1: \documentclass[aps,prb,twocolumn,showpacs]{revtex4}
2: \usepackage{graphicx}
3: %\documentstyle[multicol,aps,epsfig]{revtex}
4:
5: \begin{document}
6:
7: \title{Hydrogen-assisted distortion of gold nanowires}
8:
9: \author{A.~Halbritter, Sz.~Csonka, and G.~Mih\'aly}
10: \affiliation{Electron Transport Research Group of the Hungarian
11: Academy of Science and Department of Physics, Budapest University
12: of Technology and Economics, 1111 Budapest, Hungary}
13:
14: \author{O.I.~Shklyarevskii, S.~Speller, and H.~van~Kempen}
15: \affiliation{NSRIM, University of Nijmegen, Toernooiveld 1, 6525
16: ED Nijmegen, the Netherlands}
17:
18: \date{\today}
19:
20: \begin{abstract}
21: In this paper the influence of adsorbed hydrogen on the behavior
22: of gold nanojunctions is investigated. It is found, that the
23: hydrogen environment has a strong effect on the conductance of
24: atomic-sized gold junctions, which is markedly reflected by the
25: growth of an additional peak in the conductance histogram at $0.5$
26: quantum unit, $2e^2/h$. The statistical analysis of the
27: conductance traces implies that the new peak arises due to the
28: hydrogen-assisted distortion of atomic gold chains.
29: \end{abstract}
30:
31: \pacs{73.63.Rt, 73.23.Ad, 85.65.+h}
32:
33: \maketitle
34:
35: \section{Introduction}
36:
37: The conductance histograms of different metals have been studied
38: under various experimental conditions.\cite{Agrait2003} It was
39: found, that the peaks in the histogram arise due to the repeated
40: establishment of some stable atomic configurations during the
41: breaking process. Usually the general shape of the histogram is
42: characteristic of the material, and it is not very sensitive to
43: the experimental parameters, like the temperature or the magnitude
44: of the bias voltage.
45:
46: On the other hand, the interaction of adsorbates with the junction
47: can influence the evolution of the nanocontact during the rupture
48: and can cause drastic changes in the shape of the histogram. It
49: was shown that the adsorption of some organic molecules on copper
50: nanojunctions results in the appearance of new peaks around $0.5$
51: and $1.5$ quantum conductance unit, G$_0=2e^2/h$.\cite{Li2000}
52: Similar phenomenon was observed using gold nanowires immersed in
53: an electrochemical cell.\cite{Shu2000} By changing the potential
54: on the nanoconstriction from positive to negative with respect to
55: the reference electrode, the authors observed the appearance and
56: growth of two new peaks, again at $0.5$ and $1.5$\,G$_0$.
57:
58: These studies have shown that the conductance histograms can serve
59: as sensitive detectors of adsorbates. The new peaks in the
60: histogram are assumably caused by the appearance of new, stable
61: atomic arrangements in the modified environment. On the other
62: hand, the conductance histograms alone cannot tell the microscopic
63: background of the phenomenon. The above systems are rather
64: complicated for theoretical calculations, so the origin of the
65: observations has remained an open question.
66:
67: In this paper we show that similar phenomenon can arise due to the
68: adsorption of hydrogen molecules on gold nanowires. Hopefully, the
69: simplicity of this system will open the opportunity for
70: theoretical studies. We also show, that the detailed statistical
71: investigation of the conductance traces supplies useful
72: information about the nature of the new atomic configurations.
73:
74: Quite recently, the influence of hydrogen molecules on platinum
75: junctions was also studied.\cite{Smit2002} In this system the
76: characteristic peak of platinum at $1.5$\,G$_0$ disappears, and a
77: new peak grows in the histogram at $1$\,G$_0$. The detailed
78: studies have shown, that the new peak is connected to the
79: conduction through a H$_2$ bridge between the Pt electrodes. The
80: comparison of these results with our observations is
81: presented.
82:
83: \section{Basic observation: the appearance of a new peak in the conductance
84: histogram}
85:
86: \begin{figure}[t!]
87: \centering
88: \includegraphics[width=\columnwidth]{AuH2hist.eps}
89: \caption{\it Conductance histograms of gold measured in an
90: ultra-high vacuum environment (a) and in the presence of adsorbed
91: hydrogen (b). Both histograms were recorded at a bias voltage of
92: $129$\,mV and a temperature of $20$\,K. The histograms were built
93: from $10000$ conductance traces.} \label{AuH2hist.fig}
94: \end{figure}
95:
96: We have performed our measurements on high-purity gold samples
97: with the mechanically controllable break junction (MCBJ)
98: technique. The experiments were done under cryogenic circumstances
99: in the temperature range of $4.2-50$\,K. A typical conductance
100: histogram of gold measured in ultra high vacuum at $T=20$\,K is
101: presented in Fig.~\ref{AuH2hist.fig}(a). It agrees with earlier
102: reported data obtained under similar conditions.\cite{Yanson2001}
103: The most pronounced feature of this histogram is a sharp peak at
104: $1$\,G$_0$. The introduction of hydrogen into the vacuum
105: pot\cite{note1} resulted in the appearance of an additional peak
106: positioned close to $0.5$\,G$_0$, as shown in
107: Fig.~\ref{AuH2hist.fig}(b). In some cases a small peak at $\sim
108: 1.5$\,G$_0$ was also visible, but due to its eventual occurrence
109: we focus our attention on the peak at $\sim 0.5$\,G$_0$.
110:
111: This basic observation was studied in detail by changing three
112: experimental parameters: the temperature, the bias voltage and the
113: amount of H$_2$ near the contact. Even though the quantity of gas
114: admitted into the vacuum pot was accurately determined, no
115: reliable estimation for the H$_2$ coverage of the junction can be
116: given due to the presence of different materials and unavoidable
117: temperature gradients in the sample holder. Fortunately, the
118: results of the measurements were not sensitive to the precise
119: amount. In contrast, the other two parameters played a rather
120: crucial role and the new peak in the histogram was only observed
121: in a restricted range of the temperature and the bias voltage.
122:
123: The study of the temperature dependence showed that the fractional
124: peak is only present in the range of approximately $10-30$\,K. The
125: bias dependence of the conductance histograms measured at a fixed
126: temperature ($20$\,K) is shown in Fig.~\ref{AuH2hist2.fig}. At
127: bias voltages $\lesssim 100$\,mV the peaks of the histogram are
128: superimposed on a large featureless background
129: [Fig.~\ref{AuH2hist2.fig}(a)]. At higher bias, both the background
130: and the relative amplitude of the peak at $0.5$\,G$_0$ compared to
131: the one at $1$\,G$_0$ decrease [Fig.~\ref{AuH2hist.fig}(b)]; and
132: finally above $\sim 200$\,mV the peak at $0.5$\,G$_0$ completely
133: disappears [Fig.~\ref{AuH2hist2.fig}(b)]. The absence of the new
134: peak at elevated temperature or bias voltage is presumably caused
135: by the desorption of hydrogen from the surface of the junction.
136: These two effects can have the same origin: at elevated bias the
137: voltage-induced heating of the junction\cite{Halbritter2002}
138: causes the desorption. At low temperatures the vapor pressure of
139: hydrogen is very small, so the amount of hydrogen near the contact
140: is strongly reduced.
141:
142: \begin{figure}[b!]
143: \centering
144: \includegraphics[width=\columnwidth]{AuH2hist2.eps}
145: \caption{\it Bias dependence of the conductance histograms of gold
146: measured at $T=20$\,K in hydrogen environment. The bias voltages
147: are $65$\,mV (a) and $240$\,mV (b). The histograms were built from
148: $10000$ conductance traces.} \label{AuH2hist2.fig}
149: \end{figure}
150:
151: The same measurements were performed on silver and copper
152: contacts. In high vacuum the histograms of these noble metals are
153: resembling the one of gold.\cite{Yanson2001} The inclusion of
154: hydrogen resulted in a similar behavior with respect to the
155: appearance of a featureless background at low bias voltage. On the
156: other hand, in Ag and Cu no indication for the appearance of a new
157: peak was observed.
158:
159: Summarizing the experimental results, the addition of H$_2$ to the
160: contact area has two basic observable consequences: (1) the growth
161: of a large background signal at low bias voltage which is present
162: for all the materials investigated; (2) the appearance of a new
163: peak at $0.5$\,G$_0$ in the conductance histogram of gold.
164:
165: The first phenomenon is presumably caused by a large amount of
166: H$_2$ between the electrodes. Due to the adsorption of hydrogen,
167: the surface energy of the junction is reduced. From this reason, a
168: larger amount of atomic configurations might occur during the
169: break. It is also possible that an increased tunnel current can
170: flow through the hydrogen layers.
171:
172: The second phenomenon -- which is in the focus of the present
173: study -- indicates the repeated establishment of a special atomic
174: configuration during the last stages of the break. This atomic
175: configurations arises or becomes stable due to the surrounding
176: hydrogen molecules. The analysis of the conductance traces can
177: provide information about the nature of this new atomic
178: configuration, as discussed in the following.
179:
180: \section{The nature of the conductance traces}
181:
182: In pure gold contacts the conductance plateau at $1$\,G$_0$ is a
183: robust feature. The conductance traces contain this plateau almost
184: without exception. Furthermore, this is the last plateau, after
185: which the contact breaks. This plateau is very flat, very long,
186: and usually it is precisely positioned at $1$\,G$_0$. These
187: features are reflected by the exceptionally sharp peak in the
188: conductance histogram [Fig.~\ref{AuH2hist.fig}(a)]. There is some
189: chance for having smaller conductance values down to $0.6$\,G$_0$,
190: but below that there are absolutely no counts in the histogram.
191:
192: \begin{figure}[b!]
193: \centering
194: \includegraphics[width=\columnwidth]{AuH2trace.eps}
195: \caption{\it Conductance traces of hydrogen-covered gold
196: junctions. Panel (a) shows a trace with a long plateau at
197: $1$\,G$_0$ followed by a shorter plateau at $0.5$\,G$_0$. The
198: traces in panel (b) show telegraph fluctuation between $1$\,G$_0$
199: and $0.5$\,G$_0$.} \label{AuH2traces.fig}
200: \end{figure}
201:
202: The appearance of a new peak around $0.5$\,G$_0$ raises the
203: question, how these strong features of the conductance traces of
204: pure gold are modified in the presence of hydrogen. Naturally, the
205: traces must contain plateaus both at $0.5$\,G$_0$ and at
206: $1$\,G$_0$. However, the conductance histogram alone cannot tell
207: how these plateaus are related to each other. Figure
208: \ref{AuH2traces.fig}(a) shows a conductance trace extracted from
209: the data set. It has plateaus at both conductance values: first
210: the conductance stays at $1$\,G$_0$, then it jumps to
211: $0.5$\,G$_0$, and finally the contact breaks. This kind of
212: behavior was frequent, but not exclusive. Another two examples are
213: presented in Fig.~\ref{AuH2traces.fig}(b). These traces show a
214: telegraph fluctuation between $0.5$\,G$_0$ and $1$\,G$_0$. This
215: telegraph fluctuation implies that the contact can choose from two
216: metastable atomic configurations. The one with unit conductance is
217: assumably the customary atomic arrangement of pure gold. The other
218: one with $G=0.5$\,G$_0$ is another -- yet unknown -- configuration
219: which is only observable if hydrogen molecules are also present.
220: It should be noted that plateaus at $0.5$\,G$_0$ are not seen in
221: each trace, several conductance curves look like those in pure
222: gold contacts. On the other hand, traces with plateau at
223: $0.5$\,G$_0$ and no plateau at $1$\,G$_0$ were not typical.
224:
225: The plateaus are not precisely placed at $1$\,G$_0$ and
226: $0.5$\,G$_0$, which is indeed expected from the finite width of
227: the peaks in the histogram. The peak at $0.5$\,G$_0$ grows above
228: the interval $\sim 0.3-0.75$\,G$_0$, while the peak at $1$\,G$_0$
229: grows above $\sim 0.75-1.1$\,G$_0$. So all the plateaus in the
230: first interval are regarded as plateaus near $0.5$\,G$_0$, while
231: the plateaus in the second interval are taken as plateaus near
232: $1$\,G$_0$.
233:
234: Further on the lengths of these two types of plateaus are analyzed
235: with statistical methods.
236:
237: \section{Exploring correlations between the peaks in the histogram}
238:
239: As a first step, we investigate those traces for that the plateau
240: at $0.5$\,G$_0$ is long. The length of this plateau was determined
241: for each trace by counting the number of data points in the
242: conductance interval $G/G_0 \in [0.3, 0.75]$. The data set of
243: Fig.~\ref{AuH2hist.fig}(b) contained $10000$ conductance traces.
244: From that those $3000$ traces were selected, which have the
245: longest plateau at $0.5$\,G$_0$. The histogram for these selected
246: curves is shown in Fig.~\ref{AuH2corrhist.fig} by the area graph.
247: This histogram is compared with the histogram for the whole data
248: set, as shown by the dashed line. In order to have a good
249: comparison, both histograms were normalized to the number of
250: traces included.
251:
252: \begin{figure}[t!]
253: \centering
254: \includegraphics[width=0.8\columnwidth]{AuH2corrhist.eps}
255: \caption{\it The area graph shows the histogram for selected
256: $30$\,\% of the traces that have the longest plateau at $\sim
257: 0.5$\,G$_0$. The dashed line shows the histogram for the original,
258: unselected data set. Both histograms are normalized to the number
259: of traces included.} \label{AuH2corrhist.fig}
260: \end{figure}
261:
262: For the selected curves the peak at $0.5$\,G$_0$ is significantly
263: larger, which is a natural consequence of the selection. The
264: interesting result of this comparison is the shrinking of the peak
265: at $1$\,G$_0$ by more than $30$\,\% due to the selection. The rest
266: of the structures at higher conductance values are exactly the
267: same on the two histograms.
268:
269: This analysis shows that the traces with long plateaus at
270: $0.5$\,G$_0$ have a smaller peak at $1$\,G$_0$ in the histogram;
271: or in other words, if the plateau at $0.5$\,G$_0$ is longer than
272: in average then the plateau at $1$\,G$_0$ is shorter than in
273: average. This kind of anticorrelation implies that the summed
274: length of the both plateaus is important, and the switching
275: between them is probably a random process. This remark will have
276: special importance in the next part, where the histograms for the
277: plateau lengths are investigated.
278:
279: \section{Plateau length histograms}
280:
281: The measurements in a pure environment have shown, that a single
282: atom gold contact has a single conductance channel with perfect
283: transmission.\cite{Scheer1998,Ludoph1999} Furthermore it was found
284: that gold can form atomic chains with single atoms in a
285: row,\cite{Yanson1998} and this chain also has a perfect
286: transmission.\cite{Rubio2003} The existence of these atomic chains
287: was deduced from histograms plotted for the lengths of the last
288: plateaus before break. These histograms revealed equidistant
289: peaks, corresponding to the break of chains with different number
290: of atoms. Therefore the question must be addressed, to what extent
291: this chain formation is influenced by the presence of hydrogen.
292:
293: As it was mentioned, even in hydrogen surrounding, a lot of traces
294: just exhibit plateaus at $1$\,G$_0$ without any additional plateau
295: around $0.5$\,G$_0$. These traces can be regarded as ``pure gold
296: behavior'', and the plateau length histogram for these traces can
297: serve as a reference [Fig.~\ref{AuH2plh.fig}(a)]. This histogram
298: shows two well-defined peaks. This means that the process of chain
299: formation is present, but no really long chains are pulled. It
300: should be noted, that in metals that cannot form chains the
301: plateau length histogram shows only a single peak. The distance
302: between the two peaks in Fig.~\ref{AuH2plh.fig}(a) defines the
303: Au-Au distance in the chain.
304:
305: \begin{figure}[t!]
306: \centering
307: \includegraphics[width=0.6\columnwidth]{AuH2plh.eps}
308: \caption{\it Plateau length histograms for gold junctions in
309: hydrogen environment. Panel (a) shows the plateau length histogram
310: for the traces that do not have any plateau in the interval $G/G_0
311: \in [0.3, 0.75]$. The plateau length is determined by the number
312: of data points in the interval $G/G_0 \in [0.75, 1.1]$. Panel (b)
313: contains plateau length histograms for those traces that do have a
314: plateau in the region $G/G_0 \in [0.3, 0.75]$. The dashed line
315: shows the histogram for the lengths of the plateaus near
316: $1$\,G$_0$. This is measured by the number of points in the
317: interval $G/G_0 \in [0.75, 1.1]$. The area graph shows the
318: histogram for the joint lengths of the plateaus around $1$ and
319: around $0.5$\,G$_0$. This length is given by the number of points
320: in the interval $G/G_0 \in [0.3, 1.1]$} \label{AuH2plh.fig}
321: \end{figure}
322:
323: We can also plot the plateau length histogram for the traces that
324: have a plateau around $0.5$\,G$_0$ as well. The anticorrelation
325: demonstrated in the previous part implies that the summed length
326: of the plateaus at $1$\,G$_0$ and at $0.5$\,G$_0$ is important.
327: Therefore, as a first step the plateau length is measured for each
328: trace by the number of data points in the conductance interval
329: $G/G_0 \in [0.3, 1.1]$. The plateau length histogram built in this
330: way is presented by the area graph in Fig.~\ref{AuH2plh.fig}(b).
331: The position of the peaks coincides with the reference histogram.
332: The plateau length histogram for the same traces can be plotted by
333: measuring just the lengths of the plateaus at $1$\,G$_0$. (In this
334: case the plateau length is the number of data points in the
335: conductance interval $G/G_0 \in [0.75, 1.1]$) This analysis
336: results in the histogram shown by the dashed line. In this
337: histogram the peaks are smeared and considerably shifted compared
338: to the reference histogram.
339:
340: A plateau at $0.5$\,G$_0$ in a conductance trace is a clear sign,
341: that the break of the nanojunction is influenced the hydrogen
342: molecules. The two peaks in the plateau length histogram in
343: Fig.~\ref{AuH2plh.fig}(b) indicate that even in this case the
344: chain formation of gold is preserved. Furthermore, a nice plateau
345: length histogram is only obtained if the plateaus at $0.5$\,G$_0$
346: are also included in the plateau length. This indicates that the
347: atomic configuration with $G=0.5$\,G$_0$ is a part of the chain
348: formation.
349:
350: \section{Discussion of the observations}
351:
352: Our measurements have shown, that the adsorption of hydrogen on
353: gold junctions results in the appearance of a new peak at
354: $0.5$\,G$_0$ in the conductance histogram. The same phenomenon was
355: not observed in silver and copper junctions, which have conduction
356: properties similar to gold. An important difference between the
357: nobel metals is the striking property that gold can form atomic
358: chains while Ag and Cu cannot.\cite{Smit2001} It has already
359: implied that the new peak at $0.5$\,G$_0$ is somehow connected to
360: the chain formation of gold. This assumption was further supported
361: by the investigation of plateau length histograms. This analysis
362: has shown that the configuration with $G=0.5$\,G$_0$ is indeed a
363: part of the chain formation.
364:
365: Local density functional simulations have shown that a small
366: increase in the interatomic distance in a Au chain causes a
367: considerable reduction of the conductance.\cite{Hakkinen1999} This
368: implies, that the configuration with perfect transmission is the
369: usual gold chain, while the conductance value of $G=0.5$\,G$_0$
370: corresponds to a distorted chain with slightly increased Au-Au
371: distance.
372:
373: This distortion can be an intrinsic property of gold, which is
374: just aided by the hydrogen environment. It was shown that
375: stretched atomic chains have a tendency for spontaneous
376: dimerization.
377: \cite{Torres1999,Okamoto1999,Hakkinen1999,Hakkinen2000,Maria2000}
378: Calculations of the conductance through a short chain of gold
379: atoms with a dimer yield values from $0.58$\,G$_0$
380: \cite{Hakkinen1999} to $0.4$\,G$_0$.\cite{Okamoto1999} Such dimers
381: have never been observed in high vacuum, but the adhesive hydrogen
382: environment might stabilize this configuration. A crucial problem
383: with this interpretation is the following. In a process called
384: ``dimerization'' at least three atoms should be included in the
385: chain. In our measurements the occurrence of such long chains was
386: not frequent (see the plateau length histograms in
387: Fig.~\ref{AuH2plh.fig}), so a plausible explanation should also
388: account for the distortion of a chain with two atoms.
389:
390: As another possibility, the binding of the hydrogen molecule on
391: the chain itself might cause the distortion. The precise nature of
392: such process cannot be predicted without microscopic model
393: calculations. The possibility of the chemical reaction of H$_2$
394: with the gold chain also cannot be excluded.
395:
396: The measurements on hydrogen covered platinum junctions have
397: demonstrated that the new peak in the histogram is connected to
398: the conductance through a hydrogen molecule.\cite{Smit2002} In
399: gold the plateau length histograms strongly indicate that the
400: conductance values near $0.5$\,G$_0$ are related to chains of gold
401: atoms; therefore, the same explanation is not supported. A similar
402: analysis of the plateau length histograms can also be performed on
403: platinum, which has the tendency of chain formation as well. Our
404: measurements have shown that in Pt the chain formation is
405: completely destroyed in the presence of hydrogen.
406:
407: In conclusion, we have shown that a new stable atomic
408: configuration appears in gold nanojunctions due to the adsorption
409: of hydrogen. The analysis of the results implies that this
410: configuration is related to a distorted chain of gold atoms.
411: Presently, the experimental observations cannot provide more
412: information about the nature of this configuration. Hopefully, the
413: simplicity of the gold-hydrogen adsorbate system will stimulate
414: extensive theoretical investigations, which will lead to a better
415: understanding of the details of molecular adsorption on nanowires.
416:
417: This work has been supported by the ``Stichting FOM'' and the
418: Hungarian research funds OTKA TS040878, T037451.
419:
420: \begin{thebibliography}{17}
421: \expandafter\ifx\csname
422: natexlab\endcsname\relax\def\natexlab#1{#1}\fi
423: \expandafter\ifx\csname bibnamefont\endcsname\relax
424: \def\bibnamefont#1{#1}\fi
425: \expandafter\ifx\csname bibfnamefont\endcsname\relax
426: \def\bibfnamefont#1{#1}\fi
427: \expandafter\ifx\csname citenamefont\endcsname\relax
428: \def\citenamefont#1{#1}\fi
429: \expandafter\ifx\csname url\endcsname\relax
430: \def\url#1{\texttt{#1}}\fi
431: \expandafter\ifx\csname
432: urlprefix\endcsname\relax\def\urlprefix{URL }\fi
433: \providecommand{\bibinfo}[2]{#2}
434: \providecommand{\eprint}[2][]{\url{#2}}
435:
436: \bibitem[{\citenamefont{Agrait et~al.}(2003)\citenamefont{Agrait, Yeyati, and
437: van Ruitenbeek}}]{Agrait2003}
438: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Agrait}},
439: \bibinfo{author}{\bibfnamefont{A.~L.} \bibnamefont{Yeyati}},
440: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van
441: Ruitenbeek}}, \bibinfo{journal}{Phys. Rep.} \textbf{\bibinfo{volume}{377}},
442: \bibinfo{pages}{81} (\bibinfo{year}{2003}).
443:
444: \bibitem[{\citenamefont{Li et~al.}(2000)\citenamefont{Li, He, Bogozi, Bunch,
445: and Tao}}]{Li2000}
446: \bibinfo{author}{\bibfnamefont{C.~Z.} \bibnamefont{Li}},
447: \bibinfo{author}{\bibfnamefont{H.~X.} \bibnamefont{He}},
448: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Bogozi}},
449: \bibinfo{author}{\bibfnamefont{J.~S.} \bibnamefont{Bunch}}, \bibnamefont{and}
450: \bibinfo{author}{\bibfnamefont{N.~J.} \bibnamefont{Tao}},
451: \bibinfo{journal}{Appl. Phys. Lett.} \textbf{\bibinfo{volume}{76}},
452: \bibinfo{pages}{1333} (\bibinfo{year}{2000}).
453:
454: \bibitem[{\citenamefont{Shu et~al.}(2000)\citenamefont{Shu, Li, He, Bogozi,
455: Bunch, and Tao}}]{Shu2000}
456: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Shu}},
457: \bibinfo{author}{\bibfnamefont{C.~Z.} \bibnamefont{Li}},
458: \bibinfo{author}{\bibfnamefont{H.~X.} \bibnamefont{He}},
459: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Bogozi}},
460: \bibinfo{author}{\bibfnamefont{J.~S.} \bibnamefont{Bunch}}, \bibnamefont{and}
461: \bibinfo{author}{\bibfnamefont{N.~J.} \bibnamefont{Tao}},
462: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{84}},
463: \bibinfo{pages}{5196} (\bibinfo{year}{2000}).
464:
465: \bibitem[{\citenamefont{Smit et~al.}(2002)\citenamefont{Smit, Noat, Untiedt,
466: Lang, van Hemert, and van Ruitenbeek}}]{Smit2002}
467: \bibinfo{author}{\bibfnamefont{R.~H.~M.} \bibnamefont{Smit}},
468: \bibinfo{author}{\bibfnamefont{Y.}~\bibnamefont{Noat}},
469: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Untiedt}},
470: \bibinfo{author}{\bibfnamefont{N.~D.} \bibnamefont{Lang}},
471: \bibinfo{author}{\bibfnamefont{M.~C.} \bibnamefont{van Hemert}},
472: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van
473: Ruitenbeek}}, \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{419}},
474: \bibinfo{pages}{906} (\bibinfo{year}{2002}).
475:
476: \bibitem[{\citenamefont{Yanson}(2001)}]{Yanson2001}
477: \bibinfo{author}{\bibfnamefont{A.~I.} \bibnamefont{Yanson}}, Ph.D. thesis,
478: \bibinfo{school}{Universiteit Leiden, The Netherlands}
479: (\bibinfo{year}{2001}).
480:
481: \bibitem[{not()}]{note1}
482: \bibinfo{note}{The admitted amount of H$_2$ was between $1\mu$mole$-1$mmole.}
483:
484: \bibitem[{\citenamefont{Halbritter et~al.}(2002)\citenamefont{Halbritter,
485: Csonka, Kolesnychenko, Mihaly, Shklyarevskii, and van
486: Kempen}}]{Halbritter2002}
487: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Halbritter}},
488: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Csonka}},
489: \bibinfo{author}{\bibfnamefont{O.~Y.} \bibnamefont{Kolesnychenko}},
490: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Mihaly}},
491: \bibinfo{author}{\bibfnamefont{O.~I.} \bibnamefont{Shklyarevskii}},
492: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{van
493: Kempen}}, \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{65}},
494: \bibinfo{pages}{45413} (\bibinfo{year}{2002}).
495:
496: \bibitem[{\citenamefont{Scheer et~al.}(1998)\citenamefont{Scheer, Agrait,
497: Cuevas, Yeyati, Ludoph, Mart{\'{\i}}n-Rodero, Bollinger, van Ruitenbeek, and
498: Urbina}}]{Scheer1998}
499: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Scheer}},
500: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Agrait}},
501: \bibinfo{author}{\bibfnamefont{J.~C.} \bibnamefont{Cuevas}},
502: \bibinfo{author}{\bibfnamefont{A.~L.} \bibnamefont{Yeyati}},
503: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Ludoph}},
504: \bibinfo{author}{\bibfnamefont{A.}~\bibnamefont{Mart{\'{\i}}n-Rodero}},
505: \bibinfo{author}{\bibfnamefont{G.~R.} \bibnamefont{Bollinger}},
506: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van Ruitenbeek}},
507: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Urbina}},
508: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{394}},
509: \bibinfo{pages}{154} (\bibinfo{year}{1998}).
510:
511: \bibitem[{\citenamefont{Ludoph et~al.}(1999)\citenamefont{Ludoph, Devoret,
512: Esteve, Urbina, and van Ruitenbeek}}]{Ludoph1999}
513: \bibinfo{author}{\bibfnamefont{B.}~\bibnamefont{Ludoph}},
514: \bibinfo{author}{\bibfnamefont{M.~H.} \bibnamefont{Devoret}},
515: \bibinfo{author}{\bibfnamefont{D.}~\bibnamefont{Esteve}},
516: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Urbina}}, \bibnamefont{and}
517: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van Ruitenbeek}},
518: \bibinfo{journal}{Phys. Rev. Lett.} \textbf{\bibinfo{volume}{82}},
519: \bibinfo{pages}{1530} (\bibinfo{year}{1999}).
520:
521: \bibitem[{\citenamefont{Yanson et~al.}(1998)\citenamefont{Yanson, Bollinger,
522: van~den Brom, Agrait, and van Ruitenbeek}}]{Yanson1998}
523: \bibinfo{author}{\bibfnamefont{A.~I.} \bibnamefont{Yanson}},
524: \bibinfo{author}{\bibfnamefont{G.~R.} \bibnamefont{Bollinger}},
525: \bibinfo{author}{\bibfnamefont{H.~E.} \bibnamefont{van~den Brom}},
526: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Agrait}}, \bibnamefont{and}
527: \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van Ruitenbeek}},
528: \bibinfo{journal}{Nature} \textbf{\bibinfo{volume}{395}},
529: \bibinfo{pages}{783} (\bibinfo{year}{1998}).
530:
531: \bibitem[{\citenamefont{Rubio-Bollinger
532: et~al.}(2003)\citenamefont{Rubio-Bollinger, de~las Heras, Bascones, Agrait,
533: Guinea, and Vieira}}]{Rubio2003}
534: \bibinfo{author}{\bibfnamefont{G.}~\bibnamefont{Rubio-Bollinger}},
535: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{de~las Heras}},
536: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Bascones}},
537: \bibinfo{author}{\bibfnamefont{N.}~\bibnamefont{Agrait}},
538: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Guinea}}, \bibnamefont{and}
539: \bibinfo{author}{\bibfnamefont{S.}~\bibnamefont{Vieira}},
540: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{67}},
541: \bibinfo{pages}{121407} (\bibinfo{year}{2003}).
542:
543: \bibitem[{\citenamefont{Smit et~al.}(2001)\citenamefont{Smit, Untiedt, Yanson,
544: and van Ruitenbeek}}]{Smit2001}
545: \bibinfo{author}{\bibfnamefont{R.~H.~M.} \bibnamefont{Smit}},
546: \bibinfo{author}{\bibfnamefont{C.}~\bibnamefont{Untiedt}},
547: \bibinfo{author}{\bibfnamefont{A.~I.} \bibnamefont{Yanson}},
548: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{van
549: Ruitenbeek}}, \bibinfo{journal}{Phys. Rev. Lett.}
550: \textbf{\bibinfo{volume}{87}}, \bibinfo{pages}{266102}
551: (\bibinfo{year}{2001}).
552:
553: \bibitem[{\citenamefont{H{\"a}kkinen et~al.}(1999)\citenamefont{H{\"a}kkinen,
554: Barnett, and Landman}}]{Hakkinen1999}
555: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{H{\"a}kkinen}},
556: \bibinfo{author}{\bibfnamefont{R.~N.} \bibnamefont{Barnett}},
557: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Landman}},
558: \bibinfo{journal}{J. Phys. Chem. B} \textbf{\bibinfo{volume}{103}},
559: \bibinfo{pages}{8814} (\bibinfo{year}{1999}).
560:
561: \bibitem[{\citenamefont{Torres et~al.}(1999)\citenamefont{Torres, Tosatti,
562: Corso, Ercolessi, Kohanoff, Tolla, and Soler}}]{Torres1999}
563: \bibinfo{author}{\bibfnamefont{J.~A.} \bibnamefont{Torres}},
564: \bibinfo{author}{\bibfnamefont{E.}~\bibnamefont{Tosatti}},
565: \bibinfo{author}{\bibfnamefont{A.~D.} \bibnamefont{Corso}},
566: \bibinfo{author}{\bibfnamefont{F.}~\bibnamefont{Ercolessi}},
567: \bibinfo{author}{\bibfnamefont{J.~J.} \bibnamefont{Kohanoff}},
568: \bibinfo{author}{\bibfnamefont{F.~D.~D.} \bibnamefont{Tolla}},
569: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{J.~M.} \bibnamefont{Soler}},
570: \bibinfo{journal}{Surface Science} \textbf{\bibinfo{volume}{426}},
571: \bibinfo{pages}{L441} (\bibinfo{year}{1999}).
572:
573: \bibitem[{\citenamefont{Okamoto and Takayanagi}(1999)}]{Okamoto1999}
574: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Okamoto}} \bibnamefont{and}
575: \bibinfo{author}{\bibfnamefont{K.}~\bibnamefont{Takayanagi}},
576: \bibinfo{journal}{Phys. Rev. B} \textbf{\bibinfo{volume}{60}},
577: \bibinfo{pages}{7808} (\bibinfo{year}{1999}).
578:
579: \bibitem[{\citenamefont{H{\"a}kkinen et~al.}(2000)\citenamefont{H{\"a}kkinen,
580: Barnett, Scherbakov, and Landman}}]{Hakkinen2000}
581: \bibinfo{author}{\bibfnamefont{H.}~\bibnamefont{H{\"a}kkinen}},
582: \bibinfo{author}{\bibfnamefont{R.~N.} \bibnamefont{Barnett}},
583: \bibinfo{author}{\bibfnamefont{A.~G.} \bibnamefont{Scherbakov}},
584: \bibnamefont{and} \bibinfo{author}{\bibfnamefont{U.}~\bibnamefont{Landman}},
585: \bibinfo{journal}{J. Phys. Chem. B} \textbf{\bibinfo{volume}{104}},
586: \bibinfo{pages}{9066} (\bibinfo{year}{2000}).
587:
588: \bibitem[{\citenamefont{Maria and Springborg}(2000)}]{Maria2000}
589: \bibinfo{author}{\bibfnamefont{L.~D.} \bibnamefont{Maria}} \bibnamefont{and}
590: \bibinfo{author}{\bibfnamefont{M.}~\bibnamefont{Springborg}},
591: \bibinfo{journal}{Chem. Phys. Lett.} \textbf{\bibinfo{volume}{323}},
592: \bibinfo{pages}{293} (\bibinfo{year}{2000}).
593:
594: \end{thebibliography}
595:
596: \end{document}
597: