1: \documentclass[prl,twocolumn,showpacs]{revtex4}
2: \usepackage{graphics}
3: \usepackage{dcolumn}
4: \begin{document}
5: \title{Enhancing T$_c$ in field-doped Fullerenes by applying uniaxial stress}
6: \author{Erik Koch}
7: \email{E.Koch@fkf.mpg.de}
8: \affiliation{Max-Planck Institut f\"ur Festk\"orperforschung,
9: Heisenbergstra\ss e 1, 70569 Stuttgart, Germany}
10: \date{\today}
11: \begin{abstract}
12: Capitalizing on the two-dimensional nature of superconductivity in
13: field-effect doped C$_{60}$, we show that it should be possible to
14: increase the transition temperature $T_c$ by applying uniaxial stress
15: perpendicular to the gate electrode. This method not only holds the
16: promise of substantially enhancing $T_c$ (by about 30 K per GPa),
17: but also provides a sensitive check of the current understanding
18: of superconductivity in the doped Fullerenes.
19: \end{abstract}
20: % 74. Superconductivity
21: % 74.70.Wz Fullerenes and related materials
22: % 74.62.Fj Pressure effects
23: \pacs{74.70.Wz,74.62.Fj}
24: \maketitle
25:
26: In a remarkable series of experiments Sch\"on and collaborators have
27: demonstrated superconductivity in field-effect devices based on Fullerene
28: crystals. In a first step they field-doped pure C$_{60}$ with
29: electrons, observing a maximum $T_c$ of 11 K at a doping of about 3 electrons
30: per molecule in the layer under the gate electrode \cite{elecdoped}.
31: Reversing the polarity of the gate voltage, they achieved hole-doping, finding
32: superconductivity with a transition temperature of up to 52 K at a doping level
33: of about 3 to 3.5 holes per molecule \cite{holedoped}. In the most recent
34: experiments, they observed superconductivity by field-doping C$_{60}$ crystals
35: that were intercalated with chloroform, CHCl$_3$, and CHBr$_3$
36: \cite{latticeexp}. In these crystals, the volume per C$_{60}$ molecule is
37: increased due to the presence of the intercalated molecules. Superconductivity
38: was observed for electron- as well as for hole-doping and it was found that in
39: both cases $T_c$ increases almost linearly with the distance between
40: neighboring C$_{60}$ molecules, reaching 80 K for C$_{60}\cdot$2CHCl$_3$
41: and 117 K for C$_{60}\cdot$2CHBr$_3$. After this discovery a 'race to beat
42: the cuprates' has been announced \cite{dagotto}, and the search is on
43: for ways to further increase the transition temperature in the Fullerenes.
44: Here we show that a feasible method for doing so is the application of
45: uniaxial stress perpendicular to the gate electrode. While this proposal
46: might at first seem counterintuitive, we will demonstrate that it is a natural
47: consequence of the two-dimensional nature of superconductivity in the
48: field-doped Fullerenes.
49:
50: So far superconductivity in field-doped Fullerenes has been discussed
51: in close analogy to alkali-doped C$_{60}$. This was motivated by the
52: remarkable similarities between these two classes of materials \cite{gunnar}:
53: % electron-doped field-effect devices and
54: %the alkali-doped crystals A$_n$C$_{60}$ \cite{gunnar}:
55: (i) in both $T_c$ is largest when the $t_{1u}$ band is half-filled, and
56: (ii) $T_c$ increases with the distance between the C$_{60}$ molecules in
57: the crystal, which is explained as a consequence of the corresponding increase
58: in the density of states at the Fermi level. Experimentally it is found that
59: for a given nearest-neighbor distance the transition temperature for the
60: alkali- and the electron-field-doped Fullerenes are almost identical
61: \cite{latticeexp}.
62: Nevertheless, there is a fundamental difference between the two classes of
63: materials: the alkali-doped Fullerenes are bulk-superconductors, while in
64: the field-effect devices superconductivity is restricted to two dimensions.
65: %
66: It appears that in the field-doped Fullerenes only the first monolayer
67: under the gate electrode is doped. This is confirmed by tight-binding
68: simulations where it is found that that the deeper layers carry
69: negligible charge density \cite{wehrli}. Experimentally it is supported
70: by the fact that for the increase in $T_c$ upon intercalation of CHCl$_3$
71: and CHBr$_3$ only the change in the nearest neighbor distance seems to matter,
72: while the change in crystal structure appears to have no effect: At the
73: temperatures where superconductivity occurs, pure C$_{60}$ is simple cubic
74: with four molecules (sitting on fcc sites) per unit cell \cite{david}, while
75: the intercalated crystals are (almost) hexagonal \cite{jansen,nodos}. This
76: suggests that superconductivity in field-doped Fullerenes is restricted to a
77: single, triangular layer: a (111) plane in C$_{60}$ and a (001) plane in
78: C$_{60}\cdot$2CHCl$_3$ and C$_{60}\cdot$2CHBr$_3$.
79:
80: Given that only a single lattice plane is involved in the superconductivity
81: and given that the transition temperature $T_c$ increases with the distance
82: between neighboring molecules, it should be possible to increase $T_c$ simply
83: by pushing on the gate electrode of the field-effect device. Such uniaxial
84: stress will decrease the spacing of the lattice planes parallel to the gate,
85: but at the same time it will {\em increase} the distance of the molecules
86: {\it in} the planes. This is the Poisson effect \cite{nye}.
87: We again stress the fundamental difference between the alkali- and the
88: field-doped Fullerenes. The alkali-doped materials are bulk superconductors.
89: Thus under pressure, even under uniaxial stress, the volume per molecule is
90: reduced, lowering the density of states at the Fermi level and hence $T_c$.
91: In contrast, in the field-doped Fullerenes only the layer under the gate
92: electrode is doped, and therefore hopping to the deeper layers is suppressed
93: by the strong electrostatic potential of the induced space charge \cite{wehrli}.
94: Thus, reducing the distance between the layers will have a negligible effect
95: on the (practically two-dimensional) density of states, which is, however,
96: strongly increased by the increasing distance of the molecules in the plane.
97:
98: At this point the obvious questions are: Will uniaxial stress really
99: affect the molecules under the gate electrode? And if so, how large will
100: the effect be?
101: The first problem that comes to mind is that the molecules under the gate
102: might be held in place through interactions with the Al$_2$O$_3$ gate oxide.
103: In this case the response of the molecules to stress would rather be determined
104: by the hard oxide than by the elastic constants of the soft organic crystal.
105: Strong bonding of the C$_{60}$ molecules to the gate dielectric seems, however,
106: quite unlikely, and there is even direct evidence for the mobility of the
107: molecules that carry the source-drain current: Upon cooling, solid C$_{60}$
108: undergoes an orientational phase transition at about 260 K, which is
109: accompanied by a substantial reduction in volume (the distance between the
110: molecules decreases by more than 0.3\%, see Fig.~\ref{thermplot}) \cite{david},
111: while the thermal expansion of aluminum oxide does not show any significant
112: structure in that temperature range \cite{thermexp}. Nevertheless, the ordering
113: transition shows up in the resistivity of field-doped C$_{60}$ (Fig.~5 of
114: Ref.~\cite{holedoped}), indicating that the molecules in the relevant
115: layer indeed remain mobile.
116: \begin{figure}
117: \resizebox{3in}{!}{\includegraphics{thermal.epsi}}
118: \caption[]{\label{thermplot}
119: Linear thermal expansion of a C$_{60}$ single crystal \cite{david} and
120: of Al$_2$O$_3$ \cite{thermexp}.}
121: \end{figure}
122:
123:
124: The next question is whether it is experimentally feasible to apply pressure
125: of any significant magnitude to field-effect devices based on delicate organic
126: crystals. As it turns out, the necessary techniques are well
127: established for the investigation of quasi-two-dimensional organic conductors:
128: The whole device is embedded in a suitable epoxy or frozen in oil, and the
129: application of pressures of up to 1 GPa at temperatures down to 0.5 K to
130: samples of millimeter dimensions is straightforward \cite{pressure1,pressure2}.
131: Ironically, one of the major problems in such experiments seems to be the
132: suppression of the Poisson effect \cite{pressure2}, the very effect which is
133: desired in the present case.
134: Thinking about C$_{60}$ under pressure one might worry about polymerization.
135: Pressures of the order of 1 GPa can, however, be considered low
136: \cite{sundqvist}. Moreover, under uniaxial stress the crystal will rather
137: loose cohesion and be squeezed, than that the resilient C$_{60}$ molecules
138: will form chemical bonds.
139:
140: How large an effect on $T_c$ can we expect? For an estimate we calculate
141: the Young modulus $E$ and the Poisson ratio $\sigma$ for C$_{60}$ under
142: uniaxial stress.
143: In terms of the elastic constants, the elastic response to stress in the
144: [100] direction is given by
145: \begin{displaymath}
146: E_{[100]}={(c_{11}-c_{12})(c_{11}+2c_{12})\over c_{11}+c_{12}}
147: \;,\quad
148: \sigma_{[100]}={c_{12}\over c_{11}+c_{12}} .
149: \end{displaymath}
150: For stress in the [111] direction we find
151: \begin{displaymath}
152: E_{[111]}={3(c_{11}+2c_{12})c_{44}\over c_{11}+2c_{12}+c_{44}}
153: \;,\quad\!
154: \sigma_{[111]}={c_{11}+2(c_{12}-c_{44})\over 2(c_{11}+2c_{12}+c_{44})} .
155: \end{displaymath}
156: %uniaxial stress along [111]: crystal becomes hexagonal.
157:
158: \begin{table}
159: \begin{center}
160: \begin{tabular}{ccc|c|cc|cc|l}
161: $c_{11}$&$c_{12}$&$c_{44}$ & $B$ & $E_{[100]}$&$\sigma_{[100]}$ &
162: $E_{[111]}$&$\sigma_{[111]}$\\\hline
163: 14.1& 6.0& 7.7 & 8.7 & 10.5&0.30 & 17.8&0.16 & \cite{yildharris} pot.~II\\
164: 14.9& 6.9& 8.1 & 9.6 & 10.5&0.32 & 19.0&0.17 & \cite{yildharris} pot.~I\\
165: 16.1& 8.2& 8.2 & 10.8 & 10.5&0.34 & 19.6&0.20 & PR potential\\
166: 24.4&12.4&12.4 & 16.4 & 16.1&0.34 & 29.7&0.20 & GF potential\\
167: 24.5&12.3&11.8 & 16.4 & 16.3&0.33 & 28.5&0.21 & \cite{burgos}
168: \end{tabular}
169: \end{center}
170: \vspace{-2ex}
171: \caption[]{\label{elastconst}
172: Elastic constants for a C$_{60}$ crystal at $T=0$.
173: Elastic constants, bulk modulus ($B=(c_{11}+2c_{12})/3$), and
174: Young moduli $E$ are in GPa; Poisson ratios $\sigma$ are dimensionless.}
175: \end{table}
176:
177: Table \ref{elastconst} gives a list of representative theoretical values
178: for the elastic constants of C$_{60}$ from the literature and from calculations
179: using the Girifalco \cite{girifalco} and the Pacheco-Ramalho \cite{pacheco}
180: potentials. There is a remarkably large spread in the predicted elastic
181: constants, which vary by up to a factor of two. The Young moduli fall in the
182: ranges $E_{[100]}\approx10\ldots16$ GPa and $E_{[111]}\approx18\ldots30$ GPa.
183: The Poisson ratios differ, however, by only about ten percent:
184: $\sigma_{[100]}\approx0.32$ and $\sigma_{[111]}\approx 0.2$.
185: Using the slope $c\approx$ 230 K/\AA\ of $T_c$ as a function of $a=\sqrt{2}\,d$,
186: where $d$ is the distance to the nearest molecule in the doped layer, that
187: was found experimentally \cite{latticeexp}, we can estimate the increase in
188: transition temperature under uniaxial stress: $\Delta T_c/p=ca_0\sigma/E$.
189: For a hole-doped C$_{60}$ crystal under uniaxial stress in the [111] direction
190: we thus find $\Delta T_c^{[111]}/p\approx 22\ldots 36$ K/GPa. Should the [100]
191: plane be the relevant for superconductivity, the effect would be even larger:
192: $\Delta T_c^{[100]}/p\approx 65\ldots105$ K/GPa.
193:
194: The effect of uniaxial stress on the transition temperature in field-doped
195: Fullerenes should thus be large, indeed. Already a modest force of
196: 20 Newton on a crystal with an area of a square millimeter should increase
197: $T_c$ by about half a Kelvin or more. Such an effect should be observable
198: when monitoring the source-drain resistivity of the field-effect device
199: just above the ambient-pressure transition temperature.
200: %
201: The increase in $T_c$ is of course limited by the yield stress of the
202: C$_{60}$ crystal. To get a feeling for the behavior of C$_{60}$ under
203: finite stress, Fig.~\ref{fig111} shows the increase in the intermolecular
204: distance $d$ in the plane and the decrease in distance between the lattice
205: planes as a function of the applied stress in [111] direction. For finite
206: stresses the increase in $d$ is even larger than expected from the elastic
207: constants. Eventually the slope of $\Delta d/d_0$ becomes, however, infinite,
208: which means that the crystal is squeezed.
209:
210: \begin{figure}
211: \centerline{\resizebox{3in}{!}{\includegraphics{fig111.eps}}}
212: \caption[]{\label{fig111}
213: Response to uniaxial stress in [111] direction
214: calculated using the Girifalco (open symbols and dotted lines) and
215: the Pacheco-Ramalho (filled symbols and dashed lines) potential.
216: The lines show the deformation expected from the Young modulus,
217: the symbols show the calculated deformations under finite stress.}
218: \end{figure}
219:
220: In our estimate of the change in transition temperature under uniaxial stress
221: we have assumed that the interpretation given in Ref.\ \onlinecite{latticeexp}
222: is correct, namely that the distance of the molecules in the doped lattice
223: plane is the only relevant parameter for determining $T_c$. A recent structural
224: analysis of the low temperature phases of C$_{60}\cdot$2CHCl$_3$ and
225: C$_{60}\cdot$2CHBr$_3$ has, however, cast doubts on this interpretation
226: \cite{nodos}. It might therefore be possible, that other effects, like
227: the different molecular orientations or the presence of the intercalants
228: substantially influence superconductivity. In that respect, investigating
229: the change in $T_c$ under uniaxial stress could help to clarify the situation.
230:
231:
232: In summary, the application of uniaxial stress is a straightforward and
233: feasible way for enhancing the transition temperatures of field-doped
234: Fullerenes even further. We have shown that the enhancement is significant,
235: about 30 K per GPa, so that already the effect of very small pressures should
236: be measurable. Besides the potential for achieving record $T_c$'s, such
237: experiments would deepen our understanding of the physics behind the increase
238: in transition temperature, since they would allow to study the effect of an
239: increased spacing between the molecules without having to introduce additional
240: molecules.
241: %
242: Due to the softness of the Fullerene crystals it is tempting to speculate that
243: it might even be possible to push the field-doped Fullerenes across the
244: Mott transition \cite{c60mott}. Then, by continuously varying the distance
245: between the molecules (by the applied stress) and the doping (by the gate
246: voltage) it should be possible to study the physics of a doped Mott insulator.
247: From the alkali-doped Fullerenes we know that $T_c$ decreases when the
248: lattice constant is increased too much \cite{screen,TcMott}. It will thus be
249: interesting to see whether $T_c$ in the field-doped Fullerenes will be limited
250: by the Mott transition, or if the maximum $T_c$ is realized in the Mott
251: insulating regime --- at some optimal doping.
252:
253: It is a privilege to thank O.~Gunnarsson for sharing his insights and
254: for his continuous support.
255: We would also like to thank R.K.~Kremer, W.~Branz, J.~Merino, K.~Syassen,
256: and J.H.~Sch\"on for helpful discussions.
257:
258: \begin{thebibliography}{}
259: \bibitem{elecdoped}
260: J.H.~Sch\"on, Ch.~Kloc, R.C.~Haddon, B.~Batlogg,
261: Science {\bf 288}, 656 (2000).
262: \bibitem{holedoped}
263: J.H.~Sch\"on, Ch.~Kloc, B.~Batlogg, Nature {\bf 408}, 549 (2000).
264: \bibitem{latticeexp}
265: J.H.~Sch\"on, Ch.~Kloc, B.~Batlogg, Science {\bf 293}, 2432 (2001).
266: \bibitem{dagotto}
267: E.~Dagotto, Science {\bf 293} 2410 (2001).
268: \bibitem{gunnar}
269: O.~Gunnarsson, Rev.\ Mod.\ Phys.\ {\bf 69}, 575 (1997).
270: \bibitem{wehrli}
271: S.~Wehrli, D.~Poilblanc, and T.M.~Rice, cond-mat/ 0106433
272: (European Journal of Physics B {\bf 23}, 345 (2002)).
273: \bibitem{david}
274: W.I.F.~David, R.M.~Ibberson, T.J.S.~Dennis, J.P.~Hare, K.~Prassides,
275: Europhys.\ Lett.\ {\bf 18}, 219 (1992).
276: \bibitem{jansen}
277: M.~Jansen, G.~Waidmann, Z.\ anorg.\ allg.\ Chem.\ {\bf 621}, 14 (1995).
278: \bibitem{nodos}
279: R.E.~Dinnebier, O.~Gunnarsson, H.~Brumm, E.~Koch, P.W.~Stephens, A.~Huq,
280: and M.~Jansen, submitted to Science.
281: \bibitem{nye}
282: J.F.~Nye, {\it Physical Properties of Crystals}
283: (Oxford University Press, 1957).
284: \bibitem{thermexp}
285: Y.S.~Touloukian, R.K.~Kirby, R.E.~Taylor, T.Y.R.~Lee,
286: {\it Thermal Properties of Matter, Vol.~13:
287: Thermal Expansion, Nonmetallic Solids}
288: (Plenum, New York, 1977).
289: \bibitem{pressure1}
290: C.E.~Campos {\it et al.},
291: %Campos, C.E., Brooks, J.S., van Bentum, P.J.M., Perenboom, J.A.A.J., Rook, J.,
292: %Klepper, S.J. \& Tokumoto, M.
293: Rev.\ Sci.\ Instrum.\ {\bf 66}, 1061 (1995).
294: \bibitem{pressure2}
295: M.~Maesato, Y.~Kaga, R.~Kondo, S.~Kagoshima,
296: Rev.\ Sci.\ Instrum.\ {\bf 71}, 176 (2000).
297: \bibitem{sundqvist}
298: B.~Sundqvist, Adv.\ Phys.\ {\bf 48}, 1 (1999).
299: \bibitem{yildharris}
300: T.~Yildirim and A.B.~Harris, Phys.\ Rev.\ B {\bf 46}, 7878 (1992).
301: \bibitem{burgos}
302: E.~Burgos, E.~Halac, and H.~Bonadeo, Phys.\ Rev.\ B {\bf 49}, 15544 (1994).
303: \bibitem{girifalco}
304: L.A.~Girifalco, J.\ Phys.\ Chem.\ {\bf 92}, 858 (1992).
305: \bibitem{pacheco}
306: J.M.~Pacheco, J.P.~Prates Ramalho, Phys.\ Rev.\ Lett.\ {\bf 79}, 3873 (1997).
307:
308: \bibitem{c60mott}
309: O.~Gunnarsson, E.~Koch, R.M.~Martin,
310: Phys.\ Rev.\ B {\bf 54}, R11026 (1996).
311: \bibitem{screen}
312: E.~Koch, O.~Gunnarsson, R.M.~Martin,
313: Phys.\ Rev.\ Lett.\ {\bf 83}, 620 (1999).
314: \bibitem{TcMott}
315: P.~Dahlke, M.S.~Denning, P.F.~Henry, M.J.~Rosseinsky,
316: J.\ Am.\ Chem.\ Soc.\ {\bf 122}, 12352 (2000).
317: \end{thebibliography}
318: \end{document}
319: