1: %\documentclass[aps,pra,groupedaddress]{revtex4}
2: \documentclass[aps,prb,twocolumn,groupedaddress]{revtex4}
3: %\documentclass[aps,prb,preprint,groupedaddress]{revtex4}
4: \usepackage{graphics,epsfig}
5: %\usepackage{dcolumn}
6: %\linespread{1.3}
7: \begin{document}
8:
9: \title{Density-functional study of adsorption of isocyanides on the gold (111) surface.}
10:
11: \author{Yulia Gilman}
12: \affiliation{Department of Physics and Astronomy, State University of New York,
13: Stony Brook, NY 11794-3800}
14:
15: \author{Philip B. Allen}
16: \affiliation{Department of Physics and Astronomy, State University of New York,
17: Stony Brook, NY 11794-3800\footnote{Permanent address.},
18: \\and Center for Functional Nanomaterials, Brookhaven National Laboratory, Upton, New York 11973-5000}
19:
20: \author{Mark S. Hybertsen}
21: \affiliation{Department of Applied Physics and Applied Mathematics,
22: and Center for Electron Transport in Molecular Nanostructures,
23: Columbia University, New York, NY 10027}
24:
25:
26: \begin{abstract}
27:
28: Density functional theory within the generalized-gradient approximation
29: is used to study the adsorption of
30: the isocyanides CNH and CNCH$_3$ on the gold (111) surface at several coverages.
31: It is found that these molecules are highly selective
32: in their adsorption site preference.
33: Adsorption is possible only at the top site,
34: although binding is rather weak and dipole-dipole repulsion prevents
35: binding at coverages of 1/3 ML and higher.
36: At all other high symmetry sites considered (hcp, fcc, bridge),
37: the isonitriles are not bound.
38: To reveal the main mechanisms of bonding,
39: the isonitriles are compared to CO and ammonia using
40: Au-X radicals as model systems.
41: The systematic trends are understood in terms of $\sigma$ donation
42: from the ligand lone pair and $\pi^*$ back donation.
43: Finally, CNH is found to be more strongly bound to an
44: undercoordinated gold atom (adatom) on the surface.
45: \end{abstract}
46:
47:
48: \maketitle
49:
50: \section{Introduction}
51:
52: Gold is widely chosen as an electrode material for studies of molecular conductance
53: for a number of practical reasons, including ease of deposition and patterning and
54: presentation of a surface that is relatively easy to clean and maintain.
55: Another important factor has been the extensive base of experience in forming
56: self assembled monolayers (SAMs) of organic molecules, linked to the gold surface via
57: a sulfur atom.
58: The initial discovery of the thiol-gold route to self assembly \cite{nuzzo}
59: led to rapid growth in the study of the formation and properties
60: of organic layers \cite{ulman,schreiber04}.
61: The S-Au bond is relatively strong ($\sim$1.7 eV), but the barrier to lateral
62: motion between adsorption sites is rather low ($\sim$0.1 eV), facilitating formation of
63: ordered monolayers \cite{ulman}.
64: Sulfur has been alsmost exclusively used as the ``alligator clip'' \cite{tour}
65: in the experiments probing conductance at the molecular scale \cite{reed,salomon}.
66: However, the substantial variation in the measured conductance
67: for basic molecules such as alkanes contacted using the Au-S link,
68: between different groups and different measurement techniques, remains a puzzle.
69: One possibility is sample to sample variation in the local structure
70: of the Au-S link \cite{basch}.
71: This is consistent with the low barriers for motion that facilitate assembly
72: and the known mobility of Au atoms
73: during formation of the thiol linked layer \cite{poirrer,schreiber00}.
74: Also, the atomic scale structure of the interface of assembled layers with Au-S linkages
75: has proved to be a very delicate problem for {\it ab-initio} methods \cite{schreiber00},
76: although a recent study appears to reconcile extensive
77: data for decanethiols \cite{fischer}.
78: Overall, despite wide usage, the Au-S link displays a number of critical issues for
79: application in molecular conductance studies.
80: Recent research has been directed to find alternative link chemistries,
81: e.g. Ru or Mo carbene \cite{nuckolls,mcbreen}.
82:
83: An alternative link chemistry for a gold electrode that shows more
84: selectivity in the local bonding geometry is an attractive possibility in
85: the context of molecular scale conductance.
86: Although Au surfaces are relatively unreactive,
87: other candidates have been explored.
88: A very recent study demonstrated the utility of amine terminated molecules
89: for molecular conductance studies \cite{venkataraman}. The Au-NH$_2$-R links
90: proved to be flexible and reproducible.
91: Another example, with a similar bonding character, is isocyanide (CN-R, e.g. CNH),
92: a less-stable isomer of cyanide (NC-R, e.g. NCH).
93: Well-established synthetic methods exist for the CN- linker
94: with various molecules (R). Bonding to Pt or Au electrodes has been observed \cite{lin}.
95: In contrast to the Au-S link,
96: adsorption of isocyanide (Au-CN) linked molecules on gold surfaces is
97: much less studied.
98: Most experimental works on this subject consider adsorption of the molecules
99: on gold nanoparticles \cite{8,9,10,11}.
100: Henderson {\it et al.} \cite{7} studied the formation of self-assembled monolayers of
101: diisocyanides on the gold (111) surface.
102: It was concluded from reflection absorption infrared (RAIR) spectra
103: that the molecules with rigid chains
104: between two CN- groups (such as biphenyldiisocyanide) attach
105: to the gold surface through only one isocyanide group while the other group
106: remains free.
107: Ellipsometry data indicate that in these cases, the molecular axis is
108: perpendicular to the surface.
109: Molecules with flexible alkyl chains can attach to the gold surface
110: through both isocyanide groups, which makes them unlikely
111: candidates for molecular conductance studies.
112: The preferred adsorption site and geometry
113: remain as unresolved questions in the literature.
114: Angelici {\it et al.} observed the peak in the infrared spectra
115: associated with N-C stretch mode to be shifted 50-70 cm$^{-1}$
116: higher upon adsorption on gold powder \cite{8,9}.
117: They concluded that isocyanides bind to the surface at the top site.
118: Henderson {\it et al.} \cite{7} observed a similar blue shift
119: for adsorption on the gold (111) surface, but concluded
120: that both atop and three-fold adsorption configurations are possible.
121:
122: The earliest of the few theoretical works on the adsorption of isocyanide
123: on gold considered phenyl isocyanide
124: attached to a single gold atom, a gold dimer and a
125: trimer \cite{13}.
126: The calculations showed that for each gold cluster, the binding was to a single Au atom.
127: Bonding motifs analogous to the bridge and hollow sites on the (111) surface
128: were not observed.
129: The results suggest site selectivity for the isocyanide link, although
130: the details of the adsorption energy and the geometry on the Au(111)
131: surface remain to be determined.
132: Recently several groups \cite{1,2} successfully fabricated self-assembled monolayers
133: of various diisocyanide molecules
134: sandwiched between two gold electrodes, and studied electrical transport
135: through them.
136: The impact of the isocyanide link on the conduction has received
137: some attention from theorists as well.
138: Conductance and voltage-current characteristics of molecules
139: connected to metallic electrodes were calculated using combined
140: density functional theory (DFT) and non-equilibrium Green's function
141: methods \cite{3}.
142: However, the adsorption geometry was not explicitly determined
143: by energy minimization.
144:
145: This paper presents a study of the adsorption of isocyanide linked
146: molecules on the gold (111) surface using a density functional theory (DFT) approach.
147: We focus on the binding of the essential link element to the metal surface,
148: considering the simplest case, a CNH molecule.
149: Calculations performed for CNCH$_3$ confirm the results.
150: The basic electronic structure of the CNH molecule suggests that the filled
151: lone pair on the C atom will be available for donor-acceptor type bonding
152: to Au complexes, in analogy with ammonia (NH$_3$) and CO.
153: We chose to draw a comparison between CNH, CO and NH$_3$
154: in order to better understand the factors that affect adsorption.
155: The case of CO has been extensively studied
156: for many transition metal surfaces \cite{trends}.
157: There are also recent calculations for the case of NH$_3$ on Au(111) \cite{ammonia}.
158: The standard Blyholder picture \cite{Blyholder} describes the bonding
159: in terms of the trade-off between $\sigma$ donation from the lone pair to the
160: metal and back donation from the metal to the empty $\pi^*$ molecular orbital.
161: Two main themes emerge: (1) the interplay of the electronic couplings
162: in the Blyholder picture with
163: the corresponding charge transfers; (2) the impact of the dipole moment, both
164: that of the molecule prior to adsorption and the induced dipole upon adsorption.
165: In the next section, the methodology and the basic results of the DFT
166: calculations are described.
167: The physical interpretation of the results is discussed in Sect. III.
168:
169:
170: \section{Methodology and results}
171:
172: The adsorption of the target molecules on the Au surface is studied
173: using a computational approach based on DFT.
174: The main calculations are done using the generalized gradient approximation (GGA)
175: of Perdew, Burke and Ernzerhof (PBE) \cite{PBE}.
176: Different forms of GGA are widely used for metal
177: surface and adsorption problems in order to address the
178: large errors in adsorption energy found
179: with the local density approximation (LDA),
180: although the accuracy still varies with the system
181: under study \cite{gross,PBEtest,hammer,trends,feibelman}.
182: For comparison, selected calculations are also done
183: with the LDA \cite{LDA,goedecker}.
184: To have an initial picture of the bonding, the target molecules bonded to a
185: single Au atom are considered
186: using the NRLMOL DFT code \cite{NRLMOL}.
187: For the light atoms (C, N, and H), an all electron basis is used.
188: A slight modification \cite{NRLMOLpsp} of norm-conserving
189: Troullier-Martins pseudopotentials is used for gold atoms.
190: NRLMOL uses Gaussian basis sets optimized for
191: density-functional calculations \cite{NRLMOLbasis}.
192: The exponents and contraction coefficients are determined by
193: optimizing the total energy of the free atom in the ground state.
194: For the valence states, the number of independent contractions used
195: (s, p, d) for each atom is as follows: H (4, 3, 1), C (5, 4, 3), N (5, 4, 3)
196: and Au (4, 2, 4).
197: It has been shown that these basis sets are well-converged and
198: that they have negligible basis set superposition error.
199: Further details of the basis sets can be found in Ref. \cite{NRLMOLbasis}.
200: Radicals were treated with spin-polarized, unrestricted calculations.
201:
202:
203: The adsorption on the surface was modeled with the standard scheme
204: of a periodically repeated slab geometry.
205: We found that the slab of four layers of gold atoms was enough to
206: reproduce key adsorption characteristics such as the binding energy.
207: The molecules are adsorbed on one side of the slab.
208: For most of the calculations, the molecular structure is relaxed,
209: with the molecular axis kept perpendicular to the surface and the Au atoms in
210: the slab frozen at their bulk positions.
211: Relaxation is carried out until the
212: maximum force is less than 0.15 eV/$\AA$.
213: For selected cases, the top layer of Au atoms was allowed to relax.
214: We also explored bonding of CNH to an Au adatom in the hcp hollow site,
215: allowing the adatom and the surface Au atoms to relax.
216: Most of these calculations were done using
217: the WIEN2k DFT code (full potential and linearized
218: augmented plane waves \cite{12}).
219: For comparison, some calculations were also done with the
220: ABINIT package \cite{abinit,gonze},
221: which employs a plane wave basis set and pseudopotentials.
222: LDA calculations were done with
223: Hartwigsen-Goedecker-Hutter (HGH) pseudopotentials \cite{HGH},
224: while GGA calculations were done with Troullier-Martins type pseudopotentials \cite{TM,FHI}.
225: It was checked that the energy is converged with respect to the number of
226: k-points as well as the size
227: of the basis set.
228: To facilitate analysis, the partial densities of state (PDOS) were plotted in selected
229: cases.
230: In the WIEN2K code, a modified tetrahedron method \cite{tetrahedron}
231: is used for density of state calculations;
232: the PDOSs are calculated by projecting the wave functions
233: onto the spherical harmonic basis functions centered on the atoms in question
234: inside a sphere centered on the atom
235: (atomic sphere radii are 1.06 $\AA$ for Au atoms, 0.58 $\AA$ for C and N, and 0.29 $\AA$ for H).
236:
237: We start by considering binding of the CNH and CNCH$_3$ molecules to a single gold atom.
238: The binding energy of the Au atom, the gold to ligand distances, and
239: the dipole moments of the free molecules and
240: the molecules bound to the gold atom are reported in Table I.
241: The results for CNH and CNCH$_3$ are very similar.
242: For comparison, results for CO and NH$_3$ are also shown.
243: The first step of relaxation allows only configurations with axial symmetry.
244: The axial CO and CNH molecules are very similar in their electronic structure.
245: The highest occupied molecular orbital (HOMO) is the lone pair on the C
246: and there is a doubly degenerate $\pi$ bond between C and the N or O.
247: For NH$_3$, the N lone pair is the HOMO, but there are no $\pi$ bonds.
248: The $\sigma$ part of the interaction in all three is due to coupling of the
249: lone pair orbital on the ligand to the $d_{z^2-r^2}$ and s states on the Au.
250: This results in partial transfer of electron density onto the Au atom.
251: Because the Au s state is half filled, the frontier, anti-bonding $\sigma$ state
252: in the complex is also half filled.
253: The initial ligand lone pair orbital energy is highest
254: for NH$_3$, comparable to the Au s state,
255: and successively lower for CNCH$_3$, CNH and lowest for CO.
256:
257: The $\sigma$ donation is partially balanced by the back donation
258: from the Au $d_{zx}$ and $d_{yz}$
259: into the empty $\pi^*$ doublet on the molecule for CO, CNH and CNCH$_3$.
260: This balance is slightly different for AuCNH and AuCNCH$_3$ relative to AuCO.
261: In the AuCO case, the C-O bond length increases by 0.004 $\AA$ relative to CO.
262: In the AuCNH (AuCNCH$_3$) case,
263: the C-N bond length decreases by 0.007 $\AA$ relative to CNH (CNCH$_3$).
264: This difference suggests less $\pi^*$ back donation in the AuCNH case.
265: The reduced back donation is consistent with the $\pi^*$ doublet being about
266: 1 eV higher in CNH and even a bit higher
267: for CNCH$_3$. This is also consistent with the trend of increasing Au-C
268: bond length from AuCO to AuCNH to CNCH$_3$.
269: The trend for net charge transfer to the Au as measured by Mulliken type analysis
270: is consistent with larger $\sigma$ donation and smaller $\pi$ back donation
271: from AuCNCH$_3$ to AuCNH to AuCO.
272: The binding energy follows the same order, but the impact
273: of larger $\sigma$ donation and smaller $\pi^*$ back donation are competitive,
274: with the $\sigma$ donation dominating the trend.
275: By comparison, the NH$_3$ molecule shows
276: the strongest $\sigma$ donation, with the largest charge transfer to Au,
277: but offers no empty $\pi$ space for back donation.
278: Its binding energy to a gold atom
279: is 0.60 eV, less than for CO.
280: Correspondingly the Au-N bond length is substantially larger (2.277 $\AA$).
281: The role of the $\pi$ states is further highlighted by the
282: results of full relaxation, which leads to bent structures for
283: Au-CNH, Au-CNCH$_3$ and Au-CO.
284: This bending is due to the pseudo Jahn-Teller effect.
285: The lowered symmetry allows mixing between the half occupied $\sigma$ antibonding level
286: in the complexes and one of the empty $\pi^*$ states.
287: The resultant energy gain depends on the gap between $\sigma$ and $\pi^*$ states.
288: It is largest for the Au-CO complex and smallest for Au-CNCH$_3$;
289: the Au-CO complex has the
290: smallest gap between $\sigma$ and $\pi^*$ states,
291: while Au-CNCH$_3$ has the largest.
292:
293: In comparison to CO,
294: a key difference is that CNH and NH$_3$ have a much larger dipole moment.
295: Table 1 shows the calculated dipole moments which agree well with the
296: measured values.
297: In all three cases,
298: there is a substantial increase in the dipole of the gold-ligand complex relative
299: to the isolated ligand.
300: The dipole moment of the AuCO complex is 1.6 D for the linear configuration,
301: compared to the
302: small dipole of the CO molecule.
303: The dipole of the AuCNH complex increases by 2.4 D over the CNH ligand,
304: while the increase is 3.2 D for the NH$_3$ case.
305: The increase in dipole can not be simply seen in a point charge approximation
306: with associated charge transfer (e.g. from a Mulliken type analysis).
307: Instead, the extension of the lone pair electrons at the C end of
308: the ligand onto the Au by quantum mechanical mixing with the Au s state is responsible,
309: partly balanced by the $\pi^*$ back donation.
310: The reduced back donation in the AuCNH case results in a larger change in dipole.
311: In the NH$_3$ case there is no back donation and the dipole change is even larger.
312: On the other hand, the distortion to the bent form in the Au-CNH
313: and Au-CO cases results in a much reduced dipole,
314: consistent with the enhanced back donation from the Au to the C.
315:
316: In addition to the hybridization considerations, the dipole also influences
317: the relative binding of Au to the ligands.
318: There is an additional attraction in the Au-CNH complex due
319: to polarization of the Au by the dipole.
320: This influences the differences in binding energy, in addition to the systematic
321: hybridization differences.
322: To estimate this contribution, we use the calculated polarizability of the
323: Au atom \cite{ammonia} together with a point charge model of the molecule
324: derived from fitting to the self consistent electrostatic potential and the
325: calculated dipole moment.
326: For the case of AuCNH, we estimate that polarization contributes 0.1 eV to
327: the binding energy. For AuNH$_3$, the effect is even smaller (0.01 eV).
328:
329: So far as we know, there are no direct measurements of the Au binding energy
330: in these radicals.
331: The Au-NH$_3$ radical was studied using both DFT approaches and quantum chemistry
332: approaches for evaluation of
333: the correlation energy (e.g. coupled cluster with singles, doubles
334: and triple corrections, CCSD(T)) \cite{lambropoulos}.
335: Substantial care was also taken to account for basis set superposition
336: errors, which can be a significant issue for these correlation techniques
337: in particular \cite{dargel}.
338: The Au binding energy in this case was deduced to be 0.78$\pm$0.06 eV,
339: similar to the GGA-PBE value we obtain.
340: There are several studies of the Au-CO radical with DFT and quantum chemistry
341: approaches \cite{schwerdtfeger, liang, mendizabal, wu}.
342: Our result for the lowest energy
343: geometry, the bent form, agrees reasonably well
344: with previous values based on DFT with GGA:
345: 0.71 eV \cite{liang} and 0.80 eV \cite{wu}.
346: However, the correlated electron calculations consistently give smaller
347: binding energies, ranging from 0.14 to 0.56 eV depending on the details
348: \cite{schwerdtfeger, mendizabal}.
349: Given concerns about basis set superposition errors, it is premature to draw
350: a final conclusion from the quantum chemistry calculations in this case.
351:
352:
353: \begin{table*}
354:
355: \caption{Comparison of Au-X complexes, where X is CO, NH$_3$, CNH and CNCH$_3$. All characteristics are calculated with NRLMOL DFT
356: code using both PBE or LDA exchange-correlation potentials. Experimental values of the dipole moments are given in parentheses when available.
357: The X dipole column corresponds to the dipole moment of the isolated molecule X in the case of linear configurations.
358: In the case of a bent configuration, it corresponds to the dipole moment of the ligand X
359: with the geomtery taken from the fully relaxed bent Au-X complex.}
360: \begin{tabular}{lcccccc}
361: \hline
362: \hline
363: Molecule X & Au-X geometry & Functional & Binding energy, eV & Au-X bond length, $\AA$ & X dipole, D & Au-X dipole, D\\
364: \hline
365: CO & linear & PBE & $0.68 $ & $1.997$ & $0.20 (0.12)$\footnote{J. S. Muenter, J. Mol. Spectrosc. {\bf 55}, 490 (1975).
366: } & $1.65$ \\
367: CO & linear & LDA & $1.27 $ & $1.948$ & $0.23$ & $1.57$ \\
368: CNH & linear & PBE & $0.93 $ & $2.002$ & $3.09 (3.05)$\footnote{G.L. Blackman, R.D. Brown, P.D. Godrey, and H.I. Gunn, Nature {\bf 261}, 395 (1976).} & $5.46$ \\
369: CNH & linear & LDA & $1.54 $ & $1.957$ & $3.20$ & $5.50$ \\
370: NH$_3$ & linear & PBE & $0.60 $ & $2.277$ & $1.53$ & $4.72$ \\
371: NH$_3$ & linear & LDA & $1.07 $ & $2.188$ & $1.54$ & $4.89$ \\
372: CNCH$_3$& linear & PBE & $ 0.99 $ & $2.011$ & $3.95 (3.83)$\footnote{S.N. Ghosh, R. Trambarulo, and W. Gordy, J. Chem. Phys. {\bf 21}, 308 (1953).
373: } & $6.97$ \\
374: CO & bent & PBE & $0.87 $ & $2.005$ & $0.14$ & $0.35$ \\
375: CNH & bent & PBE & $1.04 $ & $1.989$ & $2.69$ & $2.89$ \\
376: CNCH$_3$ & bent & PBE & $ 1.05 $ & $2.019$ & $3.94$ & $5.74$ \\
377: \hline
378: \hline
379:
380: \end{tabular}
381: \end{table*}
382:
383: \begin{figure}
384: \includegraphics{fig1.eps}
385: \caption{Total energy (GGA-PBE) per unit cell for HNC monolayers with $1/1$ and $1/3$ coverages on the gold slab { \it vs.} distance $d$
386: to the surface. Molecules are at the top or filled hollow site with molecular axis perpendicular to the surface. The indicated adsorption energy is relative to a uniform molecular layer well separated from the surface.
387: }
388: \end{figure}
389:
390: Turning to the adsorption on the surface,
391: we initially consider a full monolayer of CNH
392: adsorbed on one side of a slab of four gold layers.
393: The supercell contains one molecule and four gold atoms, two of them being
394: surface atoms.
395: First, the molecules are placed at the top site (above the surface Au atom)
396: and the distance from the carbon atom to the surface is varied. The molecular axis is kept perpendicular to the
397: surface. The observed
398: dependence of total energy { \it vs.} distance $d$ is shown in Fig.1.
399: The binding energy $E_{ads}$ is the energy per molecule required to separate
400: the layer of CNH from the Au (111) surface, leaving the free standing
401: CNH layer otherwise unchanged.
402: The value of $E_{ads}$ is found to be 0.43 eV and the optimal distance is 2.01 $\AA$.
403: In order to check that four layers of gold are enough to represent
404: the surface, we repeated the calculation for three, five, and six
405: gold layers in the slab. The results value of $E_{ads}$ varies
406: from 0.453 to 0.431 eV,
407: indicating that the choice of four layers is justified.
408:
409: When a molecule is placed at the filled (hcp) hollow site, the
410: energy {\it vs.} distance (see Fig. 1) is
411: qualitatively different from that at the top site. Although there is a
412: local minimum at around 1.6 $\AA$, the energy
413: is minimal when the molecule is far away from the surface.
414: This indicates that the CNH molecule cannot be stabilized
415: at the hollow hcp site. The same result is obtained for the empty hollow (fcc)
416: and for the bridge sites.
417:
418:
419: The next candidate for a stable configuration at uniform commensurate coverage
420: is a monolayer with one molecule per three surface atoms, forming a
421: $\sqrt{3}\times\sqrt{3}R30^{\circ}$ equilateral triangular lattice on the surface,
422: where R=2.883 $\AA$ is the distance between nearest neighbor gold atoms.
423: The distance between molecules in this case is 4.944 $\AA$.
424: For the CNH molecules positioned at the top site, we relax the monolayer keeping the molecular axis normal
425: to the surface and gold atoms frozen.
426: The adsorption energy is found to be 0.18 eV relative to the separated slab-monolayer system.
427: For the hcp site and $\sqrt{3}\times\sqrt{3}R30^{\circ}$ surface unit cell, similar to the $1\times 1$ case,
428: the energy is minimal
429: when molecules are far away from the surface, although there is a local minimum in energy curve (see Fig. 1).
430: This shows that adsorption is not possible at the hcp site.
431: Similar results are obtained for the CNH monolayer with 1/4 coverage ($2\times 2$ surface unit cell) and
432: the CNCH$_3$ monolayer with 1/3 coverage
433: (see Table II).
434:
435:
436: \begin{table}
437: \caption{Stabilization energies from WIEN2K calculations for various
438: molecules obtained with GGA-PBE exchange-correlation potential.}
439: \begin{tabular}{lccc}
440: \hline
441: \hline
442: Molecule & Coverage & Site & Stabilization energy, eV \\
443: \hline
444: CNH & 1/1 & top & $0.43 $ \\
445: CNH & 1/1 & hcp & no adsorption \\
446: CNH & 1/1 & fcc & no adsorption \\
447: CNH & 1/1 & bridge & no adsorption \\
448: CNH & 1/3 & top & $0.18 $ \\
449: CNH & 1/3 & hcp & no adsorption \\
450: CNH & 1/4 & top & $0.20 $ \\
451: CNH & 1/4 & hcp & no adsorption \\
452: CNCH$_3$ & 1/3 & top & $0.22 $ \\
453: CNCH$_3$ & 1/3 & hcp & no adsorption \\
454: CO & 1/3 & top & $0.20 $ \\
455: \hline
456: \hline
457: \end{tabular}
458: \end{table}
459:
460:
461: So far binding of pre-formed monolayers was considered.
462: For adsorption from the gas phase, the energy cost of forming the free monolayer
463: must be taken into account.
464: The CNH molecule has a large dipole moment, as noted above.
465: In the case of 1 ML coverage the distance between molecules in the
466: monolayer is 2.883 $\AA$.
467: The point dipole approximation gives a crude estimate of about 1.4 eV per
468: molecule for the energy cost of monolayer formation.
469: This is much larger than the energy gained upon adsorption. Therefore HNC molecules
470: cannot adsorb on the (111) gold surface at 1 ML coverage.
471: For smaller coverages (1/3 and 1/4 ML) the distance between molecules is large enough for the point
472: dipole approximation to be accurate.
473: We calculate an energy cost of 0.26 eV per molecule to form a
474: CNH monolayer at 1/3 ML coverage and 0.17 eV at 1/4 ML coverage.
475: When compared to the adsorption energies in Table II,
476: the formation of 1/3 ML is endothermic, while the 1/4 ML coverage is
477: slightly exothermic, about 0.03 eV per molecule.
478:
479: Just as noted for the AuCNH complex, adsorption leads to a change in dipole moment,
480: due to both the charge transfer and the polarization
481: of the molecules and the gold surface.
482: For sparse coverages (less than 1/3), the induced dipole should not depend much on the
483: distance between molecules.
484: This extra dipole gives a coverage dependent contribution which is included in the
485: adsorption energy reported in Table II.
486: The scaling of this term is inversely
487: proportional to the cube of the intermolecular distance.
488: Using this approach and the data in Table II,
489: we extrapolate the adsorption energy in the dilute limit to be 0.23 eV
490: and the effective dipole moment of the adsorbed monolayer to be 3.35 D.
491: Thus the induced dipole moment in the monolayer is about 0.3 D, a considerably
492: smaller change than found for the AuCNH complex.
493: We also estimate the contribution to the binding that derives from
494: the classical polarization response of the metal surface to the
495: initial dipole on the CNH.
496: We use the point charge model mentioned previously for CNH
497: and assume a classical image potential to arrive at about 0.1 eV.
498: In comparison to the relatively small extrapolated net adsorption energy,
499: polarization plays a noticeable role.
500:
501: Given the extra repulsion from the induced dipole,
502: one might have expected the CNH to adopt a bent configuration on the surface.
503: In the radical, this dramatically reduces the dipole, as seen in Table I.
504: This has been investigated by considering small distortions of the CNH
505: away from the vertical a-top structure at 1/3 ML coverage (molecular axis was tilted
506: by 10$^{\circ}$, while the geometry of the molecule was otherwise unchanged).
507: Relaxation of this initial configuration results in an essentially vertical structure.
508: Alternatively, we have started from a bent CNH configuration at 1/4 ML coverage
509: (with Au-C bond vertical, CNH molecule positioned as in the bent Au-CNH complex)
510: and relaxed the structure.
511: We find a smooth path back towards the vertical geometry.
512: Evidently the back donation driving force, characterized further in the next section,
513: is weaker in the adsorbed case than in the radical.
514:
515: In the neutral radical, the net effect of the donor-acceptor bonding is relatively weak
516: due to the partially occupied Au s-orbital.
517: For an Au cation, or a complex which partially withdraws the Au s-electron, the
518: binding can be substantially stronger.
519: Experiments and calculations have been analyzed for a series of
520: ligands bonding to the gold cation \cite{schroder}.
521: The estimated bond energies for Au$^+$CO, Au$^+$NH$_3$, and Au$^+$CNCH$_3$
522: are 2.1, 3.1 and 3.1 eV respectively.
523: These are all substantially larger than for the corresponding neutral radicals.
524: Furthermore, it is known that isocyanides as well as amines and phosphines passivate
525: gold nanoparticles \cite{8, aminecapped,phosphinecapped}.
526: This raises the interesting question of the character of CNH bonding to
527: undercoordinated gold atoms on the surface.
528: To probe this question, we calculated the binding of CNH to an Au adatom
529: using ABINIT.
530: The adatom was relaxed, together with the first full surface layer,
531: in the hcp hollow site of a 4 ML slab with 1/4 coverage.
532: Then the CNH was added to the system and fully relaxed.
533: The final geometry is close to vertical with a reduced Au-C bond
534: length (1.958 $\AA$) and slightly longer C-N bond length (by $<$0.01 $\AA$)
535: relative to the original a-top geometry.
536: The binding energy
537: is increased by 0.9 eV, relative to the a-top adsorption site on the flat surface.
538: This binding energy refers to an array of CNH molecules in the same
539: unit cell relative to the gold surface including the adatoms.
540: The formation energy of the adatom is not explicitly included
541: since we consider the adatom to be a proxy for undercoordinated gold
542: atoms on the surface of a nanoparticle or other rough gold surfaces.
543:
544: Finally, it is well known that the DFT method using
545: the LDA overestimates binding energies.
546: However it is usually expected that
547: general trends are reproduced correctly.
548: Some calculations for the radicals were repeated with
549: the LDA exchange-correlation functional (Table 1).
550: As expected, LDA systematically overestimates binding energies by about 0.6 eV and
551: underestimates the Au-ligand bond lengths.
552: Also, some of the slab calculations of CNH adsorption on Au(111)
553: were repeated using LDA.
554: Adsorption energies calculated with ABINIT for a 1/4 ML coverage
555: are 0.95 eV, 1.09 eV and 1.12 eV for the top, hcp and fcc sites respectively.
556: Similar results are obtained with WIEN2k code for 1/3 ML coverage:
557: 0.80 and 1.03 eV for the top and hcp sites.
558: There is a qualitative difference between the GGA-PBE and LDA results, the GGA being more realistic where experimental
559: tests are available.
560: With GGA the top site is preferred and there is no adsorption at the hcp or fcc sites.
561: With LDA, on the contrary, adsorption at the hcp site is possible
562: and it is energetically favored over adsorption at the top site.
563:
564:
565:
566: \section{Discussion}
567:
568: We have found that isocyanides selectively bind to the atop site on the flat
569: gold (111) surface, with a small adsorption energy.
570: The C-N bond in the isocyanide is vertical.
571: This binding geometry is consistent with experiments described in the introduction:
572: top site adsorption is observed \cite{8,9} and ellipsometry data
573: indicate vertical orientation of adsorbed isocyanides \cite{7}.
574: At the present there are no measurements of binding energies
575: or saturation coverage of isocyanide molecules on the gold surface
576: for direct comparison to our calculated adsorption energy.
577: The CO molecule, on the other hand, is very well studied.
578: As noted, its electronic structure is similar to that of the CNH molecule.
579: We performed calculations identical to those described above for
580: CO on the gold (111) surface and obtained 0.2 eV for the adsorption energy,
581: in good agreement with previous calculations,
582: but less than the experimental value 0.4 eV \cite{trends}.
583:
584: The bonding of both CO and CNH to transition metals is often discussed in terms of the
585: Blyholder picture \cite{Blyholder}.
586: Binding involves donation from the filled lone pair on the C into
587: the partially occupied Au s shell together with $2\pi^*$ back donation into the
588: $2\pi^*$ molecular orbital due to interaction with gold $d_{xz}$ and $d_{yz}$.
589: We probe this picture for the bonding of CNH on the flat surface
590: by analyzing the angular character of the local density of states
591: for the adsorbed molecule and the gold slab.
592: The projected density of states (PDOS) for the monolayer adsorbed onto a
593: Au(111) slab at 1/3 coverage in the atop configuration is plotted in Fig. 2
594: along with the PDOS for a free CNH monolayer and a clean Au(111) slab.
595: Also shown are isosurface plots of selected wavefunctions ($\Gamma$-point)
596: corresponding to certain energy regions in the PDOS of the Au(111)-CNH system.
597:
598: In the PDOS of the free CNH monolayer several very sharp peaks are observed
599: which can be associated with the molecular orbitals $4\sigma$, $1\pi$, $5\sigma$ and
600: $2\pi^*$ (the last being unoccupied).
601: After adsorption the $4\sigma$ level does not change much
602: and there is no discernable interaction with gold.
603: This is not surprising considering that this level lies well below the bottom
604: of the gold valence band.
605: In contrast, $1\pi$ and $5\sigma$ levels undergo substantial reorganization.
606: The $5\sigma$ state strongly interacts with
607: hybridized $s$ and $d_{z^2}$ orbitals of gold.
608: A localized state is formed below the bottom of the gold valence band.
609: In addition, there is a state of anti-bonding character that can be identified
610: above the occupied gold d-band, but below the Fermi energy
611: (around -1 eV in Fig. 2).
612: These states are very similar to the hybrids that forms in the AuCNH radical.
613: In the radical, the anti-bonding state is half occupied.
614: As a result of the strong $\sigma$ interaction,
615: the $1\pi$ level now lies above $5\sigma$ derived level.
616: The $1\pi$ level interacts with gold $d_{xz}$ and $d_{yz}$ orbitals.
617: This results in two peaks visible in the PDOS at
618: around $-7$ eV and $-5$ eV, of bonding and anti-bonding character respectively,
619: as can be seen from the wavefunction plots.
620: The bonding combination appears to form a localized state just below the bottom
621: of the Au d-bands, while the anti-bonding combination is resonant with the d-bands.
622: Finally, the $2\pi^*$ peak from the CNH derived state density
623: extends somewhat below Fermi energy upon adsorption.
624: The corresponding wavefunction isosurface plot shows
625: that these occupied states are of bonding character between gold p-like states available
626: near the Fermi energy and the CNH $2\pi^*$ orbital.
627: Hence back donation into the $2\pi^*$ orbital contributes to the bonding.
628: Above the Fermi energy, near the center of the resonance, the coupling
629: is weak, but clearly of anti-bonding character to gold $d_{xz}$ and $d_{yz}$ orbitals.
630: This qualitative picture from the PDOS is very similar to previous results
631: for CO adsorbed on the a-top site of gold (111) \cite{trends}.
632:
633: It is hard to quantify this analysis of bonding character since the PDOS is calculated
634: using only electron densities inside spheres surrounding the atoms.
635: However, another metric related to bonding trends is the change in C-N bond
636: length upon adsorption on gold.
637: The calculation for the Au-CNH complex shows a C-N bond length of 1.173 $\AA$ in
638: the straight configuration.
639: Upon full relaxation to the bent configuration,
640: the C-N bond is 1.198 $\AA$, compared to 1.176 $\AA$ in the free CNH.
641: This net weakening of the C-N bond upon bending traces to the pseudo Jahn-Teller effect
642: driven by increased occupation of the antibonding 2$\pi^*$ orbital.
643: On the other hand, upon adsorption on the Au(111) surface,
644: the calculated C-N bond length is
645: slightly decreased to 1.16 $\AA$.
646: This contrasts with results for CO adsorption,
647: where the C-O bond length slightly increases
648: upon adsorption \cite{trends}.
649: The decrease in C-N bond length suggests slightly less $\pi$ back donation
650: for CNH adsorption on the surface relative to the radical.
651: However, careful study of metal-carbonyl complexes
652: reveals a more complex picture \cite{AuCO}.
653: The final C-N bond length (and the corresponding stretch frequency)
654: reflects the balance between $\sigma$ bond polarization due to the metal
655: and $\pi^*$ back donation effects.
656: This balance is affected by electrostatics, particularly in the limit of
657: the positively charged carbonyl complex.
658: The positive charge near the C influences the heteropolar C-O
659: $\sigma$ bond, driving it to be more symmetrical and stronger.
660: This also drives an increase in the C-O stretch frequency in the carbonyls.
661: One would expect similar effects on the C-N bond for the CNH case.
662: This is consistent with the observed increase
663: of the C-N stretch frequency in isocyanides adsorbed
664: on gold \cite{7,8}.
665:
666: The initial dipole on the molecule and the induced dipole upon adsorption
667: both play a role.
668: The CNH molecule has a substantial dipole moment, in contrast to the CO molecule.
669: Therefore, in the case of CNH, the binding
670: energy contains some contribution from the polarization energy due to the interaction
671: between the molecular dipole and the gold surface.
672: Our estimates indicated a modest effect (0.1 eV),
673: which is however a non-trivial fraction of the final adsorption energy
674: for an isolated CNH on the flat surface (0.23 eV).
675: Furthermore, the net dipole inhibits formation of a dense film on the surface.
676: The inferred induced dipole on the metal surface
677: is substantially smaller than that found in the isolated Au-CNH radical.
678: This is indicative of
679: less net charge transfer in comparison to the radical.
680: This is not surprising from two points of view.
681: First, the work function of Au(111) (5.31 eV \cite{michaelson}) is smaller
682: than the chemical potential of the Au atom
683: (conventionally the average of the ionization potential
684: and the electron affinity, 5.76 eV \cite{augas}).
685: This inhibits the sigma donation from the lone pair.
686: Second, the surface Au s-state is substantially involved in band formation, having nine
687: nearest neighbors.
688: This interferes with hybridization to the lone pair, reducing the net energy
689: gain for the sigma donation process.
690:
691: The role of charge transfer has been controversial in the analysis of
692: these weakly adsorbed systems and the impact of the hybridization
693: implicit in the Blyholder picture has been debated
694: \cite{trends, ammonia, pyridine, bipyridine}.
695: Bilic {\it et al.} argue that the main mechanisms
696: of binding of NH$_3$ to the gold surface are
697: polarization effects and dispersive interactions,
698: not the covalent bonding \cite{ammonia}.
699: In their work on pyridine \cite{pyridine}, they find more evidence for
700: charge transfer, but still argue that covalent effects are minimal.
701: A more recent study of pyridine binding to Au \cite{bipyridine} analyzed the impact of
702: Au coordination and suggested a more prominent role for hybridization effects.
703: The utility of the Blyholder picture becomes more apparent when examining
704: trends in the binding, e.g. for bonding to different metals \cite{trends}.
705: Alternatively, it gives a way to rationalize the trends for the binding
706: of different molecules to the same metal, as described for the Au radicals in Section II.
707: Finally, the sigma donation becomes more prominent for the case
708: of binding to an Au adatom on the surface.
709: We found a large increase in binding energy of CNH compared to the flat surface (0.9 eV).
710: A similar increase was recently reported for NH$_3$ (0.4 eV) \cite{venkataraman}.
711: The donor-acceptor binding
712: to the adatom is also enhanced due to the slight positive charge
713: on the Au adatom, similar to the increased binding energy for the Au(I) cation to
714: several ligands \cite{schroder}.
715:
716: The small calculated value of adsorption energy of isocyanides on
717: the flat gold surface seems to contradict the experimental observation of
718: SAM formation \cite{7}.
719: Two caveats are due here.
720: First, layer formation takes place in a solution.
721: Solvation effects can screen the dipole-dipole interactions
722: and facilitate layer growth from nuclei.
723: Also, a layer of larger molecules (e.g. alkanes)
724: may be more stable than the binding energy of the
725: CNH link moiety would suggest due to attractive inter-molecular interactions
726: between the extended molecules in the layer, e.g. due to dispersive interactions
727: between neighboring alkanes.
728: Second, the presence of undercoordinated Au binding sites at step edges
729: on the surface may facilitate nucleation of the isonitrile layers.
730: We have found that the binding energy to the Au adatom is substantially larger
731: than the binding energy to the top site on the flat surface.
732: This may also account for the binding to powdered gold samples and nanoparticles
733: where undercoordinated Au sites would be common.
734:
735: Assessing the accuracy of DFT calculations for these weakly bound systems
736: is difficult.
737: The information for the Au radicals was already assessed in Sec. II.
738: In the case of
739: Au-NH$_3$, the GGA Au to NH$_3$ binding energy
740: is smaller than accurate quantum chemistry calculations
741: by about 0.2 eV.
742: Bilic {\it et al.} studied
743: the binding of NH$_3$ to the gold surface
744: using the GGA with the PW91 exchange-correlation functional,
745: which in this particular case gave an adsorption energy in
746: quantitative agreement with experiment (within 0.1 eV) \cite{ammonia}.
747: For the Au-CO radical, the binding energy is less well established, as discussed
748: in Sec. II.
749: Gajdo {\it et al.} studied adsorption of
750: CO on (111) surface of various transition and noble metals \cite{trends}.
751: In particular, they report results for CO adsorption
752: on Au and Ag surfaces obtained with PW91 and
753: a revised form of the PBE exchange-correlation potentials
754: which has in other cases given more accurate adsorption energies \cite{hammer}.
755: Adsorption energies were very sensitive to the choice of the
756: functional.
757: Both functionals underestimate adsorption energies.
758: The revised PBE leads to endothermic adsorption while PW91 underestimates
759: the binding energy by about 0.2 eV.
760: The essential mechanisms of binding are very similar for CO and CNH molecules,
761: as it can be seen from the comparison of PDOS pictures.
762: The experience with CO on the flat Au surface suggests that the
763: present DFT calculations may underestimate the binding energy
764: for the CNH case as well.
765: Also, in the CNH case, we have found that the LDA overestimates
766: the radical binding energies and predicts the wrong binding site on the flat surface.
767: The role of dispersion forces in the binding
768: has been debated.
769: The evidence just summarized points to the GGA calculations
770: underestimating the binding energy of CO and CNH on the Au surface.
771: This is consistent with experience in the case of intermolecular interactions
772: for closed shell molecules which are dominated by dispersion forces.
773: The GGA approximations are known to underestimate
774: binding in these cases while LDA over estimates them \cite{meijer}.
775:
776:
777: \begin{figure*}
778: \includegraphics{fig2.eps}
779: \caption{Projected density of states onto:
780: (a) the HNC molecule in free-standing monolayer;
781: (b) the surface Au atom in the clean Au(111) slab;
782: (c) the Au atom bound to the molecule in the Au(111)-CNH system (1/3 ML, atop site);
783: (d) the HNC molecule in the Au(111)-CNH system.
784: Panel (e) shows isosurface plots of selected wavefunctions (at $\Gamma$-point)
785: of the Au(111)-CNH system at energies indicated by the arrows.
786: }
787: \end{figure*}
788:
789: \noindent {\bf Acknowledgements} We thank Tao Sun for help with the
790: WIEN2k calculations. We thank J. Davenport for help and for time
791: on computers at the Computational Science Center at Brookhaven National Laboratory. We thank K. K. Likharev
792: for use of Njal supercomputer cluster at Stony Brook. We thank M. R. Pederson and T. Baruah for
793: doing some useful molecular calculations. Work at Stony Brook was supported in part by
794: NSF Grant NIRT-0304122. Work at BNL was supported by U.S. DOE under contract No. DEAC 02-98 CH 10886.
795: Work at Columbia University was supported by the
796: Nanoscale Science and Engineering Initiative of the National Science Foundation
797: under NSF Award Number CHE-0117752 and by the New York State Office of Science,
798: Technology, and Academic Research (NYSTAR).
799:
800:
801: \begin{thebibliography}{99}
802:
803: \bibitem{nuzzo} R.G. Nuzzo and D.L. Allara, J. Am. Chem. Soc. {\bf 105},
804: 4481 (1983).
805:
806: \bibitem{ulman} A. Ulman, Chem. Rev. {\bf 96}, 1533 (1996).
807:
808: \bibitem{schreiber04} F. Schreiber, J. Phys.: Condens. Matter {\bf 16}, R881 (2004).
809:
810: \bibitem{tour} L. Jones, II, J.S. Schumm, and J.M. Tour, J. Organic Chem. {\bf 62},
811: 1388 (1997).
812:
813: \bibitem{reed} M.A. Reed, C. Zhou, C.J. Muller, T.P. Burgin, and J.M. Tour,
814: Sci. {\bf 278}, 252 (1997).
815:
816: \bibitem{salomon} A. Salomon, D. Cahen, S. Lindsay, J. Tomfohr, V.B. Engelkes,
817: and C.D. Frisbie, Adv. Mater. {\bf 15}, 1881 (2003).
818:
819: \bibitem{basch} H. Basch, R. Cohen, and M.A. Ratner, Nano Lett. {\bf 5}, 1668 (2005).
820:
821: \bibitem{poirrer} G.E. Poirer, Chem. Rev. {bf 97}, 1117 (1997).
822:
823: \bibitem{schreiber00} F. Schreiber, Prog. Surf. Sci. {\bf 65}, 151 (2000).
824:
825: \bibitem{fischer} D. Fischer, A. Curioni, and W. Andreoni, Lang. {\bf 19}, 3567 (2003).
826:
827: \bibitem{nuckolls} G.S. Tulevski, M.B. Myers, M.S. Hybertsen, M.L. Steigerwald,
828: and C. Nuckolls, Sci. {\bf 309}, 591 (2005).
829:
830: \bibitem{mcbreen} M. Siaj and P.H. McBreen, Sci. {\bf 309}, 588 (2005).
831:
832: \bibitem{venkataraman} L. Venkataraman, J.E. Klare, I.W. Tam,
833: C. Nuckolls, M.S Hybertsen and M. Steigerwald, Nano Lett. {\bf 5}, 458 (2006).
834:
835: \bibitem{lin} S. Lin and R.L. McCarley, Lang. {\bf 15}, 151 (1999).
836:
837: \bibitem{8} M. J Robertson, R. J. Angelici, Langmuir {\bf 10}, 1488, (1994).
838:
839: \bibitem{9} K.Shih, R. J. Angelici, Langmuir {\bf 11}, 2539 (1995).
840:
841: \bibitem{10} S.J. Bae, Ch. Lee, I.S. Choi, Ch.-S. Hwang, M. Gong, K. Kim,
842: and S.-W. Joo, J. Phys. Chem. B {\bf 106}, 7076 (2002).
843:
844: \bibitem{11} H. S. Kim, S. J. Lee, N. H. Kim, J. K. Yoon, H. K. Park,
845: and K. Kim, Langmuir {\bf 19}, 6701 (2003).
846:
847: \bibitem{7} J. I. Henderson, S. Feng, T. Bein, and C. P. Kubiak,
848: Langmuir {\bf 16}, 6183 (2000).
849:
850: \bibitem{13} J. M. Seminario, A.G. Zacarias, and J.M. Tour, J. Am. Chem. Soc.
851: {\bf 121}, 411 (1999).
852:
853: \bibitem{1} J. Chen, L.C. Calvet, M.A. Reed, D.W. Carr, D.S. Grubisha,
854: D.W. Bennett, Chem. Phys. Lett. {\bf 313}, 741 (1999).
855:
856: \bibitem{2} J.-O Lee, G. Lientschnig, F. Wiertz, M. Struijk, R.A.J. Janssen,
857: R. Egberink, D.N. Reinhoudt, P. Hadley, and C. Dekker,
858: Nano Letters, {\bf 3}, 113 (2003).
859:
860: \bibitem{3} Y. Xue and M. A. Ratner, Phys. Rev. B {\bf 69}, 085403 (2004).
861:
862: \bibitem{trends} M. Gajdos, A. Eichler and J. Hafner,
863: J. Phys. Condens. Matter {\bf 16}, 1141 (2004).
864:
865: \bibitem{ammonia} A. Bilic, J. R. Reimers, N. S. Hush, and J. Hafner, J. Chem. Phys. {\bf 116}, 8981 (2002)
866:
867: \bibitem{Blyholder} G. Blyholder, J. Phys. Chem. {\bf 68}, 2772 (1964).
868:
869: \bibitem{PBE} J. P. Perdew, K. Burke, and M. Ernzerhof,
870: Phys. Rev. Letters {\bf 77}, 3865 (1996).
871:
872: \bibitem{hammer} B. Hammer, L.B. Hansen, and J.K. Norskov, Phys. Rev. B
873: {\bf 59}, 7413 (1999).
874:
875: \bibitem{PBEtest} S. Kurth, J. P. Perdew, P. Blaha, Int. J. of Quant. Chem. {\bf 75}, 889 (1999).
876:
877: \bibitem{feibelman} P. J. Feibelman, B. Hammer, J. K. Norskov, F. Wagner, M. Scheffler, R. Stumpf, R. Watwe, and J. Dumesic, J. Phys. Chem. B {\bf 105}, 4018 (2001).
878:
879: \bibitem{gross} A. Gross, Surf. Sci. Rep. {\bf 32}, 291 (1998).
880:
881: \bibitem{LDA}
882: The parameterization of Ref. \cite{goedecker} is used with
883: the Abinit slab calculations. Perdew and Wang parametrization \cite{PW92} is used in NRLMOL and WIEN2k
884: calculations.
885:
886: \bibitem{PW92} J.P. Perdew and Y. Wang, Phys.Rev. B {\bf 45}, 13244 (1992).
887:
888: \bibitem{goedecker} S. Goedecker, M. Teter, and J. Hutter, Phys. Rev. B
889: {\bf 54}, 1703 (1996).
890:
891: \bibitem{NRLMOL} M.R. Pederson and K.A. Jackson, Phys. Rev. B {\bf 41}, 7453
892: (1990); K.A. Jackson and M.R. Pederson, Phys. Rev. B
893: {\bf 42}, 3276 (1991).
894:
895: \bibitem{NRLMOLpsp} D. Porezag and M.R. Pederson, Phys. Stat. Solidi (b) {\bf 217}, 219 (2000).
896:
897: \bibitem{NRLMOLbasis} D. Porezag and M.R. Pederson, Phys. Rev. A, {\bf 60}, 2840 (1999); D. V. Porezag, PhD thesis: http://archiv.tu-hemnitz.de/pub/1997/0025.
898:
899: \bibitem{12} P. Blaha, K. Schwarz, G. K. H. Madsen, D. Kvasnicka and J. Luitz,
900: WIEN2k, An Augmented Plane Wave + Local Orbitals Program for Calculating
901: Crystal Properties (Karlheinz Schwarz, Techn. Universitat Wien, Austria),
902: 2001. ISBN 3-9501031-1-2
903:
904: \bibitem{abinit} The ABINIT code is a common project of the Université Catholique de Louvain, C. I., and other contributors. URL http://www.abinit.org.
905:
906: \bibitem{gonze} X. Gonze, J.M. Beuken, R. Caracas, F. Detraux, M. Fuchs,
907: G.M. Rignanese, L. Sindic, M. Verstraete, G. Zerah, F. Jollet, M. Torrent,
908: A. Roy, M. Mikami, P. Ghosez, J.Y. Raty, and D.C. Allan,
909: Computational Materials Science {\bf 25}, 478 (2002).
910:
911: \bibitem{HGH} C. Hartwigsen, S. Goedecker, and J. Hutter, Phys. Rev. B { \bf 58}, 3641 (1998).
912:
913: \bibitem{TM} N. Troullier and J.L. Martins, Phys. Rev. B { \bf 43}, 8861 (1991).
914:
915: \bibitem{FHI} M. Fuchs and M. Scheffler, Computer Physics Communications { \bf 119}, 67 (1999).
916:
917: \bibitem{tetrahedron} P. E. Blochl, O. Jepsen and O. K. Andersen, Phys. Rev B {\bf 49}, 16223 (1994).
918:
919: \bibitem{lambropoulos} N.A. Lambropoulos, J.R. Reimers, and N.S. Hush, J. Chem. Phys.
920: {\bf 116}, 10277 (2002).
921:
922: \bibitem{dargel} T.K. Dargel, R.H. Hertwig, W. Koch, and H. Horn,
923: J. Chem. Phys. {\bf 108}, 3876 (1998).
924:
925: \bibitem{schwerdtfeger} P. Schwerdtfeger and G.A. Bowmaker, J. Chem. Phys.
926: {\bf 100}, 4487 (1994).
927:
928: \bibitem{liang} B. Liang and L. Andrews, J. Phys. Chem. A
929: {\bf 104}, 9156 (2000).
930:
931: \bibitem{mendizabal} F. Mendizabal, Organometallics
932: {\bf 20}, 261 (2001).
933:
934: \bibitem{wu} X. Wu, L. Senapati, S.K. Nayak, A. Selloni, and M. Hajaligol,
935: J. Chem. Phys. {\bf 117}, 4010 (2002).
936:
937: \bibitem{schroder} D. Schroder, H. Schwarz, J. Hrusak, and P. Pyykko,
938: Inorg. Chem., {\bf 37}, 624 (1998).
939:
940: \bibitem{aminecapped} D. V. Leff, L. Brandt, and J. R. Heath, Langmuir {\bf 12}, 4723 (1996).
941:
942: \bibitem{phosphinecapped} W. W. Weare, S. M. Reed, M. G. Warner, and J. E. Hutchison, J. Am. Chem. Soc. {\bf 122}, 12890 (2000).
943:
944: \bibitem{AuCO} A.J. Lupinetti, S. Fau, G. Frenking, and S.H.Strauss, J. Phys. Chem. A {\bf 101}, 9551 (1997).
945:
946: \bibitem{michaelson} H. B. Michaelson, J. Appl. Phys. {\bf 48}, 4729 (1977).
947:
948: \bibitem{augas} J. M. Dyke, N. K. Fayad, A. Morris, and I. R. Trickle,
949: J. Phys. B {\bf 12}. 2985 (1979);
950: G. Gantefor, S. Kraus, and W. Eberhardt, J. Electron Spectrosc. Relat.
951: phenom. {\bf 88}, 35 (1998).
952:
953: \bibitem{pyridine} A. Bilic, J.R. Reimers, and N.S. Hush,
954: J. Phys. Chem. B {\bf 106}, 6740, (2002).
955:
956: \bibitem{bipyridine} R. Stadler, K.S. Thygesen, and K.W. Jacobsen, Phys. Rev. B
957: {\bf 72}, 241401(R) (2005).
958:
959: \bibitem{meijer} E.J. Meijer and M. Sprik, J. Chem. Phys.
960: {\bf 105}, 8684 (1996).
961:
962: \end{thebibliography}
963:
964: \end{document}
965:
966:
967: