1. Introduction
The interest in vicinal crystal surfaces, whose planes form a small angle with a low-index plane (e.g., (001), (111) or (110) in cubic crystals), goes much beyond pure crystallography. Ideal vicinal surfaces are characterized by sets of equally-spaced steps separated by low-index terraces. Their structure and stability with respect to faceting, reconstruction and roughening transitions have been the subject of several experimental studies with He atom scattering (HAS) and related theoretical works [
1,
2,
3,
4,
5,
6,
7,
8,
9,
10,
11,
12,
13,
14,
15,
16,
17,
18,
19,
20]. The early interest in vicinal surfaces was motivated by their expected role as natural templates for the epitaxial growth of functional nanostructures [
21,
22,
23,
24], and in heterogeneous catalysis [
25]. The topological and the quasi-one-dimensionality (quasi-1D) features of steps are argued to be reflected in their electronic and vibrational properties, hence in the local electron-phonon (el-ph) interaction, thus opening new horizons for quasi-1D superconductivity in topological materials [
26,
27,
28,
29].
In a previous work [
30], it has been shown that helium-atom scattering (HAS) can provide a direct information on the surface el-ph interaction at a conducting surface, as expressed by the total mass-enhancement factor
λ, via the temperature- or incident-momentum-dependence of the HAS Debye-Waller (DW) exponent. The value of
λ derived in this way (hereafter denoted
λHAS) has been reported for several metal surfaces [
31] and overlayers [
32], graphene [
33], topological semimetals [
34,
35,
36,
37], layered dichalcogenides [
38,
39], 2D superconductors [
40], and low- and multi-dimensional surfaces [
41]. In this paper we extend this study to high-index metal surfaces, by re-analysing previous HAS diffraction measurements on the vicinal surfaces of copper Cu(11α) (α = 3,5,7 [
42,
43], α = 2,5 [
44,
45,
46] (
Figure 1, left)) and aluminium Al(221) and Al(332) [
46,
47] (
Figure 1, right). It is shown that the DW exponent -2
W(
T,ki), whether as a function of the surface temperature
T or of the incident He beam wavevector
ki, carries distinct information about the el-ph interaction associated with either steps or terraces. Lapujoulade et al. [
42,
43] observed that the typical linear slope of -2
W(
T,ki) for increasing temperature Cu(11α),
, becomes suddenly steeper above a certain temperature
TR, of the order or above room temperature, indicative of a surface roughening transition. In the light of the el-ph theory of the DW exponent, it is shown in
Section 3 that this kind of roughening transition actually yields an increase of the local el-ph interaction, similarly to what recently reported for a semiconductor surface [
48].
2. Theory
The specular HAS intensity is written as a function of the incident wavevector
ki and surface temperature
T in the form
The wavevector dependence of the prefactor
is conveniently described at small
ki by a power law,
, where
η (
> 1 ) depends on the atom-surface interaction as well as on the supersonic beam source design and operating conditions [
30,
50].
The present el-ph theory of HAS from conducting surfaces [
30,
31] links the surface el-ph mass-enhancement factor
λHAS to the dependence of the DW exponent on temperature or incident momentum through the two equations
respectively. Here
φ is the surface work function,
ns the effective number of surface atomic layers involved in the surface el-ph interaction, and
ac the surface unit cell area. For vicinal surfaces,
ac is here approximated by that of the terraces between two neighbour parallel steps. The power law for
ensures that the measured
at a given surface temperature
T vanishes for both
ki → 0 and
ki → ∞. It will therefore have a maximum at
ki =
ki, max, where
By combining this condition with Equations (1) and (3), the latter can be expressed as:
Note that, by inserting Equation (4) into (3),
can be obtained from:
and is therefore dependent on the surface temperature.
As discussed in Ref. [
30], when the effects of the attractive surface potential depth
D are not completely negligible in specular HAS, they can be approximately accounted for by replacing
in Equation (2) with
(Beeby correction [
51]), where
m is the He atom mass and the potential depth
D can directly be obtained from the analysis of HAS bound-state resonances [
52]. Classically, the Beeby correction is equivalent to assuming that the He-surface interaction occurs at the maximum speed acquired by the He atom under the effect of the attractive potential. When the
ki—dependence of the HAS intensity is used instead of its temperature dependence to derive
λHAS, Equation (4) has the advantage that the information on the electron-phonon interaction is all within the factor
, to which the Beeby correction can be directly added.
3. The Copper Vicinal Surfaces Cu(11α)
A selection of HAS specular reflectivity data measured by Lapujoulade et al. [
42,
43] as a function of the surface temperature for the vicinal surfaces of copper (11α), with α = 3,5,7, is reproduced in
Figure 2 (left) and compared with the data for the low-index surfaces (110), (111) and (001) (α = 0,1,∞) (right). The reflectivity data, normalized to the extrapolated zero-temperature value
I(
ki,0), are plotted on a logarithmic scale so as to give the profile of the DW exponent, with the corresponding He beam incident wavevectors
ki and angles
θi indicated in the panels. The HAS experimental points have been fitted by Lapujoulade et al. (full lines) [
42,
43] with a theory for the DW temperature dependence based on a He-surface atom phenomenological potential, the surface dynamics represented by a surface Debye temperature of the order of 230 K, and adjustable parameters for the Beeby correction, for the so-called Armand effect, and for anharmonic corrections, the latter in order to account for the high-temperature deviations from the expected linearity.
All this appears to be largely insufficient to explain the sudden increase of slope observed above some temperature
TR of the order of room temperature for Cu(117), or larger for Cu(115), Cu(113) and even Cu(110), while the fit is excellent for the densely-packed surfaces (111) and (001), which show a regular behaviour up the highest measured temperature of 800 K. This has led Lapujoulade et al. [
8,
9] to consider these data as a clear evidence for a roughening transition, promoted by, and affecting first, the step rows, with
TR its critical temperature.
The values of
λHAS derived by means of Equation (2) from the data of
Figure 2 below (light face) and above (bold face) the roughening transition are given in
Table 1, with the input data taken from the indicated references. The work function
φ = 4.53 eV measured by Gartland et al. [
49] for the Cu(112) has been adopted for the other vicinal surfaces. This is however very close to that of Cu(001), which actually is the terrace plane for α = 3,5,7. The Cu(001) terrace values are then used also for
ns, and
ac.
Below the roughening transition the intensity slope used in the calculation is that of the interpolating line calculated by Lapujoulade et al. shown in
Figure 2 [
30] and including anharmonic corrections, while above the transition the experimental intensity ratio at the two temperatures
T1 and
T2 given in
Table 1 has been used. No evidence of any roughening was found for the closed-packed surface Cu(111). For this surface two different sets of data from Ref. [
30] are reproduced in
Figure 2, corresponding to the different incident angles. Although the slopes are rather different, the corresponding values of
λHAS given in
Table 1 are about the same, as expected. For the Cu(001) surface, two different values of
λHAS are associated with the slight increase of the intensity slope observed with respect to the fitting curve, although this small effect can be hardly associated to a real transition rather than to a gradual increase of surface disorder.
An interesting result from this analysis is the appreciable increase of the electron-phonon interaction above the roughening transition. As anticipated above, this appears to be consistent with what was recently reported for a CdTe surface [
48], although the mechanism for the electron-phonon interaction at a roughened metal surface may be rather different from that for an intrinsic semiconductor surface. It should be noticed that disorder activates additional scattering channels at the expense of ordinary specular and diffraction channels. Thus, a steeper decay of the DW exponent observed at increasing temperature above the disorder threshold may be in part an effect of opening new competitive scattering channels, rather than an indication of larger electron-phonon interaction. On the other hand, below that threshold the observed increase of
λHAS has a solid basis, relying on the localization of both electronic and phonon excitations, conferring a quasi-one-dimensional character to step dynamics. Phonon softening at steps [
59] may be viewed as a manifestation of a larger local electron-phonon interaction.
The HAS angular distributions reported by Miret-Artés et al. [
45] for the Cu(112) and the Cu(115) vicinal surfaces, respectively measured along the
and
directions, give the specular intensity from the vicinal plane (
θi = 45°) for seven different values of the incident
ki at the same temperature of about 130 K. The measured specular intensities, plotted in
Figure 3, give indication of a maximum value
around 15 Å
-2 for both surfaces [
45]. For the (115) surface the fit of the specular intensity with the function
where the exponential represents the DW factor and
A and
C are two fitting constants, suggests
η = 2 with
Å
-1 rather than
η = 1 with
Å
-1. For the (112) surface the dip in the intensity at
ki = 7.5 Å
-1 is presumably due to a bound-state resonance, and does not allow to clearly distinguish between the fits with the two values of
η and the respective values of
ki,max given in
Table 2. The corresponding values of
λHAS for the two surfaces Cu(112) and Cu(115) at
T = 130 K, obtained from Equation (4) for both
η = 1 and 2, are given in
Table 2. Clearly the values of
λHAS increase with
η . The quality of the fits in
Figure 3 for Cu(115) would suggest
η = 2, but the resulting values of
λHAS , even with
η = 1, turn out to be systematically larger than those obtained from the temperature dependence of the DW exponent, and more consistent with the values found above the roughening transition temperature.
It should be noted that the terraces of the Cu(112) surfaces are (111) surfaces, while the terraces of the other Cu(11α
) surfaces (α=3,5,7) are (001) planes. For vicinal surfaces with a short inter-step period and a densely packed terraces like Cu(112) the actual crystallographic unit cell area (
ac = 15.9 Å
2) should be a better choice. In this case
λHAS drops to 0.10 for
η = 1 (close to that of Cu(001) and Cu(111)) and 0.16 for
η = 2 (
Table 2).
It appears, however, as a general fact that
λHAS increases in moving from high-index to intermediate vicinal surfaces. The largest
λHAS of the Cu(11α) series derived from the DW temperature-dependence is found for α = 5. Such an increase of the surface el-ph interaction can be associated to the presence of steps, as long as they are sufficiently localized but not too far apart. Localization of the surface electronic states at the Fermi level can also stay behind the increase of
λHAS above the roughening transition.
Table 1.
shown in
Figure 2 (from ref. [
42]) for ordered (low-temperature interval) and roughened (high-temperature interval, bold face figures) Cu(11α), Al(221) and Al(332) vicinal surfaces. No roughening transition has been observed for Cu(111). For comparison, the mass-enhancement factor reported for Al(111) is
λHAS = 0.30 [
30].
Table 1.
shown in
Figure 2 (from ref. [
42]) for ordered (low-temperature interval) and roughened (high-temperature interval, bold face figures) Cu(11α), Al(221) and Al(332) vicinal surfaces. No roughening transition has been observed for Cu(111). For comparison, the mass-enhancement factor reported for Al(111) is
λHAS = 0.30 [
30].
Table 2.
Electron-phonon mass-enhancement factor λHAS derived from the dependence of the specular HAS intensity on the incident wavevector (Eqs. (3,4)). Values equal to those of the line above are omitted.
Table 2.
Electron-phonon mass-enhancement factor λHAS derived from the dependence of the specular HAS intensity on the incident wavevector (Eqs. (3,4)). Values equal to those of the line above are omitted.
Surface |
φ |
ns |
ac |
ki,max
|
D |
T |
η |
λHAS
|
|
[eV] |
|
[Å21 |
[Å-1] |
[meV] |
[K] |
|
|
Cu(112) a
|
4.53 b
|
6.8 c
|
5.64 d
|
~5.0 |
8.25 e
|
130 |
1 |
0.29 |
|
|
|
|
~6.5 |
|
|
2 |
0.45 |
|
|
|
15.9 i |
~5.0 |
8.25 e
|
130 |
1 |
0.10 |
|
|
|
|
~6.5 |
|
|
2 |
0.16 |
Cu(115) a
|
4.53 b |
6.8 c
|
6.52 d |
~5.5 |
6.35 e
|
130 |
1 |
0.26 |
|
|
|
|
~7.0 |
|
|
2 |
0.38 |
Al(221) f
|
4.26 h
|
1.6 c
|
7.09 d
|
~6.5 |
7.0 e
|
135 |
1 |
0.72 |
Al(332) g
|
4.26 h
|
1.6 c
|
7.09 d
|
~7.2 |
7.0 e
|
130 |
1 |
0.61 |
4. The Aluminium Vicinal Surfaces Al(221) and Al(332)
The HAS intensity at a scattering angle
= 91.5º has been measured for the vicinal surfaces Al(221) [
46,
54,
55] and Al(332) [
46,
47,
55], both as a function of the surface temperature and incident wavevector.
Figure 4a–c reproduces some of the HAS spectra from the Al(221) surface measured by Witte et al. [
54,
55] at three different surface temperatures and given incident wavevector
ki = 6.2 Å
-1. They are plotted as functions of the parallel wavevector change
in the direction
normal to the steps. The specular scattering from the crystallographic surface occurs at
= 0, while the specular scattering from terraces occurs at about ±2.4 Å
-1, the sign corresponding to either the up-hill or down-hill scattering configuration. Panel (d) shows the DW exponent derived from the full set of data points reported by Witte et al. and a
fit (red full line), where
θD = 790 K is the aluminium Debye temperature [
58]. The corresponding
λHAS from Equation (2) and the parameters listed in
Table 1, when referred to the approximate linear behaviour in the temperature interval 232-550 K, is 0.71. The experimental point at 712 K clearly deviates from the fitting curve, with the larger DW slope in the high-
T range giving
λHAS = 1.33. Such a large increase above 550 K could be interpreted as an effect of surface roughening, similar to what is observed in Cu(11α). Clearly, more measurements in this range should be made available in order to confirm this interpretation.
The value of
λHAS found for Al(221) below 550 K is, however, much larger than the one found for the Al(111) surface (
λHAS = 0.30, see
Table 2 of Ref. [
30]), as well as the bulk value
λ = 0.43 ± 0.05 [
57], but it almost coincides with the value of
λHAS = 0.72 derived via Equations (3,4) from the wavevector dependence of the specular HAS intensity at
T = 135 K and for
η = 1 (
Figure 5). It may be argued that also for the vicinal surface Al(221) a robust increase of the el-ph interaction occurs with respect to that of the low-index surfaces, with a further increase above 550 K as a possible effect of roughening. Note that in Al(221) the specular scattering from the crystallographic (221) surface is still larger than (or comparable to) that from the (111) terraces (small arrows in
Figure 4 and
Figure 5), despite the fairly large inter-step distance of 8.74 Å.
The vicinal surface Al(332) (
Figure 6), with its 14.30 Å inter-step distance, exhibits instead a dominant HAS specular scattering from (111) terraces (
, see
Figure 1), and only small sharp peaks from the periodic array of steps of the (332) surface (
= 0).
Figure 6, adapted from Lock
et al. [
46,
47], shows the changes of the HAS specular intensities from both terraces and step arrays observed by changing either temperature at a given
ki (
Figure 6(a)), or
ki at a given temperature (
Figure 6(b)). In the first case, Equation (2), applied to scattering from terraces with the parameters listed in
Table 1 and the temperature interval 308–606 K,
λHAS = 0.61. The same value is obtained in the second case from Eqs. (3-5), with
η = 1 and the
ki interval 6.61-10.37 Å
-1. The same consistency was found for Al(221), which strongly supports the choice
η = 1.
It is noted that
λHAS for Al(332), although still larger than that of Al(111), turns out to be smaller than that found for Al(221). An interesting question is whether this is due to the different sources of data used for Al(332) (specular HAS from (111) terraces) and Al(221) (specular HAS from the (221) surface, actually from the periodic array of steps). The temperature- and incident wavevector-dependence of the small peaks at
= 0 in
Figure 6 permits to compare what amounts to terraces and those of steps and to assess the respective contributions to the surface electron-phonon coupling. The second row of Al(332) data in
Table 1 shows indeed that
λHAS derived from the temperature dependence of the small
= 0 peak is increased to 1.10, which brings the surface electron-phonon coupling of Al(223) above that of Al(221) and much above that of the Al(111) low-index surface. A similar increase can probably be extracted from the
ki—dependence of
= 0 peak, although the very small peak at the higher wavevector is hidden in the background.
The temperature dependence of the elastic scattering spectra from Al(223) terraces measured by Lock et al. [
46,
47] around
for a fixed incident wavevector
ki = 4.93 Å
-1, reproduced in
Figure 7, can hardly be used to extract
λHAS due to the partial superposition to an additional peak growing with increasing temperature at positive values of
. This has been attributed to a temperature-driven instability towards faceting at steps. Nevertheless, the value of
λHAS = 0.66 obtained from Equation 2 in the lowest temperature interval (414-494 K) (see
Table 1, 3rd row for Al(332)), where the specular scattering intensity from (111) terraces is still dominant, is consistent with the terrace values derived from
Figure 6. At larger temperatures, where faceting instability occurs, it can make sense to consider the areas of the double peak features and to take the logarithm of the ratio of two areas measured at two different temperatures, in order to have an overall qualitative information on the electron-phonon interaction. The 4th row for Al(332) in
Table 1 gives
λHAS as obtained in this way over the entire 414-711 K temperature range. The comparatively large value of
λHAS = 1.42 is consistent with what found at the higher temperature for Al(221) (1.33) and attributed to a roughening transition, in analogy to what was reported for copper vicinal surfaces.
5. Conclusions
The method of extracting the electron-phonon interaction (mass enhancement factor
λHAS) at conducting surfaces from the Debye-Waller exponent of helium-atom scattering [
30], has been applied to the analysis of existing HAS data on Cu and Al vicinal surfaces. It is found that the electron-phonon interaction of the vicinal surfaces analysed in the present work is generally larger than that of the corresponding low-index surface of the terraces between two neighbouring steps. Such increase of
λHAS is attributed to the specific contribution of steps. The case of Al(332) is illuminating in this respect, because it permits to compare
λHAS obtained from the HAS reflectivity of (111) terraces to that obtained from the HAS reflectivity of the crystallographic surface (332).
More intriguing is the pronounced increase of
λHAS at higher temperature where some form of disorder appears such as roughening or terrace faceting. It has been noticed above that disorder activates additional scattering channels, which subtract intensity to specular and diffraction scattering, producing a steeper decay with temperature of the DW exponent. It is expected that disorder induced by increasing temperature starts from steps were atoms are more loosely bound at their lattice positions. On the other hand, the larger contributions of step atoms to
λHAS than that of terrace atoms, as found in the present analysis, is likely to be due to the step localization of both electronic and phonon excitations, which acquire a quasi-one-dimensional character, so that the phonon softening at steps [
59] may be linked to the larger local electron-phonon interaction.
The present work has been stimulated by existing HAS measurements which were not intended for the derivation of λHAS, the method being unknown at those times. There have been very interesting HAS investigations also on the clean and Fe-covered Pt(997)
surfaces [
60] where the normalization of HAS diffraction spectra measured at several different temperatures does not permit the evaluation of
λHAS. We hope that this study will stimulate new HAS studies explicitly designed for the derivation of both
λHAS and the mode-selected
λQν. The recent studies on the possible superconductivity enhancement as an effect of surface disorder [
61] offer further good reasons for new HAS studies specifically aiming at the derivation of the electron-phonon interaction at stepped surfaces and conducting surfaces with various types of disorder.