps between nearest-neighbour sites, as <strong>in</strong><br />

Fig. 32(a), but also double and more elaborate transitions,<br />

shown <strong>in</strong> Fig. 32(b) and (c). Evangelakis and Papanicolaou [82]<br />

also saw that double jumps started to be active on the (100)<br />

plane <strong>of</strong> copper, an fcc metal, at temperatures above 750 K.<br />

Above 800 K, more complicated <strong>processes</strong>, such as quadruple<br />

exchange began play<strong>in</strong>g a role as well.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 51<br />

the length <strong>of</strong> atomic jumps can be obta<strong>in</strong>ed, as the probability<br />

that after a time t an atom will be at a displacement x from the<br />

orig<strong>in</strong> was given by Wrigley et al. [114] as<br />

Fig. 33. Schematic <strong>of</strong> s<strong>in</strong>gle, double and triple transitions <strong>in</strong> one-dimensional<br />

atom motion.<br />

In 1996, Ferrando [104] looked at self-<strong>diffusion</strong> on Ag(110)<br />

and found that s<strong>in</strong>gle jumps represented ∼90% <strong>of</strong> the total, the<br />

rest were double or more complicated transitions. Investigations<br />

<strong>of</strong> long jumps were done by Montalenti and Ferrando [32] who<br />

looked at self-<strong>diffusion</strong> on the (110) planes <strong>of</strong> gold, silver and<br />

copper. Although the activation energy for movement on gold<br />

and silver is almost the same, and for copper slightly lower,<br />

their behaviour with respect to long jumps differed greatly. At<br />

450 K, long jumps were absent on gold, there were 3% long<br />

jumps on silver, and 6% on copper. For copper, the fraction<br />

<strong>of</strong> long jumps <strong>in</strong>creased to 15% at 600 K, but never got to<br />

this value on the Ag(110) <strong>surface</strong>. Ferrón et al. [113] saw long<br />

jumps as well as rebound jumps <strong>in</strong> self-<strong>diffusion</strong> on Cu(111).<br />

They claimed that at 500 K, 95% <strong>of</strong> jumps were correlated; at<br />

100 K this decreased to 50%.<br />

What is quite clear from these simulations is that at elevated<br />

temperatures, larger than 0.2T m , long jumps should participate<br />

significantly <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>. The question still rema<strong>in</strong>ed,<br />

however, how to detect the transitions <strong>in</strong> an experiment.<br />

3.2. Experimental studies<br />

Information about the characteristics <strong>of</strong> <strong>surface</strong> <strong>diffusion</strong><br />

has usually been obta<strong>in</strong>ed from measurements <strong>of</strong> the meansquare<br />

displacement. However, these measurements provide<br />

no simple connection to the types <strong>of</strong> transitions carried out<br />

by atoms diffus<strong>in</strong>g over the <strong>surface</strong>. Usually an <strong>in</strong>crease <strong>in</strong><br />

the diffusivity is expected due to long transitions, s<strong>in</strong>ce the<br />

diffusivity <strong>in</strong> Eq. (4) depends on the jump length squared, but<br />

so far this expectation has not been realized <strong>in</strong> experiments.<br />

More <strong>in</strong>formation is def<strong>in</strong>itely needed. From measurements <strong>of</strong><br />

the distribution <strong>of</strong> displacements <strong>in</strong> one dimension <strong>in</strong>sight <strong>in</strong>to<br />

p x (t) = exp[−2(α + β + γ )t]<br />

∞∑<br />

∞∑<br />

× I k (2γ t)<br />

k=−∞<br />

j=−∞<br />

I j (2βt)I x−2 j−3k (2αt). (8)<br />

Here it is assumed that s<strong>in</strong>gle jumps at the rate α, double jumps<br />

at the rate β, and triple jumps at the rate γ take place on<br />

the <strong>surface</strong>, as is illustrated <strong>in</strong> Fig. 33; I m (u) is the modified<br />

Bessel function <strong>of</strong> the first k<strong>in</strong>d, <strong>of</strong> order m and argument u.<br />

The jump rates clearly affect the probability <strong>of</strong> f<strong>in</strong>d<strong>in</strong>g a set<br />

<strong>of</strong> atom displacements, and that is more immediately evident<br />

from the plots <strong>in</strong> Fig. 34, where s<strong>in</strong>gle and double transitions<br />

participate <strong>in</strong> <strong>diffusion</strong>. All that needs to be done is to carry out<br />

an adequate number <strong>of</strong> observations <strong>of</strong> atomic displacements,<br />

and from these deduce the probability p x (t). To derive the<br />

<strong>in</strong>dividual jump rates, Eq. (8) is not all that useful, however,<br />

as it holds for <strong>diffusion</strong> on an <strong>in</strong>f<strong>in</strong>ite plane. For experiments <strong>in</strong><br />

the field ion microscope the <strong>in</strong>dividual planes are quite small,<br />

and may extend over only ∼20 spac<strong>in</strong>gs, so that Monte Carlo<br />

simulations must be employed to extract rates.<br />

The first attempt to derive jump rates from the measured<br />

distribution <strong>of</strong> displacements was made <strong>in</strong> 1989 by Wang<br />

et al. [115] <strong>in</strong> one-dimensional <strong>diffusion</strong> on W(211). Tested<br />

were Re, Mo, Ir, and Rh atoms. As shown for rhenium <strong>in</strong><br />

Fig. 35, the best fit to the distribution measured at 300 K<br />

was obta<strong>in</strong>ed assum<strong>in</strong>g only s<strong>in</strong>gle jumps. Even though long<br />

jumps were not found, for the first time the picture <strong>of</strong> <strong>diffusion</strong><br />

as occurr<strong>in</strong>g by random jumps between nearest-neighbor sites<br />

had been demonstrated. Much the same also held for the other<br />

atoms studied, although for iridium and rhodium there were<br />

negligibly small contributions from double jumps.<br />

The next step was taken by Senft [116–118], who surmised<br />

that energy transfer between the mov<strong>in</strong>g adatom and the<br />

lattice would affect the participation <strong>of</strong> long jumps <strong>in</strong> <strong>surface</strong><br />

<strong>diffusion</strong>. She therefore elected to study palladium as well as<br />

nickel, with low values <strong>of</strong> the <strong>diffusion</strong> barriers amount<strong>in</strong>g to<br />

0.314±0.006 eV for Pd and 0.46±0.06 eV for Ni, which imply<br />

rather poor energy transfer. The displacement distribution was<br />

tested for self-<strong>diffusion</strong> <strong>of</strong> tungsten on W(211) at 307 K, and as<br />

Fig. 34. Effect <strong>of</strong> double jumps on the distribution <strong>of</strong> atomic displacements <strong>in</strong> <strong>diffusion</strong> with a mean-square displacement <strong>of</strong> two.


52 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 35. Distribution <strong>of</strong> Re adatom displacements at 300 K on W(211), obta<strong>in</strong>ed from field ion observations [115]. Best fit to experiments obta<strong>in</strong>ed with an entirely<br />

negligible contribution from long jumps.<br />

Fig. 36. Displacement distribution for W adatom on W(211) at 307 K [117].<br />

Best fit to field ion experiments derived with negligible contribution <strong>of</strong> β double<br />

or longer jumps.<br />

Fig. 37. Distribution <strong>of</strong> Pd atom displacements on W(211) at 133 K [118]. Best<br />

fit with double/s<strong>in</strong>gle jumps equal to 0.20, and triple/s<strong>in</strong>gle <strong>of</strong> 0.13.<br />

shown <strong>in</strong> Fig. 36 was found to be due entirely to s<strong>in</strong>gle jumps<br />

between nearest-neighbor sites, as expected at the time.<br />

The same was found for the <strong>diffusion</strong> <strong>of</strong> palladium at 114<br />

and 122 K. However, at 133 K, the distribution <strong>in</strong> Fig. 37 for<br />

the first time gave a clear <strong>in</strong>dication <strong>of</strong> significant contributions<br />

from long jumps; the ratio <strong>of</strong> double to s<strong>in</strong>gle jumps was 0.20,<br />

Fig. 38. Distribution <strong>of</strong> Ni adatom displacements on W(211) plane at 179<br />

K [117]. Best fit <strong>of</strong> observations with double/s<strong>in</strong>gle jumps equal to 0.058.<br />

and even triple jumps were detected, at a ratio <strong>of</strong> 0.13 for triple<br />

to s<strong>in</strong>gle transitions. In the <strong>diffusion</strong> <strong>of</strong> nickel atoms on W(211),<br />

Senft [117] found a distribution shown <strong>in</strong> Fig. 38, best fit with<br />

a ratio <strong>of</strong> doubles to s<strong>in</strong>gles <strong>of</strong> 0.058 at T = 179 K. What<br />

was surpris<strong>in</strong>g about these f<strong>in</strong>d<strong>in</strong>gs is not just the detection<br />

<strong>of</strong> long jumps, but long jumps at quite a low temperature, <<br />

0.1T m , and at a rate very temperature sensitive. For palladium,<br />

a dim<strong>in</strong>ution <strong>of</strong> the temperature by 11 K sufficed to elim<strong>in</strong>ate<br />

all long transitions.<br />

With long jumps now firmly established <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>,<br />

L<strong>in</strong>deroth et al. [119] decided to explore their rates <strong>in</strong> self<strong>diffusion</strong><br />

on the reconstructed Pt(110)-(1 × 2) plane, shown<br />

<strong>in</strong> Fig. 39. <strong>Jump</strong>s were observed with the scann<strong>in</strong>g tunnel<strong>in</strong>g<br />

microscope, which yielded the distribution at 375 K <strong>in</strong> Fig. 40,<br />

with a ratio <strong>of</strong> “double” to s<strong>in</strong>gle jumps <strong>of</strong> 0.095. They also<br />

made measurements over a temperature range <strong>of</strong> 60 K to come<br />

up with the Arrhenius plot <strong>in</strong> Fig. 41, which gave a barrier<br />

<strong>of</strong> 0.81 ± 0.01 eV for s<strong>in</strong>gle jumps and a somewhat higher<br />

value, 0.89 ± 0.06 eV, identified as com<strong>in</strong>g from doubles.<br />

This identification turned out to be premature, however. A year<br />

passed and Montalenti and Ferrando [120] did simulations <strong>of</strong><br />

<strong>diffusion</strong> on Au(110)-(1×2), rely<strong>in</strong>g on RGL <strong>in</strong>teractions [102,<br />

103]. They discovered two prevalent jumps, illustrated <strong>in</strong>


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 53<br />

Fig. 39. Hard-sphere model <strong>of</strong> (1×2) reconstructed fcc(110) <strong>surface</strong>, <strong>in</strong> which<br />

every second 〈111〉 row has been removed.<br />

Fig. 41. Arrhenius plot for s<strong>in</strong>gle and double jump rates presumed to occur <strong>in</strong><br />

self-<strong>diffusion</strong> on Pt(110)-(1 × 2) [119].<br />

Fig. 40. Distribution <strong>of</strong> atomic displacements <strong>in</strong> self-<strong>diffusion</strong> on Pt(110)-<br />

(1 × 2) plane [119]. Best fit obta<strong>in</strong>ed with ratio <strong>of</strong> “double” to s<strong>in</strong>gle jumps<br />

<strong>of</strong> 0.095.<br />

Fig. 42: transitions along the bottom <strong>of</strong> the <strong>diffusion</strong> channel,<br />

and <strong>in</strong> addition transitions <strong>in</strong> which the atom jumps to the<br />

(111) sidewalls and cont<strong>in</strong>ued <strong>diffusion</strong> there. The number <strong>of</strong><br />

these metastable transitions exceeded that <strong>of</strong> long jumps <strong>in</strong> the<br />

channel. Another year later, Lorensen et al. [121] published<br />

density functional estimates for <strong>diffusion</strong> on Pt(110)-(1 × 2),<br />

which confirmed what had previously been found for gold—<br />

metastable jumps were the likely explanation for the f<strong>in</strong>d<strong>in</strong>gs<br />

<strong>of</strong> L<strong>in</strong>deroth et al. [119].<br />

Up to this po<strong>in</strong>t, long jumps had been identified<br />

experimentally only on the channeled, one-dimensional <strong>surface</strong><br />

<strong>of</strong> the W(211) plane, and only for rapidly diffus<strong>in</strong>g atoms. Oh<br />

et al. [122] <strong>in</strong> 2002 undertook tests to see if they also occurred<br />

on W(110), <strong>in</strong> the two-dimensional <strong>diffusion</strong> <strong>of</strong> palladium. Of<br />

course this is a rather more complicated system, and jumps may<br />

take place <strong>in</strong> quite a number <strong>of</strong> ways, as <strong>in</strong>dicated <strong>in</strong> Fig. 43.<br />

The Arrhenius plot <strong>in</strong> Fig. 44 did not show any evidence <strong>of</strong><br />

long jumps. However, an Arrhenius plot is not a good <strong>in</strong>dicator<br />

<strong>of</strong> the jump <strong>processes</strong> <strong>in</strong> <strong>diffusion</strong>. For this we have to rely<br />

on the distribution <strong>of</strong> displacements, shown <strong>in</strong> Fig. 45. At a<br />

Fig. 42. Trajectories <strong>of</strong> Au adatom on reconstructed Au(110) <strong>surface</strong> with<br />

〈110〉 rows miss<strong>in</strong>g, obta<strong>in</strong>ed <strong>in</strong> molecular dynamics simulations at 450<br />

K [120]. Left column: <strong>in</strong>-channel jumps s<strong>in</strong>gle, double, and triple. Right<br />

column: metastable transitions s<strong>in</strong>gle, double, and triple.<br />

temperature <strong>of</strong> 210 K this distribution gave a significant number<br />

<strong>of</strong> double β jumps along 〈111〉 as well as vertical δ y transitions.<br />

After correction for effects due to transient temperatures dur<strong>in</strong>g<br />

sample heat<strong>in</strong>g and cool<strong>in</strong>g, the ratio <strong>of</strong> β/α proved to be<br />

0.12 ± 0.06, and 0.11 ± 0.10 was found for δ y /α, the first<br />

experimental demonstration <strong>of</strong> long jumps <strong>in</strong> two-dimensional<br />

<strong>diffusion</strong>. Worth not<strong>in</strong>g here is the difference between vertical


54 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 43. Schematic <strong>of</strong> possible atom jumps on W(110) plane.<br />

Fig. 45. Displacement distribution <strong>of</strong> Pd adatom on W(110) plane at 210<br />

K [122]. Double/s<strong>in</strong>gle jump rate β/α is 0.12, δ y /α is 0.11.<br />

Fig. 44. Dependence <strong>of</strong> the diffusivity <strong>of</strong> s<strong>in</strong>gle Pd adatom on W(110) upon<br />

the reciprocal temperature [122]. Data corrected for edge effects as well as for<br />

migration dur<strong>in</strong>g transients.<br />

and horizontal jumps, which is huge. It has been possible to<br />

derive a significant energy difference between the two k<strong>in</strong>ds <strong>of</strong><br />

jumps.<br />

Oh et al. [123] also exam<strong>in</strong>ed the self-<strong>diffusion</strong> <strong>of</strong> tungsten<br />

atoms on W(110), and surpris<strong>in</strong>gly found behaviour similar to<br />

that <strong>of</strong> palladium. At 365 K, the distribution yielded β/α =<br />

0.22, δ x /α = 0.36, and δ y /α = 0.43. Half <strong>of</strong> the transitions<br />

now were long jumps. Of particular <strong>in</strong>terest are the vertical δ y<br />

and horizontal δ x transitions. It is not yet clear how they take<br />

place, but they can be envisioned as start<strong>in</strong>g as jumps <strong>in</strong> the<br />

〈111〉 direction, which are then deviated either toward the x- or<br />

y-axis.<br />

Although long jumps had now been found for a variety<br />

<strong>of</strong> atoms <strong>in</strong> both one- and two-dimensional <strong>diffusion</strong>, noth<strong>in</strong>g<br />

was known about the rates <strong>of</strong> these transitions. This matter<br />

was tackled by Antczak [124]. In 2004 she exam<strong>in</strong>ed <strong>in</strong> detail<br />

the distribution <strong>of</strong> displacements <strong>of</strong> tungsten atoms on W(110)<br />

Fig. 46. Arrhenius plot for diffusivities <strong>of</strong> W atom on W(110) along 〈100〉 and<br />

〈110〉 direction [124]. Best fit is obta<strong>in</strong>ed with a straight l<strong>in</strong>e.<br />

over a range <strong>of</strong> temperatures and with very good statistics <strong>of</strong><br />

1200 observations. From an Arrhenius plot <strong>of</strong> the diffusivities,<br />

<strong>in</strong> Fig. 46, she obta<strong>in</strong>ed straight l<strong>in</strong>es, <strong>in</strong>dicat<strong>in</strong>g a barrier<br />

<strong>of</strong> 0.92 ± 0.02 eV for <strong>diffusion</strong> along 〈100〉 and much the<br />

same barrier <strong>of</strong> 0.93 ± 0.01 eV along 〈110〉. Aga<strong>in</strong> there<br />

were no <strong>in</strong>dications <strong>in</strong> the diffusivity <strong>of</strong> anyth<strong>in</strong>g to suggest<br />

contributions from long jumps. However, the distribution <strong>of</strong><br />

displacements at elevated temperatures clearly showed such<br />

transitions, as is evident from Fig. 47. It must be noted that at<br />

these high temperatures significant displacements occur dur<strong>in</strong>g<br />

the temperature rise before the <strong>diffusion</strong> <strong>in</strong>terval and dur<strong>in</strong>g the<br />

fall at the end. The distribution <strong>of</strong> displacements dur<strong>in</strong>g these<br />

temperature transients therefore has to be determ<strong>in</strong>ed and is<br />

also shown <strong>in</strong> Fig. 47. The f<strong>in</strong>al rates were obta<strong>in</strong>ed us<strong>in</strong>g the<br />

relation<br />

r = Rt − r ot o<br />

t e<br />

(9)


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 55<br />

Table 2<br />

<strong>Jump</strong> rate parameters for W on W(110) [124]<br />

Rate Activation energy (eV) Frequency factor υ o (s −1 )<br />

α (low temperature) 0.94 ± 0.03 5.92(×2.5 ±1 ) × 10 12<br />

β 1.24 ± 0.13 8.06(×8.1 ±1 ) × 10 15<br />

δ x 1.28 ± 0.13 3.78(×8.6 ±1 ) × 10 16<br />

δ y 1.37 ± 0.13 1.00(×4.4 ±1 ) × 10 18<br />

Fig. 49. Arrhenius plots for δ x and δ y jumps <strong>of</strong> W adatom on W(110) [124].<br />

Fig. 47. Distribution <strong>of</strong> W displacements on W(110) at 364 K [124]. Best fit<br />

obta<strong>in</strong>ed with significant contributions from long jumps. Inset gives distribution<br />

dur<strong>in</strong>g temperature transients. Rates not corrected for effects from transients.<br />

Fig. 48. Arrhenius plots for rates <strong>of</strong> s<strong>in</strong>gle α jumps and double β jumps <strong>of</strong><br />

W adatom on W(110) [124]. S<strong>in</strong>gle jump rate is fitted at low temperatures. At<br />

elevated temperatures, rate drops significantly below this straight l<strong>in</strong>e.<br />

where R is the rate measured <strong>in</strong> the experiment for an <strong>in</strong>terval<br />

t, r o and t o give these quantities determ<strong>in</strong>ed for the transients,<br />

and t e is the effective time <strong>in</strong>terval.<br />

In Fig. 48 are shown the Arrhenius plots for the rates <strong>of</strong><br />

tungsten s<strong>in</strong>gle jumps α and double jumps β. What is most<br />

<strong>in</strong>terest<strong>in</strong>g here is that above ∼340 K, the rate α drops below<br />

the straight l<strong>in</strong>e extrapolated from low temperatures. This is<br />

not the case for the other rates β, nor for δ x or δ y shown<br />

<strong>in</strong> Fig. 49. Clearly the rate <strong>of</strong> s<strong>in</strong>gle transitions is <strong>in</strong>fluenced<br />

by the occurrence <strong>of</strong> the other jumps. The k<strong>in</strong>etics <strong>of</strong> these<br />

transitions are listed <strong>in</strong> Table 2. The barriers to the long jumps<br />

are all considerably higher than for s<strong>in</strong>gle transitions; so are<br />

the frequency prefactors, 5.92 × 10 12 s −1 for s<strong>in</strong>gle jumps and<br />

1.00 × 10 18 s −1 for vertical transitions.<br />

How can all <strong>of</strong> this be understood? The important th<strong>in</strong>g to<br />

recognize is that the various rates are not <strong>in</strong>dependent. If the<br />

sum <strong>of</strong> all the rates is displayed on an Arrhenius plot, as <strong>in</strong><br />

Fig. 50, a straight l<strong>in</strong>e is obta<strong>in</strong>ed, not a l<strong>in</strong>e curved concave<br />

upward. This can be understood <strong>in</strong> terms <strong>of</strong> the schematic <strong>in</strong><br />

Fig. 51. An atom starts a transition toward a nearest-neighbor<br />

site, but may cont<strong>in</strong>ue beyond this. Every time one <strong>of</strong> these<br />

other transitions β, δ x , or δ y occurs, the number <strong>of</strong> s<strong>in</strong>gle jumps<br />

is dim<strong>in</strong>ished. At higher temperatures, where the higher barriers<br />

to these transitions can be overcome, the rate α will therefore<br />

become smaller.<br />

How can we account for the high barriers and prefactors<br />

<strong>of</strong> the long transitions? For the range <strong>of</strong> temperatures <strong>in</strong> the<br />

experiments, we can write the rate r i <strong>of</strong> transition i as<br />

r i = R × p i , (10)<br />

where R is the basic jump rate, given by<br />

(<br />

R = υ o exp − E )<br />

o<br />

, (11)<br />

kT


56 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 50. Arrhenius plot <strong>of</strong> the sum <strong>of</strong> all jump rates for W adatoms on W(110),<br />

giv<strong>in</strong>g a straight l<strong>in</strong>e fit [124].<br />

Fig. 52. Arrhenius plot for Ir adatom diffusivities on W(110) plane along 〈100〉<br />

and 〈110〉 [125].<br />

Fig. 51. Schematic <strong>of</strong> jump rates on W(110) plane [124]. Basic jump is α<br />

transition along 〈111〉. At elevated temperatures, this can proceed beyond<br />

nearest neighbour end po<strong>in</strong>t, giv<strong>in</strong>g β, δ x , or δ y .<br />

and p i is the probability <strong>of</strong> a particular transition i. This<br />

probability is given by a normalized expression similar to<br />

Eq. (11), so that the activation energy for the given rate r i<br />

is just the sum <strong>of</strong> the barriers, and the prefactor the product<br />

<strong>of</strong> the separate prefactors, which makes the results obta<strong>in</strong>ed<br />

understandable.<br />

The <strong>diffusion</strong> <strong>of</strong> tungsten on W(110) is not the only system<br />

considered so far. Antczak [125] also exam<strong>in</strong>ed the behaviour<br />

<strong>of</strong> iridium atoms on W(110). Arrhenius plots <strong>in</strong> Fig. 52 for<br />

<strong>diffusion</strong> along 〈100〉 and 〈110〉 are aga<strong>in</strong> quite straight, without<br />

any <strong>in</strong>dication <strong>of</strong> multiple jump <strong>processes</strong>. Nevertheless, longer<br />

jumps occur frequently, as is clear from the distribution <strong>of</strong><br />

displacements dur<strong>in</strong>g experiments at 366 K, shown <strong>in</strong> Fig. 53<br />

together with zero-time experiments, to catch displacements<br />

dur<strong>in</strong>g temperature transients. For the long jumps were found<br />

β/α = 0.15, δ x /α = 0.38, and δ y /α = 0.32. Just as previously<br />

with tungsten atoms above 350 K, the rate <strong>of</strong> s<strong>in</strong>gle α transitions<br />

falls below the straight extrapolation from low temperatures;<br />

the other jump rates β, δ x , and δ y fit on normal Arrhenius<br />

plots. The explanation is the same as for tungsten atoms—the<br />

Fig. 53. Distribution <strong>of</strong> Ir adatom displacements on W(110) <strong>surface</strong> at<br />

366 K [125]. Both normal experiments, and experiments dur<strong>in</strong>g transient<br />

temperatures are shown. Values for ratios <strong>of</strong> jump rates shown obta<strong>in</strong>ed after<br />

corrections for transient effects.<br />

jumps are not <strong>in</strong>dependent and arise by conversion from one<br />

elementary process.<br />

What is important here is that the long transitions <strong>of</strong> both<br />

iridium and tungsten replace the s<strong>in</strong>gle jumps on W(110)<br />

gradually and at quite a low temperature, less than a tenth<br />

<strong>of</strong> the melt<strong>in</strong>g po<strong>in</strong>t. There they start play<strong>in</strong>g a lead<strong>in</strong>g role<br />

<strong>in</strong> <strong>diffusion</strong>, and at higher temperatures s<strong>in</strong>gle transitions<br />

disappear completely. Surpris<strong>in</strong>g is the difference <strong>in</strong> the<br />

activation energies <strong>of</strong> vertical and horizontal jumps on W(110),<br />

which as shown <strong>in</strong> Fig. 54 is huge for palladium atoms, much<br />

lower for tungsten, and disappears for iridium atoms. It turns<br />

out that this difference has a l<strong>in</strong>ear dependence on mass, but it<br />

is so far based on only three po<strong>in</strong>ts.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 57<br />

Fig. 54. Differences <strong>in</strong> the activation energies for vertical and horizontal jumps<br />

<strong>of</strong> Pd, W, and Ir adatoms on W(110).<br />

Fig. 56. One-dimensional distribution <strong>of</strong> W adatom displacements at 325 K<br />

on W(211) [126]. Zero-time measurements to detect transient contributions (<strong>in</strong><br />

<strong>in</strong>set) have been used to give listed rates.<br />

Fig. 55. Arrhenius plot for self-<strong>diffusion</strong> <strong>of</strong> W adatom on W(211) plane [126].<br />

On tungsten, long jumps are not limited to two-dimensional<br />

<strong>diffusion</strong>. Senft [116–118] already showed this for Pd and<br />

Ni on W(211), but with tungsten she came to the conclusion<br />

that long jumps did not occur. More recently Antczak [126]<br />

has probed one-dimensional self-<strong>diffusion</strong> on W(211), but at<br />

higher temperatures than Senft, and longer transitions were<br />

discovered for this system as well. A normal Arrhenius plot<br />

for tungsten, with an activation energy <strong>of</strong> 0.81 ± 0.02 eV and<br />

a prefactor <strong>of</strong> 3.41(×2.40 ±1 ) × 10 −3 cm 2 /s was found, as<br />

shown <strong>in</strong> Fig. 55. At an elevated temperature <strong>of</strong> 325 K the<br />

distribution <strong>of</strong> displacements <strong>in</strong> Fig. 56 now revealed a ratio<br />

β/α <strong>of</strong> double to s<strong>in</strong>gle jumps <strong>of</strong> 0.66; transitions dur<strong>in</strong>g the<br />

temperature transients are given <strong>in</strong> the same figure. The plot <strong>of</strong><br />

the s<strong>in</strong>gle jump rate α <strong>in</strong> Fig. 57 drops sharply at temperatures<br />

<strong>of</strong> 300 K and above. At 325 K the rate <strong>of</strong> s<strong>in</strong>gle jumps has gone<br />

Fig. 57. Arrhenius plot for α s<strong>in</strong>gle jumps <strong>of</strong> W on W(211) [126]. At ∼320 K,<br />

rate α has decreased so much it is no longer discernable.<br />

to zero and only long jumps contribute to <strong>diffusion</strong>; the rate <strong>of</strong> β<br />

double jumps behaves normally with temperature, as is evident<br />

from Fig. 58.<br />

Long jumps are clearly demonstrated <strong>in</strong> this onedimensional<br />

system, but there is also a big surprise. As<br />

<strong>in</strong>dicated <strong>in</strong> Fig. 59, the sum <strong>of</strong> the two rates measured does<br />

not plot as a l<strong>in</strong>ear Arrhenius graph, as had been found on<br />

W(110); it appears that here not all types <strong>of</strong> jumps have been<br />

detected. What is miss<strong>in</strong>g are rebound transitions, illustrated <strong>in</strong><br />

Fig. 60. An atom starts on a jump; at the adjacent site it can<br />

settle down, creat<strong>in</strong>g a s<strong>in</strong>gle jump. It can also cont<strong>in</strong>ue on to<br />

the next site to form a double transition, or else it can rebound<br />

to return to the start<strong>in</strong>g po<strong>in</strong>t. No displacements arise from


58 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

Fig. 60. Schematic <strong>of</strong> different types <strong>of</strong> jump <strong>processes</strong> for an adatom on<br />

W(211) [126]. β R denotes rebound transition.<br />

Fig. 58. Normal Arrhenius plot for β double jumps <strong>of</strong> W adatom on<br />

W(211) [126].<br />

Fig. 61. Rate <strong>of</strong> rebound jumps, obta<strong>in</strong>ed as the difference between the<br />

sum <strong>of</strong> s<strong>in</strong>gle plus double jumps and total jump rate extrapolated from low<br />

temperatures [129].<br />

Fig. 59. Arrhenius plot for the sum <strong>of</strong> all jump rates measured for W adatom<br />

on W(211). For temperatures ≥310 K, rate falls below l<strong>in</strong>ear plot [126].<br />

such a rebound transition, and it is therefore not detected <strong>in</strong><br />

the normal <strong>diffusion</strong> measurements, but rebounds were found <strong>in</strong><br />

molecular dynamics simulations <strong>in</strong> 1989 by DeLorenzi [127],<br />

and a few years later by Sanders and DePristo [128] and Ferrón<br />

et al. [113].<br />

A way to measure such rebounds has recently been<br />

discovered. Antczak [129] has po<strong>in</strong>ted out that the sum <strong>of</strong> the<br />

measured jump rates has to be subtracted from the straightl<strong>in</strong>e<br />

Arrhenius plot obta<strong>in</strong>ed by extrapolat<strong>in</strong>g the sum from<br />

low temperatures. The rebound rate so obta<strong>in</strong>ed is shown<br />

<strong>in</strong> Fig. 61; it has an activation energy <strong>of</strong> 1.03 ± 0.06 eV<br />

and a frequency prefactor <strong>of</strong> 1.40(×10.3 ±1 ) × 10 16 s −1 , and<br />

thus lies midway between s<strong>in</strong>gle and double jumps. What<br />

is surpris<strong>in</strong>g is that rebounds were not detected <strong>in</strong> twodimensional<br />

<strong>diffusion</strong> on W(110) at all, suggest<strong>in</strong>g it may be an<br />

effect tied to the channelled structure <strong>of</strong> W(211) that is <strong>in</strong>volved<br />

<strong>in</strong> these transitions. However, they have been seen previously <strong>in</strong><br />

simulations <strong>of</strong> self-<strong>diffusion</strong> on the Cu(111) plane [113].<br />

It should be noted that long jumps are not limited to the<br />

movement <strong>of</strong> s<strong>in</strong>gle atoms. As is apparent from Fig. 62,<br />

long jumps were observed by Wang [130,131] for clusters <strong>of</strong><br />

iridium atoms on Ir(111), and for the large organic molecules<br />

decacyclene and hexa-tert-butyl-decacyclene by Schunack<br />

et al. [132]. However, <strong>in</strong>formation about the k<strong>in</strong>etics <strong>of</strong> these<br />

<strong>processes</strong> is limited and cry<strong>in</strong>g for more work.<br />

F<strong>in</strong>ally, it should be said that theoretical predictions about<br />

long jumps have been very important <strong>in</strong> po<strong>in</strong>t<strong>in</strong>g to their<br />

contributions <strong>in</strong> <strong>diffusion</strong>. However, even the limited number<br />

<strong>of</strong> experiments done so far have already revealed that these<br />

transitions are rather more varied and more significant than<br />

predicted, especially at low temperatures.<br />

4. Summary<br />

This survey should make it clear that a whole variety <strong>of</strong><br />

different jumps contributes to <strong>surface</strong> <strong>diffusion</strong> on metals. Atom<br />

exchange <strong>processes</strong> are <strong>of</strong> course very much affected by the<br />

chemistry <strong>of</strong> the <strong>in</strong>teract<strong>in</strong>g partners, but have so far only<br />

been observed <strong>in</strong> experiments with fcc metals. On channelled<br />

<strong>surface</strong>s the direction favoured <strong>in</strong> <strong>diffusion</strong> is still puzzl<strong>in</strong>g.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 59<br />

Fig. 62. Centre-<strong>of</strong>-mass displacements for Ir 19 cluster dur<strong>in</strong>g 10 s <strong>in</strong>tervals at<br />

690 K [131]. Observed number <strong>of</strong> displacements shown bold, best fit <strong>in</strong> outl<strong>in</strong>e<br />

numerals just below. α, β, as well as γ transitions contribute significantly.<br />

Very <strong>in</strong>terest<strong>in</strong>g are the multiple exchange <strong>processes</strong> seen <strong>in</strong><br />

simulations at elevated temperatures, but these have not yet<br />

been found <strong>in</strong> experiments.<br />

The situation may well be different for the rate <strong>of</strong> long<br />

jumps. These have so far only been detected <strong>in</strong> experiments on<br />

tungsten <strong>surface</strong>s, on which it has been possible to determ<strong>in</strong>e<br />

jump rates. There it is clear that <strong>in</strong> self-<strong>diffusion</strong> these jumps<br />

already occur at low temperatures less than 1/10 the melt<strong>in</strong>g<br />

po<strong>in</strong>t; at more elevated temperatures they supplant s<strong>in</strong>gle atom<br />

jumps as carriers <strong>of</strong> the <strong>diffusion</strong> process. So far these long<br />

transitions have not been observed on other metals, even though<br />

the expectation is that long jumps will prove to be quite<br />

general, <strong>in</strong>dependent <strong>of</strong> the substrate, and important at elevated<br />

temperatures, where new types <strong>of</strong> jumps may also be found.<br />

Surface <strong>diffusion</strong> has now been explored on the atomic level<br />

for forty years, but new effects are still be<strong>in</strong>g discovered, and<br />

that suggests important work is ahead.<br />

Acknowledgements<br />

This work was done while supported by the Petroleum<br />

Research Fund, adm<strong>in</strong>istered by the ACS under Grant ACS<br />

PRF 44958-AC5. We are also <strong>in</strong>debted to Mary Kay Newman<br />

for her help with the literature.<br />

References<br />

[1] J.B. Taylor, I. Langmuir, The evaporation <strong>of</strong> atoms, ions and electrons<br />

from caesium films on tungsten, Phys. Rev. 44 (1933) 423.<br />

[2] G.L. Kellogg, Field ion microscope studies <strong>of</strong> s<strong>in</strong>gle-atom <strong>surface</strong><br />

<strong>diffusion</strong> and cluster nucleation on metal <strong>surface</strong>s, Surf. Sci. Rep. 21<br />

(1994) 1.<br />

[3] G. Ehrlich, F.G. Hudda, Atomic view <strong>of</strong> <strong>surface</strong> self-<strong>diffusion</strong>: Tungsten<br />

on tungsten, J. Chem. Phys. 44 (1966) 1039.<br />

[4] G. Ayrault, G. Ehrlich, Surface self-<strong>diffusion</strong> on an fcc crystal: An<br />

atomic view, J. Chem. Phys. 60 (1974) 281.<br />

[5] D.W. Bassett, P.R. Webber, Diffusion <strong>of</strong> s<strong>in</strong>gle adatoms <strong>of</strong> plat<strong>in</strong>um,<br />

iridium and gold on plat<strong>in</strong>um <strong>surface</strong>s, Surf. Sci. 70 (1978) 520.<br />

[6] J.D. Wrigley, G. Ehrlich, Surface <strong>diffusion</strong> by an atomic exchange<br />

mechanism, Phys. Rev. Lett. 44 (1980) 661.<br />

[7] E.W. Müller, T.T. Tsong, Field Ion Microscopy Pr<strong>in</strong>ciples and<br />

Applications, American Elsevier, New York, 1969.<br />

[8] T. Halicioglu, An atomistic calculation <strong>of</strong> two-dimensional <strong>diffusion</strong> <strong>of</strong><br />

a Pt adatom on a Pt(110) <strong>surface</strong>, Surf. Sci. 79 (1979) L346.<br />

[9] T. Halicioglu, G.M. Pound, A calculation <strong>of</strong> the <strong>diffusion</strong> energies for<br />

adatoms on <strong>surface</strong>s <strong>of</strong> FCC metals, Th<strong>in</strong> Solid Films 57 (1979) 241.<br />

[10] G. DeLorenzi, G. Jacucci, V. Pontikis, <strong>in</strong>: D.A. Degras, M. Costa (Eds.),<br />

Proc. ICSS-4 and ECOSS-3, Cannes, 1980, p. 54.<br />

[11] G. DeLorenzi, G. Jacucci, V. Pontikis, Diffusion <strong>of</strong> adatoms and<br />

vacancies on otherwise perfect <strong>surface</strong>s: A molecular dynamics study,<br />

Surf. Sci. 116 (1982) 391.<br />

[12] M.R. Mruzik, G.M. Pound, A molecular dynamics study <strong>of</strong> <strong>surface</strong><br />

<strong>diffusion</strong>, J. Phys. F 11 (1981) 1403.<br />

[13] S.H. Gar<strong>of</strong>al<strong>in</strong>i, T. Halicioglu, Mechanism for the self-<strong>diffusion</strong> <strong>of</strong> Au<br />

and Ir Adatoms on Pt(110) <strong>surface</strong>, Surf. Sci. 104 (1981) 199.<br />

[14] J.D. Wrigley, Surface Diffusion by an Atomic Exchange Mechanism,<br />

Ph.D. Thesis, <strong>University</strong> <strong>of</strong> Ill<strong>in</strong>ois at Urbana-Champaign, 1982.<br />

[15] R.T. Tung, W.R. Graham, S<strong>in</strong>gle atom self-<strong>diffusion</strong> on nickel <strong>surface</strong>s,<br />

Surf. Sci. 97 (1980) 73.<br />

[16] R.T. Tung, Atomic Structure and Interactions at S<strong>in</strong>gle Crystal Metal<br />

Surfaces, Ph.D. Thesis, <strong>University</strong> <strong>of</strong> Pennsylvania, Philadelphia, 1981,<br />

171.<br />

[17] C.L. Liu, J.M. Cohen, J.B. Adams, A.F. Voter, EAM study <strong>of</strong> <strong>surface</strong><br />

self-<strong>diffusion</strong> <strong>of</strong> s<strong>in</strong>gle adatoms <strong>of</strong> fcc metals Ni, Cu, Al, Ag, Au, Pd,<br />

and Pt, Surf. Sci. 253 (1991) 334.<br />

[18] G.L. Kellogg, Field-ion microscope observations <strong>of</strong> <strong>surface</strong> self<strong>diffusion</strong><br />

and atomic <strong>in</strong>teractions on Pt, Microbeam Anal. (1986) 399.<br />

[19] C.L. Chen, T.T. Tsong, Self-<strong>diffusion</strong> on the reconstructed and<br />

nonreconstructed Ir(110) <strong>surface</strong>s, Phys. Rev. Lett. 66 (1991) 1610.<br />

[20] G.L. Kellogg, Direct observations <strong>of</strong> adatom–<strong>surface</strong>-atom replacement:<br />

Pt on Ni(110), Phys. Rev. Lett. 67 (1991) 216.<br />

[21] C.L. Chen, T.T. Tsong, L.H. Zhang, Z.W. Yu, Atomic replacement and<br />

adatom <strong>diffusion</strong>: Re on Ir <strong>surface</strong>s, Phys. Rev. B 46 (1992) 7803.<br />

[22] L. Pedemonte, R. Tatarek, G. Bracco, Surface self-<strong>diffusion</strong> at<br />

<strong>in</strong>termediate temperature: The Ag(110) case, Phys. Rev. B 66 (2002)<br />

045414 1.<br />

[23] P.A. Gravil, S. Holloway, Exchange mechanisms for self-<strong>diffusion</strong> on<br />

alum<strong>in</strong>ium <strong>surface</strong>s, Surf. Sci. 310 (1994) 267.<br />

[24] R. Stumpf, M. Scheffler, Ab <strong>in</strong>itio calculations <strong>of</strong> energies and self<strong>diffusion</strong><br />

on flat and stepped <strong>surface</strong>s <strong>of</strong> alum<strong>in</strong>um and their implications<br />

on crystal growth, Phys. Rev. B 53 (1996) 4958.<br />

[25] Y.-J. Sun, J.-M. Li, Self-<strong>diffusion</strong> mechanisms <strong>of</strong> adatom on Al(001),<br />

(011) and (111) <strong>surface</strong>s, Ch<strong>in</strong>ese Phys. Lett. 20 (2003) 269.<br />

[26] C.L. Liu, J.B. Adams, Diffusion mechanisms on Ni <strong>surface</strong>s, Surf. Sci.<br />

265 (1992) 262.<br />

[27] P. Stoltze, Simulation <strong>of</strong> <strong>surface</strong> defects, J. Phys.: Condens. Matter 6<br />

(1994) 9495.<br />

[28] U.T. Ndongmouo, F. Hont<strong>in</strong>f<strong>in</strong>de, Diffusion and growth on fcc(110)<br />

metal <strong>surface</strong>s: A computational study, Surf. Sci. 571 (2004) 891.<br />

[29] L. Hansen, P. Stoltze, K.W. Jacobsen, J.K. Nørskov, Self-<strong>diffusion</strong> on<br />

copper <strong>surface</strong>s, Phys. Rev. B 44 (1991) 6523.<br />

[30] C. Mottet, R. Ferrando, F. Hont<strong>in</strong>f<strong>in</strong>de, A.C. Levi, A Monte Carlo<br />

simulation <strong>of</strong> submonolayer homoepitaxial growth on Ag(110) and<br />

Cu(110), Surf. Sci. 417 (1998) 220.<br />

[31] G.A. Evangelakis, D.G. Papageorgiou, G.C. Kall<strong>in</strong>teris, C.E. Lekka, N.I.<br />

Papanicolaou, Self-<strong>diffusion</strong> <strong>processes</strong> <strong>of</strong> copper adatom on Cu(110)<br />

<strong>surface</strong> by molecular dynamics simulations, Vacuum 50 (1998) 165.<br />

[32] F. Montalenti, R. Ferrando, <strong>Jump</strong>s and concerted moves <strong>in</strong> Cu, Ag, and<br />

Au(110) adatom self-<strong>diffusion</strong>, Phys. Rev. B 59 (1999) 5881.<br />

[33] S. Durukanoglu, O.S. Trush<strong>in</strong>, T.S. Rahman, Effect <strong>of</strong> step–step<br />

separation on <strong>surface</strong> <strong>diffusion</strong> <strong>processes</strong>, Phys. Rev. B 73 (2006)<br />

125426.<br />

[34] L.S. Perk<strong>in</strong>s, A.E. DePristo, Self-<strong>diffusion</strong> <strong>of</strong> adatoms on fcc(110)<br />

<strong>surface</strong>s, Surf. Sci. 317 (1994) L1152.


60 G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61<br />

[35] F. Hont<strong>in</strong>f<strong>in</strong>de, R. Ferrando, A.C. Levi, Diffusion <strong>processes</strong> relevant to<br />

the epitaxial growth <strong>of</strong> Ag on Ag(110), Surf. Sci. 366 (1996) 306.<br />

[36] K.-D. Shiang, C.M. Wei, T.T. Tsong, A molecular dynamics study <strong>of</strong><br />

self-<strong>diffusion</strong> on metal <strong>surface</strong>s, Surf. Sci. 301 (1994) 136.<br />

[37] C.M. Chang, C.M. Wei, S.P. Chen, Model<strong>in</strong>g <strong>of</strong> Ir adatoms on Ir<br />

<strong>surface</strong>s, Phys. Rev. B 54 (1996) 17083.<br />

[38] M. Villarba, H. Jónsson, Diffusion mechanisms relevant to metal crystal<br />

growth: Pt/Pt(111), Surf. Sci. 317 (1994) 15.<br />

[39] U.T. Ndongmouo, F. Hont<strong>in</strong>f<strong>in</strong>de, R. Ferrando, Numerical study <strong>of</strong> the<br />

stability <strong>of</strong> (111) and (331) micr<strong>of</strong>acets on Au, Pt, and Ir(110) <strong>surface</strong>s,<br />

Phys. Rev. B 72 (2005) 115412.<br />

[40] M. Karimi, G. Vidali, I. Dal<strong>in</strong>s, Energetics <strong>of</strong> the formation and<br />

migration <strong>of</strong> defects <strong>in</strong> Pb(110), Phys. Rev. B 48 (1993) 8986.<br />

[41] G. DeLorenzi, G. Jacucci, The migration <strong>of</strong> po<strong>in</strong>t defects on bcc <strong>surface</strong>s<br />

us<strong>in</strong>g a metallic pair potential, Surf. Sci. 164 (1985) 526.<br />

[42] D.L. Price, K.S. S<strong>in</strong>gwi, M.P. Tosi, Lattice dynamics <strong>of</strong> alkali metals <strong>in</strong><br />

the self-consistent screen<strong>in</strong>g theory, Phys. Rev. B 2 (1970) 2983.<br />

[43] D.L. Price, Effects <strong>of</strong> a volume-dependent potential on equilibrium<br />

properties <strong>of</strong> liquid sodium, Phys. Rev. A 4 (1971) 358.<br />

[44] G.L. Kellogg, P.J. Feibelman, Surface self-<strong>diffusion</strong> on Pt(001) by an<br />

atomic exchange mechanism, Phys. Rev. Lett. 64 (1990) 3143.<br />

[45] C. Chen, T.T. Tsong, Displacement distribution and atomic jump<br />

direction <strong>in</strong> <strong>diffusion</strong> <strong>of</strong> Ir atoms on the Ir(001) <strong>surface</strong>, Phys. Rev. Lett.<br />

64 (1990) 3147.<br />

[46] G.L. Kellogg, Temperature dependence <strong>of</strong> <strong>surface</strong> self-<strong>diffusion</strong> on<br />

Pt(001), Surf. Sci. 246 (1991) 31.<br />

[47] G.L. Kellogg, A.F. Wright, M.S. Daw, Surface <strong>diffusion</strong> and adatom<strong>in</strong>duced<br />

substrate relaxations <strong>of</strong> Pt, Pd, and Ni atoms on Pt(001), J. Vac.<br />

Sci. Technol. A 9 (1991) 1757.<br />

[48] T.T. Tsong, C. Chen, Atomic replacement and vacancy formation and<br />

annihilation on iridium <strong>surface</strong>s, Nature 355 (1992) 328.<br />

[49] G.L. Kellogg, Surface <strong>diffusion</strong> <strong>of</strong> Pt adatoms on Ni <strong>surface</strong>s, Surf. Sci.<br />

266 (1992) 18.<br />

[50] A. Friedl, O. Schütz, K. Müller, Self-<strong>diffusion</strong> on iridium (100). A<br />

structure <strong>in</strong>vestigation by field-ion microscopy, Surf. Sci. 266 (1992) 24.<br />

[51] T.-Y. Fu, T.T. Tsong, Structure and <strong>diffusion</strong> <strong>of</strong> small Ir and Rh clusters<br />

on Ir(001) <strong>surface</strong>s, Surf. Sci. 421 (1999) 157.<br />

[52] R. Stumpf, M. Scheffler, Ab <strong>in</strong>itio calculations <strong>of</strong> energies and self<strong>diffusion</strong><br />

on flat and stepped <strong>surface</strong>s <strong>of</strong> alum<strong>in</strong>um and their implications<br />

on crystal growth, Phys. Rev. B 53 (1996) 4958.<br />

[53] J.-M. Li, P.-H. Zhang, J.-L. Yang, L. Liu, Theoretical study <strong>of</strong> adatom<br />

self-<strong>diffusion</strong> on metallic fcc{100} <strong>surface</strong>s, Ch<strong>in</strong>ese Phys. Lett. 14<br />

(1997) 768.<br />

[54] S. Valkealahti, M. Mann<strong>in</strong>en, Diffusion on alum<strong>in</strong>um-cluster <strong>surface</strong>s<br />

and the cluster growth, Phys. Rev. B 57 (1998) 15533.<br />

[55] O.S. Trush<strong>in</strong>, P. Salo, M. Alatalo, T. Ala-Nissila, Atomic mechanisms <strong>of</strong><br />

cluster <strong>diffusion</strong> on metal fcc(100) <strong>surface</strong>s, Surf. Sci. 482–485 (2001)<br />

365.<br />

[56] S. Ovesson, A. Bogicevic, G. Wahnstrom, B.I. Lundqvist, Neglected<br />

adsorbate <strong>in</strong>teractions beh<strong>in</strong>d <strong>diffusion</strong> prefactor anomalies on metals,<br />

Phys. Rev. B 64 (2001) 125423.<br />

[57] N.I. Papanicolaou, V.C. Papathanakos, D.G. Papageorgiou, Self<strong>diffusion</strong><br />

on Al(100) and Al(111) <strong>surface</strong>s by molecular-dynamics<br />

simulation, Physica B 296 (2001) 259.<br />

[58] T. Fordell, P. Salo, M. Alatalo, Self-<strong>diffusion</strong> on fcc(100) metal <strong>surface</strong>s:<br />

Comparison <strong>of</strong> different approximations, Phys. Rev. B 65 (2002)<br />

233408.<br />

[59] P.M. Agrawal, B.M. Rice, D.L. Thompson, Predict<strong>in</strong>g trends <strong>in</strong> rate<br />

parameters for self-<strong>diffusion</strong> on FCC metal <strong>surface</strong>s, Surf. Sci. 515<br />

(2002) 21.<br />

[60] C.M. Chang, C.M. Wei, Self-<strong>diffusion</strong> <strong>of</strong> adatoms and dimers on<br />

fcc(100) <strong>surface</strong>s, Ch<strong>in</strong>ese J. Phys. 43 (2005) 169.<br />

[61] P.G. Flahive, W.R. Graham, Pair potential calculations <strong>of</strong> s<strong>in</strong>gle atom<br />

self–<strong>diffusion</strong> activation energies, Surf. Sci. 91 (1980) 449.<br />

[62] D.E. Sanders, A.E. DePristo, Predicted <strong>diffusion</strong> rates on fcc(001) metal<br />

<strong>surface</strong>s for adsorbate/substrate comb<strong>in</strong>ations <strong>of</strong> Ni, Cu, Rh, Pd, Ag, Pt,<br />

Au, Surf. Sci. 260 (1992) 116.<br />

[63] K.-D. Shiang, Molecular dynamics simulation <strong>of</strong> adatom <strong>diffusion</strong> on<br />

metal <strong>surface</strong>s, J. Chem. Phys. 99 (1993) 9994.<br />

[64] L.S. Perk<strong>in</strong>s, A.E. DePristo, Self-<strong>diffusion</strong> mechanisms for adatoms on<br />

fcc(100) <strong>surface</strong>s, Surf. Sci. 294 (1993) 67.<br />

[65] L.S. Perk<strong>in</strong>s, A.E. DePristo, The <strong>in</strong>fluence <strong>of</strong> lattice distortion on atomic<br />

self-<strong>diffusion</strong> on fcc(001) <strong>surface</strong>s: Ni, Cu, Pd, Ag, Surf. Sci. 325 (1995)<br />

169.<br />

[66] G. Boisvert, L.J. Lewis, A. Yelon, Many-body nature <strong>of</strong> the Meyer-<br />

Neidel compensation law for <strong>diffusion</strong>, Phys. Rev. Lett. 75 (1995) 469.<br />

[67] Z.-P. Shi, Z. Zhang, A.K. Swan, J.F. Wendelken, Dimer shear<strong>in</strong>g as a<br />

novel mechanism for cluster <strong>diffusion</strong> and dissociation on metal (100)<br />

<strong>surface</strong>s, Phys. Rev. Lett. 76 (1996) 4927.<br />

[68] J. Merikoski, I. Vattula<strong>in</strong>en, J. He<strong>in</strong>onen, T. Ala-Nissila, Effect <strong>of</strong> k<strong>in</strong>ks<br />

and concerted <strong>diffusion</strong> mechanisms on mass transport and growth on<br />

stepped metal <strong>surface</strong>s, Surf. Sci. 387 (1997) 167.<br />

[69] H. Mehl, O. Biham, I. Furman, M. Karimi, Models for adatom <strong>diffusion</strong><br />

on fcc(001) metal <strong>surface</strong>s, Phys. Rev. B 60 (1999) 2106.<br />

[70] S.Y. Davydov, Calculation <strong>of</strong> the activation energy for <strong>surface</strong> self<strong>diffusion</strong><br />

<strong>of</strong> transition-metal atoms, Phys. Solid State 41 (1999) 8.<br />

[71] C.M. Chang, C.M. Wei, Self-<strong>diffusion</strong> <strong>of</strong> adatoms and dimers on<br />

fcc(100) <strong>surface</strong>s, Ch<strong>in</strong>ese J. Phys. 43 (2005) 169.<br />

[72] U. Kürpick, A. Kara, T.S. Rahman, Role <strong>of</strong> lattice vibrations <strong>in</strong> adatom<br />

<strong>diffusion</strong>, Phys. Rev. Lett. 78 (1997) 1086.<br />

[73] C.M. Chang, C.M. Wei, J. Hafner, Self-<strong>diffusion</strong> <strong>of</strong> adatoms on Ni(100)<br />

<strong>surface</strong>s, J. Phys.: Condens. Matter 13 (2001) L321.<br />

[74] P. Wynblatt, N.A. Gjoste<strong>in</strong>, A calculation <strong>of</strong> relaxation, migration and<br />

formation energies for <strong>surface</strong> defects <strong>in</strong> copper, Surf. Sci. 12 (1968)<br />

109.<br />

[75] J.E. Black, Z.-J. Tian, Complicated exchange-mediated <strong>diffusion</strong><br />

mechanisms <strong>in</strong> and on a Cu(100) substrate at high temperatures, Phys.<br />

Rev. Lett. 71 (1993) 2445.<br />

[76] J.M. Cohen, Long range adatom <strong>diffusion</strong> mechanism on fcc(100) EAM<br />

modeled materials, Surf. Sci. Lett. 306 (1994) L545.<br />

[77] C. Lee, G.T. Barkema, M. Breeman, A. Pasquarello, R. Car, Diffusion<br />

mechanism <strong>of</strong> Cu adatoms on a Cu(001) <strong>surface</strong>, Surf. Sci. Lett. 306<br />

(1994) L575.<br />

[78] C.-L. Liu, Energetics <strong>of</strong> <strong>diffusion</strong> <strong>processes</strong> dur<strong>in</strong>g nucleation and<br />

growth for the Cu/Cu(100) system, Surf. Sci. 316 (1994) 294.<br />

[79] M. Karimi, T. Tomkowski, G. Vidali, O. Biham, Diffusion <strong>of</strong> Cu on Cu<br />

<strong>surface</strong>, Phys. Rev. B 52 (1995) 5364.<br />

[80] P.V. Kumar, J.S. Raul, S.J. Warakomski, K.A. Fichthorn, Smart Monte<br />

Carlo for accurate simulation <strong>of</strong> rare-event dynamics: Diffusion <strong>of</strong><br />

adsorbed species on solid <strong>surface</strong>s, J. Chem. Phys. 105 (1996) 686.<br />

[81] G. Boisvert, L.J. Lewis, Self-<strong>diffusion</strong> <strong>of</strong> adatoms, dimers, and vacancies<br />

on Cu(100), Phys. Rev. B 56 (1997) 7643.<br />

[82] G.A. Evangelakis, N.I. Papanicolaou, Adatom self-<strong>diffusion</strong> <strong>processes</strong><br />

on (001) copper <strong>surface</strong>s by molecular dynamics, Surf. Sci. 347 (1996)<br />

376.<br />

[83] O.S. Trush<strong>in</strong>, K. Kokko, P.T. Salo, W. Hergert, M. Kotrla, Step<br />

roughen<strong>in</strong>g effect on adatom <strong>diffusion</strong>, Phys. Rev. B 56 (1997) 12135.<br />

[84] Q. Xie, Dynamics <strong>of</strong> adatom self-<strong>diffusion</strong> and island morphology<br />

evolution at a Cu(100) <strong>surface</strong>, Phys. Status Solidi 207 (1998) 153.<br />

[85] G. Boisvert, N. Mousseau, L.J. Lewis, Surface <strong>diffusion</strong> coefficients by<br />

thermodynamic <strong>in</strong>tegration: Cu on Cu(100), Phys. Rev. B 58 (1998)<br />

12667.<br />

[86] O. Biham, I. Furman, M. Karimi, G. Vidali, R. Kennett, H. Zeng, Models<br />

for <strong>diffusion</strong> and island growth <strong>in</strong> metal monolayers, Surf. Sci. 400<br />

(1998) 29.<br />

[87] J.B. Adams, Z. Wang, Y. Li, Model<strong>in</strong>g Cu th<strong>in</strong> film growth, Th<strong>in</strong> Solid<br />

Films 365 (2000) 201.<br />

[88] R. Pentcheva, Ab <strong>in</strong>itio study <strong>of</strong> microscopic <strong>processes</strong> <strong>in</strong> the growth <strong>of</strong><br />

Co on Cu(001), Appl. Phys. A 80 (2005) 971.<br />

[89] M.O. Jahma, M. Rusanen, A. Karim, I.T. Koponen, T. Ala-Nissila,<br />

T.S. Rahman, Diffusion and submonolayer island growth dur<strong>in</strong>g<br />

hyperthermal deposition on Cu(100) and Cu(111), Surf. Sci. 598 (2005)<br />

246.<br />

[90] H. Yildirim, A. Kara, S. Durukanoglu, T.S. Rahman, Calculated preexponential<br />

factors and energetics for adatom hopp<strong>in</strong>g on terraces and<br />

steps <strong>of</strong> Cu(100) and Cu(110), Surf. Sci. 600 (2006) 484.


G. Antczak, G. Ehrlich / Surface Science Reports 62 (2007) 39–61 61<br />

[91] J.E. Müller, H. Ibach, Migration <strong>of</strong> po<strong>in</strong>t defects at charged Cu, Ag, and<br />

Au (100) <strong>surface</strong>s, Phys. Rev. B 74 (2006) 085408.<br />

[92] P.J. Feibelman, R. Stumpf, Adsorption-<strong>in</strong>duced lattice relaxation and<br />

<strong>diffusion</strong> by concerted substitution, Phys. Rev. B 59 (1999) 5892.<br />

[93] A.V. Evteev, A.T. Kosilov, S.A. Solyanik, Atomic mechanisms and<br />

k<strong>in</strong>etics <strong>of</strong> self-<strong>diffusion</strong> on the Pd(001) <strong>surface</strong>, Phys. Solid State 46<br />

(2004) 2003.<br />

[94] B.D. Yu, M. Scheffler, Anisotropy <strong>of</strong> growth <strong>of</strong> the close-packed <strong>surface</strong>s<br />

<strong>of</strong> silver, Phys. Rev. Lett. 77 (1996) 1095.<br />

[95] B.D. Yu, M. Scheffler, Ab <strong>in</strong>itio study <strong>of</strong> step formation and self<strong>diffusion</strong><br />

on Ag(100), Phys. Rev. B 55 (1997) 13916.<br />

[96] R.C. Nelson, T.L. E<strong>in</strong>ste<strong>in</strong>, S.V. Khare, P.J. Reus, Energetics <strong>of</strong> steps,<br />

k<strong>in</strong>ks, and defects on Ag{100} and Ag{111} us<strong>in</strong>g the embedded atom<br />

method, and some consequences, Surf. Sci. 295 (1993) 462.<br />

[97] Z. Chvoj, C. Ghosh, T.S. Rahman, M.C. Tr<strong>in</strong>gides, Prefactors for<br />

<strong>in</strong>terlayer <strong>diffusion</strong> on Ag/Ag(111), J. Phys.: Condens. Matter 15 (2003)<br />

5223.<br />

[98] R.M. Lynden-Bell, Migration <strong>of</strong> adatoms on the (100) <strong>surface</strong> <strong>of</strong> facecentered-cubic<br />

metals, Surf. Sci. 259 (1991) 129.<br />

[99] G. Boisvert, L.J. Lewis, Self-<strong>diffusion</strong> on low-<strong>in</strong>dex metallic <strong>surface</strong>s:<br />

Ag on Au(100) and (111), Phys. Rev. B 54 (1996) 2880.<br />

[100] P.J. Feibelman, Diffusion path for an Al adatom on Al(100), Phys. Rev.<br />

Lett. 65 (1990) 729.<br />

[101] S.M. Foiles, M.I. Baskes, M.S. Daw, Embedded-atom-method functions<br />

for the fcc metals Cu, Ag, Ni, Pd, Pt, and their alloys, Phys. Rev. B 33<br />

(1986) 7983.<br />

[102] V. Rosato, M. Guillope, B. Legrand, Thermodynamical and structural<br />

properties <strong>of</strong> f.c.c. transition metals us<strong>in</strong>g a simple tight-b<strong>in</strong>d<strong>in</strong>g model,<br />

Philos. Mag. A 59 (1997) 321.<br />

[103] F. Cleri, V. Rosato, Tight-b<strong>in</strong>d<strong>in</strong>g potentials for transition metals and<br />

alloys, Phys. Rev. B 48 (1993) 22.<br />

[104] R. Ferrando, Correlated jump-exchange <strong>processes</strong> <strong>in</strong> the <strong>diffusion</strong> <strong>of</strong> Ag<br />

on Ag(110), Phys. Rev. Lett. 76 (1996) 4195.<br />

[105] W. Xiao, P.A. Greaney, D.C. Chrzan, Adatom transport on stra<strong>in</strong>ed<br />

Cu(001): Surface Crowdions, Phys. Rev. Lett. 90 (2003) 146102.<br />

[106] W. Xiao, P.A. Greaney, D.C. Chrzan, Pt adatom <strong>diffusion</strong> on stra<strong>in</strong>ed<br />

Pt(001), Phys. Rev. B 70 (2004) 033402.<br />

[107] J.C. Tully, G.H. Gilmer, M. Shugard, Molecular dynamics <strong>of</strong> <strong>surface</strong><br />

<strong>diffusion</strong>. I. The motion <strong>of</strong> adatoms and clusters, J. Chem. Phys. 71<br />

(1979) 1630.<br />

[108] A.F. Voter, J.D. Doll, Dynamical corrections to transition state theory<br />

for multistate systems: Surface self-<strong>diffusion</strong> <strong>in</strong> the rare-event regime, J.<br />

Chem. Phys. 82 (1985) 80.<br />

[109] A.F. Voter, Simulation <strong>of</strong> the layer-growth dynamics <strong>in</strong> silver films:<br />

Dynamics <strong>of</strong> adatom and vacancy clusters on Ag(100), Proc. SPIE 821<br />

(1987) 214.<br />

[110] U. Kürpick, Surface <strong>diffusion</strong> on (100), (110), and (111) <strong>surface</strong>s <strong>of</strong> Ni<br />

and Cu: A detailed study <strong>of</strong> prefactors and activation energies, Phys. Rev.<br />

B 64 (2001) 075418.<br />

[111] U. Kürpick, T.S. Rahman, Vibrational free energy contribution to self<strong>diffusion</strong><br />

on Ni(100), Cu(100) and Ag(100), Surf. Sci. 383 (1997) 137.<br />

[112] U. Kürpick, T.S. Rahman, Diffusion <strong>processes</strong> relevant to homoepitaxial<br />

growth on Ag(100), Phys. Rev. B 57 (1998) 2482.<br />

[113] J. Ferrón, L. Gómez, J.J. de Miguel, R. Miranda, Nonstochastic behavior<br />

<strong>of</strong> atomic <strong>surface</strong> <strong>diffusion</strong> on Cu(111) down to low temperatures, Phys.<br />

Rev. Lett. 93 (2004) 166107.<br />

[114] J.D. Wrigley, M.E. Twigg, G. Ehrlich, Lattice walks by long jumps, J.<br />

Chem. Phys. 93 (1990) 2885.<br />

[115] S.C. Wang, J.D. Wrigley, G. Ehrlich, Atomic jump lengths <strong>in</strong> <strong>surface</strong><br />

<strong>diffusion</strong>: Re, Mo, Ir, and Rh on W(211), J. Chem. Phys. 91 (1989)<br />

5087.<br />

[116] D.C. Senft, G. Ehrlich, Long jumps <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>: Onedimensional<br />

migration <strong>of</strong> isolated adatoms, Phys. Rev. Lett. 74 (1995)<br />

294.<br />

[117] D.C. Senft, Long <strong>Jump</strong>s <strong>in</strong> Surface Diffusion on Tungsten(211), Ph.D.<br />

Thesis, <strong>University</strong> <strong>of</strong> Ill<strong>in</strong>ois at Urbana-Champaign, Urbana, 1994, p.<br />

118.<br />

[118] D.C. Senft, Atomic jump length <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>: Experiment and<br />

theory, Appl. Surf. Sci. 94–95 (1996) 231.<br />

[119] T.R. L<strong>in</strong>deroth, S. Horch, E. Laegsgaard, I. Stensgaard, F. Besenbacher,<br />

Surface <strong>diffusion</strong> <strong>of</strong> Pt on Pt(110): Arrhenius behavior <strong>of</strong> long jumps,<br />

Phys. Rev. Lett. 78 (1997) 4978.<br />

[120] F. Montalenti, R. Ferrando, Compet<strong>in</strong>g mechanisms <strong>in</strong> adatom <strong>diffusion</strong><br />

on a channeled <strong>surface</strong>: <strong>Jump</strong>s versus metastable walks, Phys. Rev. B 58<br />

(1998) 3617.<br />

[121] H.T. Lorensen, J.K. Norskov, K.W. Jacobsen, Mechanism <strong>of</strong> self<strong>diffusion</strong><br />

on Pt(110), Phys. Rev. B 60 (1999) R5149.<br />

[122] S.-M. Oh, S.J. Koh, K. Kyuno, G. Ehrlich, Non-nearest-neighbor jumps<br />

<strong>in</strong> 2D <strong>diffusion</strong>: Pd on W(110), Phys. Rev. Lett. 88 (2002) 236102.<br />

[123] S.-M. Oh, K. Kyuno, S.J. Koh, G. Ehrlich, Atomic jumps <strong>in</strong> <strong>surface</strong> self<strong>diffusion</strong>:<br />

W on W(110), Phys. Rev. B 66 (2002) 233406.<br />

[124] G. Antczak, G. Ehrlich, Long jump rates <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>: W on<br />

W(110), Phys. Rev. Lett. 92 (2004) 166105.<br />

[125] G. Antczak, G. Ehrlich, Long jumps <strong>in</strong> <strong>diffusion</strong> <strong>of</strong> iridium on W(110),<br />

Phys. Rev. B 71 (2005) 115422.<br />

[126] G. Antczak, Long jumps <strong>in</strong> one-dimensional <strong>surface</strong> self-<strong>diffusion</strong>:<br />

Rebound transitions, Phys. Rev. B 73 (2006) 033406.<br />

[127] G. DeLorenzi, Dynamics <strong>of</strong> atom jumps on <strong>surface</strong>s,<br />

http://www.lanl.gov/orgs/t/tl/<strong>surface</strong> <strong>diffusion</strong>, 1989.<br />

[128] D.E. Sanders, A.E. DePristo, A non-unique relationship between<br />

potential energy <strong>surface</strong> barrier and dynamical <strong>diffusion</strong> barrier: fcc(111)<br />

metal <strong>surface</strong>, Surf. Sci. 264 (1992) L169.<br />

[129] G. Antczak, K<strong>in</strong>etics <strong>of</strong> atom rebound<strong>in</strong>g <strong>in</strong> <strong>surface</strong> <strong>diffusion</strong>, Phys. Rev.<br />

B 74 (2006) 153406.<br />

[130] S.C. Wang, G. Ehrlich, Diffusion <strong>of</strong> large <strong>surface</strong> clusters: Direct<br />

observations on Ir(111), Phys. Rev. Lett. 79 (1997) 4234.<br />

[131] S.C. Wang, U. Kürpick, G. Ehrlich, Surface <strong>diffusion</strong> <strong>of</strong> compact and<br />

other clusters: Ir x on Ir(111), Phys. Rev. Lett. 81 (1998) 4923.<br />

[132] M. Schunack, T.R. L<strong>in</strong>deroth, F. Rosei, E. Laegsgaard, I. Stensgaard,<br />

F. Besenbacher, Long jumps <strong>in</strong> the <strong>surface</strong> <strong>diffusion</strong> <strong>of</strong> large molecules,<br />

Phys. Rev. Lett. 88 (2002) 156102.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!