1: \documentclass[aps,twocolumn,showpacs]{revtex4}
2: \usepackage{graphicx}
3:
4: \begin{document}
5:
6: \title{Carbon cage-like materials as potential
7: low work function metallic compounds: \\
8: Case of clathrates.}
9:
10: \author{V. Timoshevskii}
11: \email{vladimir@dpm.univ-lyon1.fr}
12: \author{D. Conn\'etable}
13: \author{X. Blase}
14: \affiliation{D\'epartement de Physique des Mat\'eriaux and CNRS,\\
15: Universit\'e Claude Bernard-Lyon 1, B\^at. 203, 43 Bd du 11
16: Novembre 1918, \\ 69622 Villeurbanne C\'edex, FRANCE }
17:
18: \begin{abstract}
19: We present an {\it ab-initio} calculation of the electronic affinity of the
20: hypothetical C-46 clathrate by studying its bare and hydrogenated (100)
21: surfaces. We show that such a system shares with the diamond phase
22: a small electronic affinity. Further, contrary to the diamond phase,
23: the possibility of doping endohedrally these cage-like systems allows
24: to significantly raise the position of the Fermi level, resulting in a true
25: metal with a small work function. This is illustrated in the case of the
26: Li$_8$@C-46 doped compound. Such a class of materials might be of much
27: interest for the design of electron-emitting devices.
28: \end{abstract}
29:
30: \pacs{71.20.Mq, 71.15.Mb, 71.24.+q}
31:
32: \maketitle
33:
34: %% Introduction: generality about emitting materials
35:
36: There is a strong incentive in trying to synthesize materials which
37: could be used as efficient electron-emitters for vacuum electronic devices.
38: While alkali metals display the lowest work function (W$\sim$2-3 eV) of all
39: elemental metals, their poor mechanical and chemical properties (melting
40: temperature, resistance to ionic bombardment, etc) forbid their use in actual
41: devices, and therefore metals with much higher work function such as tungsten
42: or molybdinum (W$\sim$4.5 eV) are preferred for their superior mechanical
43: and chemical properties.
44:
45: %%
46:
47: A promising class of materials which have been extensively studied
48: are carbon diamond based systems. Due to their large gap,
49: carbon diamond based
50: systems have been shown to display a small, or even negative,
51: electronic affinity \cite{Weide94,Pickett94}. Further, their mechanical
52: properties are superior to the one of any existing metal. However,
53: the difficulty of doping them "n-type" in order to obtain a significant density
54: of emitting electrons limits their efficiency as good emitters.
55:
56: %% Introduction: clathrates
57:
58: In the last few years, a novel structure has attracted much
59: attention for its specific structural and electronic properties.
60: Column-IV clathrates \cite{Kasper65} are cage-like materials
61: composed of face-sharing X$_{20}$, X$_{24}$ and X$_{28}$ clusters
62: (where X=Si,Ge). All atoms are in the same sp$^3$ environment as
63: in the diamond phase, but the network is composed of 87$\%$ of
64: pentagonal rings. As a result, it has been shown in particular
65: that Si clathrates present a band gap which is $\sim$ 0.7 eV
66: larger than the one of the diamond phase \cite{Adams94,Gryco00}. Furthermore,
67: the possibility of filling each cage by a dopant atom allows to
68: significantly raise the position of the Fermi level to yield a
69: metallic system with a large density of states (eDOS) at the Fermi
70: level (E$_F$) \cite{Saito95,Smelyansky97}.
71:
72:
73: %% Outline
74: The possibility of efficiently doping such systems has incited us to study
75: the case of the hypothetical carbon clathrates. Even though such phases
76: have not been synthesized so far, the reports on the synthesis
77: of connected C$_{36}$ clusters \cite{Piskoti98}, of diamond-like thin films
78: synthesized by deposition of small carbon clusters \cite{Paillard93} or even
79: of polymerized C$_{60}$ suggests that the synthesis of such phases might be at
80: hand or that the results presented below may apply to the above-mentioned
81: existing phases.
82:
83: \begin{figure}
84: \begin{center}
85: \includegraphics[width=8.5cm]{Figure1.eps}
86: \end{center}
87: \caption{ Symbolic representation of the Li$_8$@C-46 (100):H slab
88: unit cell. The structure consists of C$_{20}$ and C$_{24}$
89: clusters with carbon atoms (not shown here) located at the corners
90: of the cages. The gray balls show Li atoms encapsulated inside the
91: carbon clusters. The black balls represent hydrogen atoms on the
92: surface of the slab. The same type of slab was used for C-46 (100)
93: and C-46 (100):H surfaces.} \label{slab}
94: \end{figure}
95:
96: %% Outline
97: We present in this Letter an {\it
98: ab-initio} study of the electronic properties of C-46 and
99: Li$_8$@C-46 within the density functional theory (DFT)
100: \cite{Hohen-Kohn-Sham}. We study in particular the C-46 (100)
101: bare and hydrogenated surfaces and analyze the effect of doping on
102: the position of the Fermi level, the affinity of the system and
103: the density of states at E$_F$. We show that under doping
104: such a class of materials
105: are true metals with a low work function, thus combining
106: the advantages of diamond-like systems with a large carrier
107: concentration.
108:
109: %% Technical details
110: The local-density approximation (LDA) \cite{CA-PZ}
111: was used for all calculations and
112: a pseudopotential approach was adopted \cite{TM-KB}. We used the
113: {\sc Siesta} package \cite{siesta}, which is a self-consistent
114: DFT code, employing numerical atomic-like orbitals (NAO) as a
115: basis set. A well-converged basis set, consisting of doubled
116: {\it \{s,p$_x$,p$_y$,p$_z$\}} orbitals plus polarization {\it
117: d}-orbitals was applied. Some of the calculations were also
118: performed using a standard plane-wave (PW) basis set.
119: All systems studied below were fully relaxed, both with
120: respect to cell size and atomic positions. To perform the
121: electronic affinity analysis, we calculated the plane-averaged
122: self-consistent potential in the direction perpendicular to the
123: surface for each of the slabs. The positions of the valence-band
124: maximum (VBM) and conduction-band minimum (CBM) were found by
125: adding the energy difference between the average self-consistent
126: potential in the bulk and the VBM (CBM) in the bulk to the average
127: self-consistent potential inside the slab.
128:
129:
130: \begin{table}
131: \caption{Calculated electronic affinity for the diamond and
132: clathrate carbon surfaces (eV).}
133: \begin{ruledtabular}
134: \begin{tabular}{lccr}
135: {surface} & LDA-PWs & LDA-NAOs & GW \\
136: \hline \\
137: C (100) & 2.17 & 1.99 & 1.3 \\
138: C (100):H & -0.45 & -0.59 & -1.3 \\
139: C-46 (100) & & 3.47 & 2.8 \\
140: C-46 (100):H & & 1.47 & 0.8 \\
141: Li$_8$@C-46 (100):H & & 2.18 & $\sim$ 1.5 \\
142: \end{tabular}
143: \end{ruledtabular}
144: \label{table}
145: \end{table}
146:
147:
148: %% Comparison PWs and LCAO
149:
150: Before studying the C-46 clathrate phase, we have tested the method by
151: exploring the electronic and structural properties for the well-known
152: bare and hydrogenated diamond (100) surfaces. Following
153: Ref.\onlinecite{Weide94}, we have adopted a slab geometry with
154: 10-layers but with 2-atoms per layer. The $\bf
155: k$-point sampling was thus increased to a 4$\times$4$\times$1 special
156: grid \cite{MonkPack}. A vacuum of 10 \AA~was kept between
157: neighboring slabs.
158:
159:
160: The results for PW and NAO-basis are reported in Table
161: \ref{table}. As in Ref.\onlinecite{Weide94}, we observe that the
162: hydrogenation, by building a surface dipole, leads to a negative
163: affinity system. Our results are in good agreement with those
164: previously published \cite{Weide94}, and the NAO-basis
165: calculations reproduce within $\sim$ 0.15 eV the results of
166: the PW calculations.
167:
168: %% GW calculations
169:
170: As the DFT-LDA is well known to underestimate the band gap of
171: insulators, we have performed further a quasiparticle study within
172: the GW approximation \cite{Hybertsen} for the bulk diamond and
173: C-46 phases. Such an approach is known to yield quasiparticle
174: energies to within 0.1 eV as compared to experimental
175: photoemission data. More details about the formalism and results
176: will be provided elsewhere \cite{xblase}. The main outcome of such
177: calculations is that the quasiparticle correction to the DFT-LDA
178: eigenstates are very similar in both systems and that the CBM is
179: pushed to higher energy by $\sim$ 0.7 eV in both phases (while the
180: VBM is corrected down by $\sim$ 0.8 eV). This allows to correct
181: the calculated affinity as reported in the right column of Table
182: \ref{table}.
183:
184: \begin{figure}
185: \begin{center}
186: \includegraphics[width=8.5cm]{Figure2.eps}
187: \caption{ Plot of the plane-averaged self-consistent potential for
188: the C-46(100) bare and hydrogenated surfaces, and for
189: Li$_8$@C-46(100):H surface. The good agreement between the bulk
190: and slab potential on the innermost atoms of Li$_8$@C-46(100):H
191: indicates that the slab is large enough.} \label{potential}
192: \end{center}
193: \end{figure}
194:
195: %% Undoped clathrates
196:
197: The stability and electronic properties of C-34 and C-46
198: clathrates have already been discussed in some details at the
199: DFT-LDA level \cite{Adams94,Nesper93,SanMiguel99,Saito97}. In
200: particular, it has been shown that for the C-34 and C-46
201: clathrates, the cohesive energy, bulk modulus and band gap are
202: respectively $\sim$ 0.1 eV/atom, $\sim$ 10 $\%$ and $\sim$ 0.2 eV
203: smaller than those of the diamond phase. As such, the stability
204: and mechanical properties of carbon clathrates are close to those
205: of the diamond phase. In the present work, we obtained for the
206: lattice parameter, bulk modulus and band gap the values of 6.64
207: \AA, 376 GPa and 3.75 eV respectively with both NAO and PW basis
208: sets, which is in good agreement with previously reported results.
209: Within the GW approximation, the gaps increase to 5.55 and 5.35 eV
210: for the diamond and clathrate phases respectively.
211:
212:
213: We now study the C-46 (100) surface represented in Fig.\ref{slab}.
214: There are several non-equivalent (100) surfaces and we adopt
215: a surface consisting of "half emerging" C$_{24}$ clusters. After structural
216: relaxation, we find that the emerging semi-clusters hardly reconstruct
217: besides a slight contraction inward of the three-fold coordinated
218: surface atoms, leading to a minimum C-C distance of 1.40 \AA~ on
219: the surface. Hydrogen coverage is performed by "passivating" these
220: surface three-coordinated atoms. We note at this point that the results
221: presented below may certainly change quantitatively with the chosen
222: surface but the purpose of the present paper is to illustrate on
223: this specific example the potentialities of such novel systems.
224: Fig.\ref{potential} shows the calculated plane-averaged,
225: self-consistent potential for bare and hydrogenated C-46 (100)
226: surfaces. The electronic affinity is reported in Table
227: \ref{table}.
228:
229: \begin{figure}
230: \begin{center}
231: \includegraphics[width=8cm]{Figure3.eps} \caption{ Electronic density of states (eDOS) for
232: bulk C-46 and Li$_8$@C-46. A 0.1 eV broadening has been used. The
233: eDOS is normalized per carbon atom.} \label{dos}
234: \end{center}
235: \end{figure}
236:
237: %% Doped clathrates
238: The results obtained for the undoped compounds suggest that carbon
239: clathrates are {\it a priori} less favorable for emission than,
240: say, the (100) diamond surface, taking the electronic affinity as
241: the only relevant criteria. However, we emphasize again that
242: clathrates, contrary to the diamond phase, can be efficiently
243: doped \cite{Saito97,Bernasconi00} by intercalating one atom in each cage
244: in order to obtain a true metallic system with a large density of
245: electrons available for emission. This is a crucial advantage as
246: compared to the diamond phase.
247:
248:
249: %%
250: By analogy with the existing doped Si clathrates and intercalated
251: carbon graphitic systems, we study the {\it n}-doped Li$_8$@C-46 phase.
252: The corresponding bulk eDOS is represented in Fig.\ref{dos} where
253: we have arbitrarily aligned the eDOS to the top of the C-46
254: valence bands for sake of comparison.
255: As can be readily seen, intercalation of Li atoms
256: leads to the filling of the conduction bands of the host carbon
257: matrix by Li 2s electrons. The system becomes metallic with the
258: Fermi level located near a peak in the density of states, which is in
259: excellent agreement with previous work \cite{Saito97}.
260:
261: %%
262: As in the case of diamond surfaces, hydrogenation leads to a significant
263: decrease in the electronic affinity of the empty phase.
264: We now explore the effect of doping on the electronic affinity
265: (Table \ref{table}) of the hydrogenated surface.
266: The important finding is that Li doping leads to
267: the formation of a true metallic system with a quite low
268: work function ($\sim$ 1.5 eV GW value).
269: This work function is significantly lower
270: than the one of alkali metals and much
271: lower than the one of e.g. tungsten. The eDOS for
272: states located around E$_F$ can be further shown to have a strong
273: weight on both bulk and surface atoms,
274: a factor which is crucial for emission.
275: One can conclude that doped carbon clathrates are potentially
276: true metals with a work function lower than the
277: lowest affinity of all elemental metallic systems but with much
278: superior mechanical properties.
279:
280: %%
281: It is interesting to note that the work function of the Li-doped
282: clathrate is actually larger than the electronic affinity of the
283: empty phase, despite the raise of the Fermi level under doping.
284: This means that the entire band structure
285: is shifted to lower energy with respect to the vacuum level under
286: doping. This shift can originate in two effects: a
287: decrease of the average potential inside the slab (bulk effect)
288: and an increase of the surface dipole barrier (surface effect).
289: An analysis of both effects show that in the present case it is
290: mainly a bulk effect.
291:
292:
293: %% Conclusion
294:
295: In conclusion, we have shown that doped carbon clathrates are
296: potentially true metallic systems with a work function significantly
297: lower than the one of alkali metals but with a stability
298: typical of the carbon diamond phase. This opens new perspectives
299: for the making of efficient electron-emitters. An alternative choice
300: of dopant atom and/or surface treatment (e.g. cesium coverage) may
301: lead to even smaller work function.
302:
303:
304: \acknowledgments Computations have been performed on the
305: supercomputer facilities (IDRIS) at the French "Centre National
306: de Recherche Scientifique" (CNRS). One of the
307: authors (VT) acknowledges a ``R\'egion Rh\^one-Alpes"
308: Postdoctoral Fellowship. The authors are indebted to Stephen Purcell
309: and C. Constancias (OpsiTech, Grenoble, France) for stimulating discussions.
310:
311:
312: %% REFERENCES
313: \begin{references}
314:
315: \bibitem{Weide94}
316: J. van der Weide, Z. Zhang, P.K. Baumann, M.G. Wensell, J. Bernholc,
317: and R.J. Nemanich,
318: Phys. Rev. Lett. {\bf 50}, 5803 (1994);
319:
320: \bibitem{Pickett94}
321: Warren E. Pickett, Phys. Rev. Lett. {\bf 73}, 1664 (1994).
322:
323: \bibitem{Kasper65}
324: J. Kasper, P. Hagenmuller, M. Pouchard, C. Cros, {\it Science} {\bf 150}, 1713
325: (1965).
326:
327: \bibitem{Adams94}
328: G.B. Adams, M. O'Keeffe, A.A. Demkov, O.F. Sankey, and Y.-M. Huang,
329: Phys.Rev.B {\bf 49}, 8048 (1994).
330:
331: \bibitem{Gryco00}
332: J. Gryco {\it et al.}, Phys. Rev. B {\bf 62}, R7707 (2000).
333:
334: \bibitem{Saito95}
335: S. Saito, A. Oshiyama, Phys. Rev.B {\bf 51}, 2628 (1995).
336:
337: \bibitem{Smelyansky97}
338: V.I. Smelyanski and J.S. Tse, Chem. Phys. Lett. {\bf 264}, 459
339: (1997).
340:
341: \bibitem{Piskoti98}
342: %% C36, a new carbon solid.
343: C. Piskoti, J. Yarger, and A. Zettl, {\it Nature} {\bf 393}, 771 (1998).
344:
345: \bibitem{Paillard93}
346: % Diamond-like films obtained by low energy cluster beam deposition: evidence of a
347: % memory effect of the properties of free carbon clusters.
348: V. Paillard, P. M\'elinon, V. Dupuis, J.P. Perez {\it et al.},
349: Phys. Rev. Lett. {\bf 71}, 4170 (1993).
350:
351: \bibitem{Hohen-Kohn-Sham}
352: P. Hohenberg and W. Kohn, Phys. Rev. B {\bf 136}, 864 (1964);
353: W. Kohn and L. J. Sham, Phys. Rev. A {\bf 140}, 1133 (1965).
354:
355: \bibitem{CA-PZ}
356: D. M. Ceperley and B. J. Alder, Phys. Rev. Lett. {\bf 45}, 566 (1980);
357: J.P. Perdew and A. Zunger, Phys. Rev. B {\bf 23}, 5048 (1981).
358:
359: \bibitem{TM-KB}
360: N. Troullier and J.L. Martins, Phys. Rev. B {\bf 43}, 1993 (1991);
361: K. Kleinman and D.M. Bylander, Phys. Rev. Lett. {\bf 48}, 1425 (1982).
362:
363: \bibitem{siesta}
364: P. Ordejo\'n, E. Artacho and J.M. Soler, Phys. Rev. B {\bf 53}, R10441 (1996);
365: D. Sa\'nchez-Portal, P. Ordejo\'n, E. Artacho and J.M. Soler, Int. J. Quantum
366: Chem. {\bf 65}, 453 (1997).
367:
368: \bibitem{MonkPack}
369: H.J. Monkhorst and J.D. Pack, Phys. Rev. B {\bf 13}, 5188 (1976).
370:
371: \bibitem{Hybertsen}
372: M. Hybertsen and S.G. Louie, Phys. Rev. B {\bf 30}, 5777 (1984); and
373: references therein.
374:
375: \bibitem{xblase}
376: In the case of the Li$_8$@C-46, the quasiparticle correction to E$_F$ is taken
377: to be the same as the one for the CBM of C-46. While in the conduction bands
378: the GW correction can be shown to slightly increase with the energy of the
379: states (see Ref.~\onlinecite{Hybertsen}), the enhanced screening
380: under doping is expected to reduce it. Despite such a compensation, the
381: quasiparticle correction to E$_F$ should be taken as an approximate value.
382: X. Blase, to be published.
383:
384: \bibitem{Nesper93}
385: R. Nesper, K. Vogel, and P. Bl\"ochl, Angew. Chem. {\bf 32}, 701 (1993).
386:
387: \bibitem{SanMiguel99}
388: A. San-Miguel, P. K\'egh\'elian, X. Blase, P. M\'elinon {\it et al.},
389: Phys. Rev. Lett. {\bf 83}, 5290 (1999).
390:
391: \bibitem{Saito97}
392: S.Saito, Proceedings of the first symposium on Atomic-Scale Surface
393: and Interface Dynamics, March 13-14, 1997, Tokyo.
394:
395: \bibitem{Bernasconi00}
396: M. Bernasconi, S. Gaito, G. Benedek,
397: Phys. Rev. B {\bf 61}, 12689 (2000).
398: \end{references}
399:
400:
401: \end{document}
402: