1:
2:
3: \documentclass[10pt]{iopart}
4: \usepackage{epsfig}
5: \begin{document}
6:
7:
8: \title[Longitudinal photocurrent spectroscopy of a single v-groove quantum wire]{Longitudinal photocurrent spectroscopy of a single GaAs/AlGaAs v-groove quantum wire}
9:
10: \author{N I Cade\dag\footnote[2]{Present address: NTT Basic Research Laboratories, 3-1 Morinosato Wakamiya,
11: Kanagawa 243-0198, Japan. Email: ncade@will.brl.ntt.co.jp}, M Hadjipanayi\dag, R Roshan\dag, A C Maciel\dag, J F Ryan\dag, F Macherey\S, Th
12: Sch\"{a}pers\S and H L\"{u}th\S}
13:
14:
15: \address{\dag\ Clarendon Laboratory, University of Oxford, Oxford, OX1 3PU, UK}
16:
17: \address{\S\ Institut f\"{u}r Schichten und Grenzfl\"{a}chen, Forschungszentrum, 52425 J\"{u}lich, Germany}
18:
19: \begin{abstract}
20: Modulation-doped GaAs v-groove quantum wires (QWRs) have been fabricated with novel electrical contacts made to two-dimensional
21: electron-gas (2DEG) reservoirs. Here, we present longitudinal photocurrent (photoconductivity / PC) spectroscopy measurements of a single
22: QWR. We clearly observe conductance in the ground-state one-dimensional subbands; in addition, a highly temperature-dependent response is
23: seen from other structures within the v-groove. The latter phenomenon is attributed to the effects of structural topography and
24: localization on carrier relaxation. The results of power-dependent PC measurements suggest that the QWR behaves as a series of weakly
25: interacting localized states, at low temperatures.
26: \end{abstract}
27:
28: \pacs{73.50.Pz, 73.63.Nm, 78.67.Lt}
29:
30: \maketitle %\clearpage
31:
32: \section{Introduction}
33: \label{PCintro}
34:
35: In recent years there has been considerable effort expended in producing high quality quasi-one-dimensional (1D) electronic systems, in
36: order to study their fundamental physical properties. Various techniques have been developed in order to realize these systems in
37: semiconductors; these include the selective depletion of a high mobility two-dimensional electron-gas (2DEG) \cite{thornton86}, cleaved
38: edge overgrowth of quantum wells (QWs) \cite{pfeiffer90}, and vicinal growth on patterned substrates \cite{colas}. There is currently
39: increasing interest in the electronic transport properties of alternative quasi-1D materials, such as carbon nanotubes \cite{tans} and DNA
40: molecules \cite{fink}.
41:
42: The optical properties of semiconductor v-groove quantum wire (QWR) structures have attracted extensive research, due to their potential
43: use in optoelectronic devices including quantum wire lasers \cite{sirigu00}; however, much less attention has been devoted to their
44: electronic properties. The possibility of constructing nanostructure device networks using low-dimensional interconnects necessitates a
45: greater understanding of carrier transport into and through such wires \cite{ferry}. Photocurrent (PC) spectroscopy provides a means of
46: directly investigating the excited state properties of a semiconductor structure. This is especially pertinent when injecting carriers from
47: nanoscale electrical contacts into low-dimensional structures. For the case of QWRs, PC spectroscopy provides information on the specific
48: real-space electronic transitions involved in transport, which cannot be obtained by standard conductance measurements.
49:
50: Previous work in this area has mainly concentrated on \textit{transverse} (i.e.\ perpendicular to the wires) photocurrent measurements of
51: v-groove \mbox{p-i-n} structures \cite{Hamoudi,fu}, involving large ensembles of QWRs \cite{cingolani}. These investigations provided
52: information on the energy levels within QWR structures but have given only limited insight into their overall electrical transport
53: properties. Saraydarov \emph{et al}.\ \cite{saraydarov} have recently reported longitudinal PC measurements of 24 wires that suffered from
54: significant parallel transport through the substrate and buffer layers.
55:
56: In this paper we report temperature- and power-dependent \emph{longitudinal} photocurrent spectroscopy measurements of a \textit{single}
57: QWR. We have grown modulation-doped GaAs QWRs in v-grooves with quasi- \{111\}A sidewalls. In order to electrically contact the wires,
58: wide grooves with an aperture of $\sim$50 $\mu $m have been fabricated at each end of the v-groove (see figure \ref{tem}(b)): as the wire
59: grows in the v-groove, a contiguous 2DEG is formed in the (100) bottom quantum well (BQW) which grows on the bottom facet of the wide
60: groove. This technique creates a tapering from 2D - 1D, and effectively eliminates the high contact resistance that occurs in sidewall
61: contact techniques due to a large wavefunction mismatch between the sidewall quantum well (SQW) and QWR \cite{lelarge}.
62:
63: We have observed conductance in the ground-state one-dimensional subbands, and we have identified additional conductance features that
64: originate from other structures within the v-groove. Small changes in lattice temperature reveal a strong dependence of excited state
65: dynamics and relaxation on structural topography, and highlight the importance of carrier localization. We find that the power-dependence
66: of the PC response can be closely fit by a rate equation model for a two-level quantum dot system. This indicates that, at low
67: temperatures, the QWR behaves as a series of quasi-0D boxes rather than a quasi-1D system.
68:
69:
70:
71:
72: \section{Experimental details}\label{Experimental}
73:
74: A semi-insulating (100) GaAs substrate was patterned with an array of $\sim$25 $\mu$m long [01$\overline{1}$]-oriented v-grooves with 5
75: $\mu$m apertures; each groove was separated by 50 $\mu $m of Si0$_{2}$ mask. A nominal 2.5 nm GaAs quantum well (QW) was grown within
76: Al$_{0.3}$Ga$_{0.7}$As barriers by low-pressure metal-organic vapor-phase epitaxy. A 10 nm thickness of the barrier was modulation-doped 30
77: nm above (below) the QW with a Si concentration of 8 (4) $\times$ 10$^{17}$ cm$%
78: ^{-3}$. Polycrystalline growth occurs between the grooves due to the residual Si0$_{2}$, which prevents the formation of a planar (100) QW
79: at the top of the SQWs and ensures that each groove is electrically isolated \cite{schapers}. Electrical contacts were made onto the 2DEG
80: regions that terminate each groove (figure \ref{tem}(b)). Low temperature current-voltage (I-V) measurements indicate that the v-groove
81: structures are ohmic at low fields, and give a total wire resistance of $\sim$5 k$\Omega$. A detailed description of the growth and
82: characterization of these quantum wires is given elsewhere \cite{schapers}.
83:
84:
85: \begin{figure}[t!]
86: \begin{center}
87: \epsfig{file=bfig1.eps,width=8.5cm}\end{center} \caption{(a) Schematic of a single groove (approximately to scale) showing the different
88: regions of interest : 1) v-groove ($\sim$25 $\mu$m length), 2) 1D-2D transition ($\sim$5 $\mu$m length), 3) end contact 2DEG. (b) SEM image
89: of one end of a v-groove showing the transition into the u-groove. (c) TEM cross-section of a v-groove showing the formation of a quantum
90: wire, side QWs and a Ga rich vertical QW.} \label{tem}
91: \end{figure}
92:
93:
94: Figure \ref{tem}(a) shows a plan-view schematic of a single groove and the three different regions of interest: 1) the v-groove, 2) a 1D-2D
95: transition region, 3) a wide u-groove containing a 2DEG onto which electrical contacts are made. A cross-section of a v-groove is shown in
96: figure \ref{tem}(c), obtained by transmission electron microscopy (TEM); a crescent-shaped GaAs QWR forms at the vertex of two SQWs,
97: separated from them by a narrow constriction or pinch-off. A vertical quantum well (VQW) also forms at the center of the groove,
98: perpendicular to the substrate, due to the higher mobility of Ga atoms in the AlGaAs \cite{biasiol96}.
99:
100: Photocurrent measurements were made using an Ar$^{+}$ pumped automatically-tuneable dye laser ($\sim $1.6 -- 1.8 eV), focussed to a spot
101: size of $\sim$35 $\mu$m on the v-groove surface. The sample was mounted on a variable temperature continuous flow cold-finger, with a base
102: temperature of 4.5 K. A DC voltage generator was used to apply a bias of a few hundred mV between the end contacts; by mechanically
103: chopping the laser-beam, the resultant photocurrent was obtained with standard lock-in techniques.
104:
105: \section{Results and discussion}\label{Results}
106:
107: \begin{figure}[t!]
108: \begin{center}
109: \epsfig{file=bfig2.eps,width=8.5cm}\end{center} \caption{PC spectrum (10 K) obtained with a maximum power of 150 $\protect\mu $W and a 80
110: mV bias. The shading indicates the region of reduced PC observed at low temperature. The dotted lines correspond to the positions of
111: luminescence features. The tuning curve for a typical scan of the dye laser is also shown.} \label{pc}
112: \end{figure}
113:
114:
115: In figure \ref{pc} we present a PC spectrum taken at 10 K. The dotted vertical lines indicate the positions of photoluminescence (PL)
116: features observed for the various groove structures \cite{cade2}. The lines at 1.595 eV and 1.65 eV originate from e$_{1}$-hh$_{1}$ and
117: e$_{2}$-hh$_{2}$ (\textit{n}=1,2) recombination in the QWR respectively. These assignments have been confirmed by spatial mapping of the
118: luminescence from the v-groove, using near-field scanning optical microscopy \cite{marc,cade}, and also by theoretical modelling and
119: PL-excitation (PLE) measurements \cite{cade2}. A strong PC response can be seen at the $n=1$ QWR energy, indicating efficient exciton
120: generation and subsequent carrier transport in the ground state subbands. This is anomalous as the structure is supposedly heavily n-doped
121: and hence this transition should be forbidden; however, there is also no obvious absorption edge seen at the Fermi energy in PLE
122: measurements taken for this sample. The issue of modulation doping in these v-groove structures is complicated and subject to ongoing
123: research. It is possible that the transfer of free carriers into the QWR is inefficient under certain growth or experimental conditions.
124:
125:
126: There is a prominent drop in the overall PC response between the low energy side of the \textit{n}=2 QWR transition and the VQW energy, as
127: indicated by the shaded region in figure \ref{pc}. This is evidence of reduced extraction efficiency of electron-hole pairs through the
128: device at low temperature, as discussed below. Absorption is also clearly evident at the SQW energy, which is consistent with parallel
129: conduction mechanisms; at low temperatures carrier relaxation from the sidewalls into the QWR is greatly restricted by the pinch-off
130: regions \cite{cade2}. Despite the large area presented by the SQWs for conduction, the PC response at this energy is relatively small. This
131: is due to discontinuity of the \{111\}A facet at the ends of the groove, which prevents efficient carrier transfer between the end BQWs and
132: the SQWs (see figure \ref{tem}(b)). Another strong peak occurs at the VQW energy; this is indicative of carrier transfer between the VQW
133: and QWR \cite{kiener}, as parallel conduction is unlikely due to strong compositional disorder in the VQW.
134:
135:
136: \subsection{Temperature characteristics}\label{pctempsec}
137:
138: PC spectra were recorded at different temperatures and are presented in figure \ref{pctemp}. With a small rise in
139: temperature the absorption features disappear, and the PC increases greatly in the middle energy range. This is
140: especially noticeable at the \textit{n}=2 QWR energy.
141:
142: \begin{figure}[t!]
143: \begin{center}
144: \epsfig{file=bfig3.eps,width=8.5cm}\end{center} \caption{PC spectra for a single groove at different temperatures (200 mV, 30 $\mu$m spot,
145: maximum power 4 $\mu$W). The dotted lines correspond to the features in figure \ref{pc}.} \label{pctemp}
146: \end{figure}
147:
148: In the QWR, the e$_{2}$ electronic wavefunction is localized over the \{311\}A facets which create the pinch-off \cite{wang1}. These facets
149: are subject to large interfacial disorder, which causes strong exciton localization \cite{bellessa}, and greatly inhibited relaxation
150: \cite{oberli}. These issues are discussed in detail elsewhere \cite{cade2}. At very low temperatures, localization effects play a dominant
151: role in limiting carrier transport; this results in a reduction of the photocurrent contribution from certain parts of the structure, as
152: seen in figure \ref{pc} between 1.63 eV and 1.75 eV. The radiative recombination rate for an exciton is found to be proportional to its
153: coherence volume in the confining potential \cite{feldmann}. Excitons localized over lengths $L$, comparable to the excitonic Bohr radius
154: ($a_{\rm B}$) have a relatively long lifetime of up to 500 ps at low temperatures \cite{bellessa}, and are thus susceptible to dissociation
155: by hot carriers. In the limits of $L$$\gg$$a_{\rm B}$ and $L$$\ll$$a_{\rm B}$, the radiative lifetime tends to that of a free exciton,
156: $\sim$150 ps \cite{citrin}. The \textit{n}=1 and \textit{n}=2 QWR states are expected to have different localization lengths as their
157: respective wavefunctions are centered on crystal facets with different interfacial roughness \cite{wang1}. Thus, at low temperature
158: excitons in the \textit{n}=2 QWR state may recombine before they are dissociated, and are not able to contribute to the photocurrent. As
159: the lattice temperature is increased, these excitons are de-trapped and can relax more efficiently to the QWR ground-state where they are
160: ionized within the recombination time. This results in an increase in the PC at the \textit{n}=2 QWR energy, as seen in figure
161: \ref{pctemp}. The PC at the SQW energy also increases with temperature due to a real-space transfer of carriers into the QWR
162: \cite{cade2,cade}, which results in increased carrier-carrier scattering and smearing of the absorption features.
163:
164: In this discussion we have assumed that radiative recombination is the dominant mechanism, as non-radiative processes are expected to
165: become significant only at temperatures greater than the range considered here \cite{akiyama}. We have performed time-resolved measurements
166: on similar QWR structures and obtain low temperature lifetimes of $\sim $350 ps and $\sim $150 ps for the \textit{n}=1 and \textit{n}=2
167: states respectively \cite{cade2}, which supports the above conclusions. Above 30 K, we observe a decrease in the relaxation time of the
168: \textit{n}=2 state to the ground state, which is also consistent with these measurements.
169:
170: \subsection{Power characteristics}\label{pcpowersec}
171:
172: In figure \ref{pcpower} we show the PC response as a function of laser power, when illuminating at various
173: energies. In all cases the PC increases rapidly with a small increase in power, and then saturates at a value that
174: depends on the bias.
175:
176: \begin{figure}[t!]
177: \begin{center}
178: \epsfig{file=bfig4.eps,width=8.5cm}\end{center} \caption{Photocurrent as a function of laser power, for different laser energies at 10 K;
179: (a) 400 mV bias across the groove, (b) 200 mV bias, for higher laser powers. The data have been fitted with a function based on a rate
180: equation model described in the text.} \label{pcpower}
181: \end{figure}
182:
183: The high sensitivity of the PC to low powers explains why the spectrum in figure \ref{pc} shows a clear peak at $\sim$ 1.6 eV
184: while those in figure \ref{pctemp} do not. The laser is at the extreme of its tuning range at this energy, and the excitation power (and
185: hence PC response) can vary significantly on different days depending on the fine tuning of the laser.
186:
187: Beham \textit{et al}.\ \cite{beham02} have observed a similar power-dependent PC response in single InGaAs quantum dots (QDs), which they
188: attributed to bleaching of the dot ground-state. We outline their model below and discuss its relevance to a 1D wire system.
189:
190: The saturation of the PC in a QD with increasing excitation power can be described by a fundamental rate equation model for a two-level
191: system. This assumes the dot is in level 1 if it is unoccupied, and in level 2 if it is occupied with one exciton. Stimulated processes are
192: assumed to occur from level 2 with a probability $M$, and radiative recombination will occur in a time $\tau_{\rm r}$. Excitons can also
193: escape out of the dot by field-dependent carrier tunnelling, with a escape time $\tau_{\rm esc}(F)$. This leads to occupation-dependent
194: absorption behavior in the QD, which produces a nonlinear power dependence of the PC resulting in saturation. At higher electric fields the
195: tunnelling time is reduced and the QD returns faster to its initial empty state; therefore, a higher absorption rate is possible leading to
196: a higher saturated PC. With an excitation power $P$, the steady-state photocurrent is found to be
197: \begin{equation}\label{PCeq}
198: I_{PC}=\left(\frac{e}{2\tau_{\rm esc}}\right)\frac{P}{P+(1/2M)(1/\tau_{\rm esc}+1/\tau_{\rm r})}.
199: \end{equation}
200:
201: This expression has been used to fit the data points in figure \ref{pcpower}; we assume $\tau_{\rm r}\equiv\tau_{\rm loc}\simeq350$ ps
202: \cite{cade2}. From the data in figure \ref{pcpower}(a), with resonant excitation of the QWR at 1.585 eV, we obtain values of $M=0.36$ and
203: an escape time $\tau_{\rm esc}=1.9$ ps for 400 mV bias (80 V\,cm$^{-1}$). At 200 mV bias, the data fits in figure \ref{pcpower}(b) give
204: $M=0.24$ (0.20) and $\tau_{\rm esc}=3.8$ (3.9) ps for illumination at 1.6 eV (1.7 eV).
205:
206: The closeness of the fits, at least over low powers, suggests that the model developed for QDs is still valid in our 1D QWR system. As
207: discussed above, this is likely to be a result of localization effects at low temperatures; monolayer growth fluctuation will generate a
208: series of quasi-0D boxes along the wire axis \cite{hasen97}, thus emulating the isolated QD model. The escape time at 400 mV bias is half
209: the value at 200 mV, which suggests that in our case $\tau_{\rm esc}^{-1}$ gives an indication of the rate of dissociation of localized
210: excitons by field-accelerated carriers. These values of $\tau_{\rm esc}$ are much smaller than those obtained by Beham \textit{et al}.\
211: even though our fields are 1000 times less; this is likely to be due to the relatively high density of '0D' localization sites along the
212: wire ($\sim5 \times 10^{4}$ cm$^{-1}$ \cite{cade2}) and the presence of a 1D delocalized exciton continuum.
213:
214: Figure \ref{pcpower}(b) shows that the saturation PC is significantly smaller when exciting at 1.70 eV rather than quasi-resonantly with
215: the $n=1$ state at 1.6 eV. This is consistent with the discussion above; at low temperatures, excitons generated in excited states of the
216: wire recombine more rapidly than ground state excitons, thus generating fewer dissociated free carriers.
217:
218:
219: There is not sufficient data to comment on the field- and excitation energy-dependence of $M$, and a detailed analysis of the true meaning
220: of $M$ and $\tau_{\rm esc}$ in a 1D system will require a more comprehensive investigation. The deviation of the PC from a saturation
221: value, observed at higher powers in figure \ref{pcpower}(b), may be due to the presence of the 1D continuum. Scattering with phonons and
222: other carriers may be significant at higher powers and has also been neglected in the model.
223:
224: \section{Conclusion}\label{Conc}
225: In summary, we have presented longitudinal photocurrent measurements of a single v-groove quantum wire. The
226: photocurrent response from certain structures within the v-groove is found to be highly temperature dependent. In
227: particular, a marked drop is observed in the contribution from the first excited state of the quantum wire, below
228: 30 K. This has been attributed to frustrated carrier relaxation resulting from localization effects.
229:
230: The total photocurrent through the v-groove has been measured as a function of incident laser power; the PC
231: response can be closely fit by a rate equation model for a two-level quantum dot system. This is consistent with
232: the scenario that the QWR behaves as a series of quasi-0D boxes, at low temperatures.
233:
234: This work was supported by the EPSRC (UK), and the EC through the ULTRAFAST network.
235:
236: \section*{References}
237: \begin{thebibliography}{99}
238:
239: \bibitem{thornton86} Thornton T J, Pepper M, Ahmed H, Andrews D and Davies G J 1986 \textit{Phys. Rev. Lett.} {\bf 56} 1198
240:
241: \bibitem{pfeiffer90} Pfeiffer L, West K W, St{\"o}rmer H L, Eisenstein J P, Baldwin K W, Gershoni D and Spector J 1990 \textit{Appl. Phys. Lett.} {\bf 56} 1697
242:
243: \bibitem{colas} Colas E, Simhony S, Kapon E, Bhat R, Hwang D M and Lin P S D 1990 \emph{Appl. Phys. Lett.} \textbf{57} 914
244:
245: \bibitem{tans} Tans S J, Devoret M H, Dai H, Thess A, Smalley R E, Geerligs L J and Dekker C \textit{Nature} 1997 \textbf{386} 474
246:
247: \bibitem{fink} Fink H-W and Sch{\"o}nenberger C 1999 \textit{Nature} \textbf{398} 407
248:
249: \bibitem{sirigu00} Sirigu L, Oberli D Y, Degiorgi L, Rudra A and Kapon E 2000 \textit{Phys. Rev.} B {\bf 61} 10575
250:
251: \bibitem{ferry} Ferry D K and Goodnick S M 1997 \textit{Transport in Nanostructures} (UK: Cambridge University Press)
252:
253: \bibitem{Hamoudi} Hamoudi A, Ogura M and Wang X L 1997 \textit{J. Appl. Phys.} \textbf{81} 6229
254:
255: \bibitem{fu} Fu Y, Willander M, Liu X Q, Lu W, Shen S C, Tan H H,
256: Jagadish C, Zou J and Cockayne D J H 2001 \textit{J. Appl. Phys.} \textbf{89} 2351
257:
258: \bibitem{cingolani} Cingolani R, Rinaldi R, De Vittorio M, Vasanelli L,
259: Cola A, Marti U and Reinhart F K 1996 \textit{J. Appl. Phys.} \textbf{80} 936
260:
261: \bibitem{saraydarov} Saraydarov M, Donchev V, Germanova K, Wang X L, Kim S J and Ogura M 2004 \textit{J. Appl. Phys.} \textbf{95} 64
262:
263: \bibitem{lelarge} Lelarge F, Kaufman D, Dwir B, Mautino S, Rudra A and Kapon E 2000 \textit{J. Crystal Growth} \textbf{221} 540
264:
265: \bibitem{schapers} Sch\"{a}pers Th, Hartmann A, Schwarz A, Hardtdegen H, Bongartz M, Dieker Ch and L\"{u}th H 1998 \textit{Appl. Surf. Sci.} \textbf{123/24} 687
266:
267: \bibitem{biasiol96} Biasiol G, Reinhardt F, Gustafsson A, Martinet E and Kapon E 1996 \textit{Appl. Phys. Lett.} \textbf{69} 2710
268:
269: \bibitem{cade2} Cade N I, Roshan R, Hauert M, Maciel A C, Ryan J F, Schwarz A,
270: Sch\"{a}pers Th and L\"{u}th H 2004 \textit{Phys. Rev.} B {\bf 70} 195308
271:
272: \bibitem{marc} Hauert M, Roshan R, Maciel A C, Kim J, Ryan J F, Schwarz A, Kaluza A, Sch\"{a}pers Th and L\"{u}th H 2001 \textit{Proc. 25th Int. Conf. on the Physics of Semiconductors (Osaka)} ed N Miura and
273: T Ando (Berlin: Springer) p 703
274:
275: \bibitem{cade} Cade N I, Roshan R, Ryan J F, Maciel A C, Schwarz A, Sch\"{a}pers Th and L\"{u}th H 2002 \textit{Physica} B \textbf{314} 413
276:
277:
278: \bibitem{kiener} Kiener C, Rota L, Maciel A C, Freyland J M and Ryan J F 1996 \textit{Appl. Phys. Lett.} \textbf{68} 2061
279:
280: \bibitem{wang1} Wang X L, Ogura M, Matsuhata H and Hamoudi A 1997 \textit{Appl. Phys.
281: Lett.} \textbf{71} 2130
282:
283: \bibitem{bellessa} Bellessa J, Voliotis V, Grousson R, Wang X L, Ogura M and Matsuhata H 1997 \textit{Appl. Phys. Lett.} \textbf{71} 2481; 1998 \textit{Phys. Rev.} B \textbf{58} 9933
284:
285: \bibitem{oberli} Oberli D Y, Vouilloz F, Ambigapathy R, Deveaud B and Kapon E 2000 \textit{Phys. Status Solidi} A \textbf{178} 211
286:
287: \bibitem{feldmann} Feldmann J, Peter G, G\"{o}bel E O, Dawson P, Moore K, Foxon C and Elliott R J 1987 \textit{Phys. Rev. Lett.} \textbf{59} 2337
288:
289: \bibitem{citrin} Citrin D S 1992 \textit{Phys. Rev. Lett.} \textbf{69} 3393
290:
291: \bibitem{akiyama} Akiyama H, Koshiba S, Someya T, Wada K, Noge H, Nakamura Y, Inoshita T, Shimizu A and Sakaki H 1994 \textit{Phys. Rev. Lett.} \textbf{72} 924
292:
293: \bibitem{beham02} Beham E, Zrenner A, Findeis F, Bichler M and Abstreiter G 2002 \textit{Physica} E \textbf{13} 139
294:
295: \bibitem{hasen97} Hasen J, Pfeiffer L N, Pinezuk A, He S, West K W and Dennis B S 1997 \textit{Nature} {\bf 390} 54
296:
297:
298: \end{thebibliography}
299:
300:
301: \end{document}
302: