1: \documentclass[aps,prl,twocolumn]{revtex4}
2: \usepackage{graphicx}
3: \usepackage{amsmath}
4: \usepackage{amssymb}
5: \usepackage{epsfig}
6: \newcommand{\ket}[1]{\lvert#1\rangle}
7:
8: \begin{document}
9:
10:
11: \title{Transition from Free to Interacting Composite Fermions away from $\nu=1/3$.}
12:
13: \author{Y. \surname{Gallais}$^{1}$}
14: \email{yann@phys.columbia.edu}
15: \author{T.H. \surname{Kirschenmann}$^{1}$}
16: \author{I. \surname{Dujovne}$^{1}$}
17: \altaffiliation{Delft University of Technology, Kavli Insitute of NanoScience, Delft, The Netherlands}
18: \author{C.F. \surname{Hirjibehedin}$^{1}$}
19: \altaffiliation{IBM Ressearch Division, Almaden Research Center, San Jose, CA 95120}
20: \author{A. \surname{Pinczuk}$^{1,2}$}
21: \author{B.S. \surname{Dennis}$^{2}$}
22: \author{L.N. \surname{Pfeiffer}$^{2}$}
23: \author{K.W. \surname{West}$^{2}$}
24: \affiliation{$^1$Departments of Physics and of Applied Physics, Columbia University, New York, NY 10027 \\
25: $^2$Bell Labs, Lucent Technologies, Murray Hill, NJ 07974}
26:
27:
28: \begin{abstract}
29: Spin excitations from a partially populated composite fermion level
30: are studied above and below $\nu=1/3$. In the range $2/7<\nu<2/5$
31: the experiments uncover significant departures from the
32: non-interacting composite fermion picture that demonstrate the
33: increasing impact of interactions as quasiparticle Landau levels are
34: filled. The observed onset of a transition from free to interacting
35: composite fermions could be linked to condensation into the higher
36: order states suggested by transport experiments and numerical
37: evaluations performed in the same filling factor range.
38: \end{abstract}
39:
40: \maketitle
41:
42:
43:
44: %[main text]
45: In the composite fermion (CF) picture of the fractional quantum
46: Hall effect (FQHE) fundamental interactions are taken into
47: account at the mean field level by mapping the system of strongly
48: interacting electrons in a perpendicular magnetic field into a a
49: system of composite fermions (CF) moving in a reduced effective
50: magnetic field \cite{Jain}. The effective magnetic field
51: experienced by CF quasiparticles is $B^*$=$B-B_{\frac{1}{\phi}}$,
52: where $\phi$ is an even integer that labels different sequences of
53: incompressible states and $B_{\frac{1}{\phi}}$ is the magnetic
54: field corresponding to the filling factor $\nu$=$\frac{1}{\phi}$.
55: In the CF framework the reduction in magnetic field follows from the binding of $\phi$
56: flux quanta to electrons. The main sequence of FQHE states
57: are at Landau level filling factor $\nu$=p/($\phi$p$\pm$1), where
58: $p$ is the CF filling factor. Each FQHE sequence is then centered
59: around an even-denominator fraction, $\nu=1/\phi$, where the
60: effective field cancels and a compressible Fermi liquid is thought
61: to form \cite{HLR}. Underlying the CF paradigm is the formation of
62: CF Landau levels with spacing that depends on the effective field
63: $B^*$ as
64: \begin{equation}
65: \omega_{CF}=\frac{e\left|{B^*}\right|}{m^{*}c}
66: \label{CF}
67: \end{equation}
68: where $m^{*}$ is an effective CF mass.
69: \par
70: In the CF framework the complex strongly interacting many-body
71: physics is transformed into the simple picture of non-interacting CF
72: at filling factors $p$. The sequence of lowest spin-split CF Landau
73: levels is shown in Fig. \ref{fig1} for quasiparticles with $\phi$=2
74: (or $^2$CFs) and $\phi$=4 (or $^4$CFs). The issue of the impact of
75: residual CF interactions is of particular relevance to states that
76: have $1<p<2$ in which there is partial population $p^*=p-1$ of an
77: excited Landau level (see Fig. \ref{fig1}). Residual CF interactions
78: in the partially populated excited CF level are believed to be
79: responsible for the formation of higher order FQHE states such as
80: $\nu=4/11$ and $\nu=4/13$ that have $p^*=1/3$ for $^2$CFs and
81: $^4$CFs \cite{Pan}. The experimental evidence for such higher order
82: states bring into question the validity of the weakly interacting CF
83: framework \cite{Quinn}. While these higher order fractions were
84: actually predicted in the earlier hierarchical model
85: \cite{Haldane,Halperin}, it was argued that the CF model may account
86: for these new fractions when CF residual interactions are
87: incorporated beyond the mean-field level
88: \cite{Wojs-Quinn,Jain-Mandal,Mandal-Jain,Goerbig1,Goerbig2,Jain-chang,
89: Lopez-Fradkin,Wojs-Quinn2}.
90: \par
91: Experimental evidence for the formation of a Landau level structure
92: of CFs is found in light scattering measurements of spin-flip (SF)
93: excitations in which Landau level number and orientation of spin
94: change simultaneously \cite{Irene1}. At filling factors away from
95: $\nu=1/3$, the lowest lying SF excitations probe the level structure
96: because the transitions in these excitations emanate from the partially
97: populated CF level as shown in Fig. \ref{fig1} \cite{Murthy,Mandal}.
98: Since SF excitations can be observed at fractional values
99: of $p$, experiments that probe lowest lying SF modes are
100: powerful tools in studies of CF interactions at filling factors away
101: from the major FQHE states.
102: \begin{figure}[b]
103: \centering \epsfig{figure=fig1.eps, width=0.89\linewidth,clip=}
104: \caption{Composite Fermion level structure for $\nu>1/3$ ($^2$CF or
105: $\phi$=2 flux attachment, left) and $\nu<1/3$ ($^4$CF or $\phi$=4
106: flux attachement, right) with the corresponding spin excitations:
107: the spin-flip (SF) and the spin-wave (SW). The dotted lines
108: represent the partially populated levels.} \label{fig1}
109: \end{figure}
110: \par
111: In this Letter we report measurements of low lying SF excitations
112: by inelastic light scattering that reveal marked quasiparticle
113: interactions when there is partial population of the $|1,\uparrow>$
114: CF Landau level ($0<p^*<1$). The experiments indicate a sharp
115: onset of CF interactions that occurs for
116: $p^*\approxeq1/5$. These results can be regarded as evidence that
117: emergence of higher order FQHE states away from filling factor
118: $\nu=1/3$ is linked to a transition from 'free' to interacting
119: behavior at relatively low occupation of the $\ket{1,\uparrow}$ CF
120: level.
121: \par
122: The sensitivity of low lying SF modes to interactions among CF
123: quasiparticles occurs because in a 'single-particle', or 'free',
124: CF paradigm (no impact of CF interactions), the mode linewidths
125: are expected to be independent of $p^*$ and their light scattering
126: intensities should be proportional to $p^*$. We find that at low
127: values of $p^*<0.2=1/5$, the SF mode remains sharp (width
128: $\gamma\leq0.06~meV$) and its light scattering intensity indeed is
129: linear in $p^*$ for both $\nu>1/3$ and $\nu<1/3$. When $p^*>1/5$,
130: however, the light scattering spectra reveal marked departures
131: from the 'free' CF behavior. The SF mode of $^2$CFs displays
132: significant broadening, with $\gamma$ reaching 0.1~meV for
133: $p^*>1/2$, and the light scattering intensities of SF modes of
134: $^2$CFs and $^4$CFs saturate when $p^* \rightarrow 1/2$.
135: \par
136: The broadening of the light scattering peak of SF modes for
137: $p^*>1/5$ suggests a small energy scale for $^2$CFs interactions
138: that is in the range of 0.1~meV. It is conceivable that
139: the broadening is linked to lower lying collective excitations
140: that exist when there is partial population of the
141: $\ket{1,\uparrow}$ $^2$CF level. In this scenario the energy range
142: for such excitations would be comparable to the enhanced
143: broadening of SF excitations. Currently there is no evaluation of
144: SF excitations as a function of $p^*$ but the observed broadening energies
145: are in the range of predictions for the energy scale of residual
146: CF interactions \cite{Goerbig1,Jain-chang}.
147:
148: \par
149: The inelastic light scattering measurements were performed on a
150: high quality GaAs single quantum well (SQW) of width 330~$\AA$.
151: The electron density is n=5.5x10$^{10}$~cm$^{-2}$ and its low
152: temperature mobility is $\mu$=7.2x10$^6$~cm$^2$/V s.The sample was
153: mounted on the cold finger of a $^3$He/$^4$He dilution
154: refrigerator with windows for optical access that is inserted in
155: the cold bore of a superconducting magnet. Cold finger
156: temperatures can reach as low as T=23~mK. The back scattering
157: geometry was used at an angle $\theta$ with the normal of the
158: sample surface as shown in the inset to Fig. \ref{fig2}. The
159: magnetic field perpendicular to the sample is B=B$_T$cos$\theta$
160: and B$_T$ is the total field. The results reported here have been
161: obtained with $\theta$=50$^\circ\pm$2. Resonant inelastic light
162: scattering spectra were obtained by tuning the incident photon
163: energy of a Ti-Sapphire laser close to the fundamental optical gap
164: of GaAs to enhance the light scattering cross-section.The power
165: density was kept below 10$^{-5}$~W/cm$^2$ to prevent heating of
166: the electron gas. The scattered signal was dispersed by a triple
167: grating spectrometer working in additive mode and analyzed by a
168: CCD camera with 26$\mu m$ wide pixels. With a slit width of
169: 30~$\mu m$, the combined resolution of the system is about
170: 0.02~meV.
171: \par
172: Figure \ref{fig2} shows that two low-lying spin excitations are
173: observed. In addition to the SF mentioned above, there is the
174: spin-wave excitation (SW) in which only spin orientation changes
175: ($\delta$S$_z$=$\pm$1)(see Fig. \ref{fig1}). By virtue of the Larmor
176: theorem, at long wavelengths the spin-wave is at the 'bare' Zeeman
177: energy E$_z$=g$\mu_BB_T$, where $\mu_B$ is the Bohr magneton and g
178: is the Land\'{e} factor of GaAs. On the other hand the energies and
179: lineshapes of SF excitations incorporate quasiparticle interactions
180: that offer probes of composite fermion physics.
181: \cite{Murthy,Mandal,Irene1,Irene2,Cyrus-thesis,Gallais}. At $\nu=1/3$, the only
182: low-energy spin excitation observed is the long wavelength SW mode
183: at E$_z$. When the electron fluid becomes compressible, for
184: \textit{both} $\nu>1/3$ and $\nu<1/3$, the SF peak appears on the
185: lower energy side of the SW mode. The narrow linewidth of SF
186: excitations on both sides of $\nu=1/3$ (Full Width at Half Maximum
187: or $\gamma<0.06~meV$) suggests that $\ket{1,\uparrow}$ Landau levels
188: of $^2$CFs and $^4$CFs are sharp.
189: \begin{figure}
190: \centering \epsfig{figure=fig2.eps, width=0.99\linewidth,clip=}
191: \caption{Light scattering spectra of the low-lying spin
192: excitations at three different filling factors: $\nu=0.343$, $\nu=1/3$
193: and $\nu=0.323$. The scattering
194: geometry is shown in inset.} \label{fig2}
195: \end{figure}
196: \par
197: Figure \ref{fig3a} displays the evolution of the spectra of SF and
198: SW modes for $\nu>1/3$ as a function of $p^*$. These spin excitations are
199: linked to ($^2$CF) quasiparticles. It is immediately apparent from
200: these spectra that the SF mode lineshape and energy are strong
201: functions of $p^*$. We constructed reasonably good fits to the
202: spectra with two gaussian functions, as shown in Fig. \ref{fig3a}. A
203: similar procedure was carried out for $^4$CF when $\nu<1/3$ (not
204: shown in Fig. \ref{fig3a}). The linewidth $\gamma$ and the
205: integrated intensity of the SF mode that are obtained by this
206: analysis are shown in Fig. \ref{fig3}.
207:
208: \begin{figure}
209: \centering \epsfig{figure=fig3a.eps,width=0.7\linewidth, clip=}
210: \caption{Evolution of the light scattering spectrum with $p^*$ for
211: $\nu>1/3$. The dotted line shows the analysis with of a
212: two-gaussian fit of the spin-flip (SF) and spin-wave (SW) peaks.}
213: \label{fig3a}
214: \end{figure}
215: \begin{figure}
216: \centering \epsfig{figure=fig3b.eps,width=0.99\linewidth,clip=}
217: \caption{Evolution of the integrated light scattering intensity
218: (open squares) and the spectral linewidth $\gamma$ (full
219: triangles) of the spin-flip (SF) excitation as a function of
220: partial filling factor $p^*$. The left panel shows the data for
221: $\nu<1/3$ or $^4$CF, while the right panel shows data for
222: $\nu>1/3$ or $^2$CF. The dotted lines are guides to the eyes. A CF
223: Landau level scheme is drawn in inset. It depicts a simple picture
224: of the transition from "free" to interacting CF regime.}
225: \label{fig3}
226: \end{figure}
227:
228: The dependence of the integrated SF intensity and $\gamma$ on $p^*$
229: in Fig. \ref{fig3} reveals two distinct regimes. At low $p^*$, the
230: integrated intensity grows linearly with $p^*$ and the linewidth
231: remains constant around 0.06~meV$\pm$0.01. These features are
232: consistent with a picture of "free" CFs that applies to both $^2$CF
233: and $^4$CF sides. For $p^*>0.2$, however, the saturation of the
234: integrated intensity seems to indicate the onset of a breakdown of
235: the "free" CF picture.
236: \par
237: The observed $p^*$-dependence of the SF linewidths is consistent
238: with the scenario of an onset of a regime dominated by CF
239: interactions for $p^*>0.2$. The linewidth of the SF light
240: scattering peak of $^2$CF, shown in in Fig. \ref{fig3}, increases
241: when $p^*>0.2$ to up to 0.1~meV for $p^*=0.55$. The enhanced SF
242: linewidwth could be regarded as a key manifestation of the onset
243: of CF interactions. By contrast the SF peak remains sharp for $^4$CF,
244: likely due to the expected weaker residual interactions among $^4$CFs.
245: \par
246: The picture we propose to interpret these results assumes that
247: states of CF change with interactions in the same way as states of
248: electrons. In a free CF picture, such as the one in Fig.
249: \ref{fig1}, the partially populated $\ket{1,\uparrow}$ level is
250: expected to be infinitely sharp and CF quasiparticles are
251: long-lived. This picture is expected to hold at low $p^*$. As
252: $p^*$ increases CF quasiparticles acquire a finite lifetime due to
253: interactions and the $\ket{1,\uparrow}$ level is expected to
254: broaden significantly as sketched in the inset to Fig.
255: \ref{fig3}. In this regime the width of the SF mode is expected
256: to increase as seen in our data for $p^*>0.2$. The
257: $\ket{1,\uparrow}$ level may eventually split into well-defined
258: higher order CF Landau levels which, in turn, can give birth to
259: higher order FQHE states like the one observed at $\nu=4/11$ by
260: Pan et al. \cite{Pan}.
261: \par
262: An estimate of the CF level broadening can be obtained from the
263: evaluations of the gap $\Delta$ of the incompressible state at
264: $\nu=4/11$. Recent calculations, that incorporate corrections to
265: account for the width of the quantum well, give a value of roughly
266: 0.1~meV \cite{Goerbig1}. This value is consistent with the
267: linewidths observed and supports the interpretation that the width
268: of the SF peak indeed increases due to the increasing impact of the
269: CF interactions for $p^*>0.2$.
270: \par
271: Additional insights can be obtained from the filling factor
272: dependence of the SF mode energy. Up to now, we have implicitly
273: considered that the level ordering is the one depicted in
274: Fig.\ref{fig1} and Fig.\ref{fig3}. This ordering assumes full spin
275: polarization in the full range $1/3<\nu<2/5$. While this assumption
276: has been debated \cite{Wojs-Quinn,Wojs-Quinn2,Mandal-Jain,Jain-chang},
277: transport experiments with
278: tilted magnetic fields indicate that at least the state at
279: $\nu=4/11$ is fully spin polarized for $B_T>10~T$.
280: \par
281: In the following we show that the filling factor dependence of the
282: SF peak energy clarifies this issue. If we neglect excitonic
283: interactions between the excited CF quasiparticle and the
284: quasihole, one can write a simple expression for the SF mode
285: energy \cite{Mandal-Jain,Irene1}
286: \begin{equation}
287: \hbar\omega_{SF}=E_z^*-\hbar\omega_{CF}
288: \label{SF}
289: \end{equation}
290: where E$_z^*$=E$_z$+E$^{\uparrow\downarrow}$ and
291: E$^{\uparrow\downarrow}$ is the spin reversal many-body energy for
292: composite fermions \cite{Mandal-Jain-2}. E$^{\uparrow\downarrow}$
293: depends strongly on the spin polarization of the system and should
294: vanish in the absence of spin polarization. Therefore any loss of
295: spin polarization should drastically decrease the spin reversal
296: many-body energy and drive the SF peak energy towards zero energy
297: \cite{Irene3}.
298: \begin{figure}
299: \centering \epsfig{figure=fig4.eps,width=0.89\linewidth,clip=}
300: \caption{Filling factor dependence of $\omega_{SF}$ (Black full
301: circles and up-triangles for $^2\omega_{SF}$ and $^4\omega_{SF}$
302: respectively) and E$_{z}$ (open squares) for $0.31<\nu<0.38$. The
303: dotted lines are guides to the eyes. The inset shows the splitting
304: between E$_z$ and $\omega_{SF}$ plotted as a function of the reduced
305: effective field, $B^*$, of CF theory (see text).} \label{fig4}
306: \end{figure}
307: \par
308: The filling factor dependence of the SF mode energy is shown in
309: Fig.\ref{fig4} for $0.31<\nu<0.38$. The SF mode energies have
310: monotonic dependence on filling factor for both $\nu>1/3$ and
311: $\nu<1/3$. These behaviors clearly indicate that there is no
312: significant loss of spin polarization in this filling factor range.
313: \par
314: It is significant that the SF mode energies show opposite trends
315: with $\nu$ (or $B$): while $^2\omega_{SF}$ increases with $\nu$,
316: $^4\omega_{SF}$ decreases. The CF framework captures the opposite
317: trends of the SF mode energies with $\nu$. Indeed, according to
318: Eq. \ref{SF} and neglecting the weak $\sqrt{B}$ dependence of the
319: spin-reversal many-body energy, the splitting between the Zeeman
320: energy and $\omega_{SF}$ should scale as
321: $\left|{B^*}\right|$. $B^*$, the reduced effective field
322: experienced by CF quasiparticles, is written as $B$-$B_{1/2}$ and
323: $B$-$B_{1/4}$ for $^2$CF and $^4$CF respectively. The evolution of
324: the splitting as a function of $\left|{B^*}\right|$ is shown in
325: the inset of Fig.\ref{fig4}. These results are remarkably
326: consistent with key predictions of the CF framework in that the
327: mode energy is proportional to $\left|{B^*}\right|$ for both SF
328: excitations.
329: \par
330: A sharp discontinuity in the SF mode energy occurs at $\nu=1/3$.
331: This observation finds a natural explanation within the CF
332: framework because when crossing the $\nu=1/3$ boundary there is an
333: abrupt change in character of the quasiparticles in the ground
334: state of the quantum fluid. The discontinuity of the SF energy at
335: exactly $\nu=1/3$ is thus regarded as a manifestation of the
336: different energy scales associated with $^2$CF and $^4$CF
337: quasiparticles. Indeed, the energies of SF excitations are linked
338: to the CF cyclotron frequency $\omega_{CF}$ and the spin
339: reversal many-body energy E$^{\uparrow\downarrow}$. These energies
340: are expected to be markedly smaller for $^4$CF quasiparticles, as
341: found in the recent light scattering observations of $^4$CF
342: spin-conserving excitations \cite{Cyrus}. We note that a similar discontinuity has been
343: reported recently in luminescence experiments around $\nu=1/3$ \cite{Byszewski}.
344: \par
345: In summary, we have observed CF SF excitations away from $\nu=1/3$
346: and use them as a tool to study CF interactions away from major fractions.
347: The evolution of the lineshape of the SF peak excitation reveals a striking transition from a weakly
348: interacting to a strongly interacting CF regime. The breakdown of
349: the weakly interacting CF picture is expected to result in the
350: formation of higher order Fractional Quantum Hall states that do
351: not belong to the primary CF sequence. The results of light scattering experiments
352: in conjunction with CF theories that go beyond the mean-field
353: non-interacting limit should reveal intriguing new physics of
354: composite quasiparticles in the regime of the FQHE.
355: \par
356: This work is supported by the National
357: Science Foundation under Grant No. NMR-0352738, by the Department
358: of Energy under Grant No. DE-AIO2--04ER46133, and by a research
359: grant from the W.M. Keck Foundation.
360:
361: \begin{thebibliography}{00}
362: \bibitem{Jain} J.K Jain, Phys. Rev. Lett. {\bf 63}, 199 (1989)
363: \bibitem{HLR} B.I. Halperin, P.A. Lee and N. Read, Phys. Rev. B {\bf 47}, 7312 (1993)
364: \bibitem{Pan} W. Pan et al., Phys. Rev. Lett. {\bf 90}, 016801 (2003)
365: \bibitem{Quinn} J.J. Quinn and A. W\'ojs, J. Phys. Condens. Matter {\bf 12}, R265 (2000)
366: \bibitem{Haldane} F.D.M. Haldane, Phys. Rev. Lett. {\bf 51}, 605 (1983)
367: \bibitem{Halperin} B.I. Halperin, Phys. Rev. Lett. {\bf 52}, 1583 (1984)
368: \bibitem{Wojs-Quinn} A. W\'ojs and J.J. Quinn, Phys. Rev. B {\bf61}, 2846 (2000)
369: \bibitem{Mandal-Jain} S.S. Mandal and J.K. Jain, Phys. Rev. B {\bf 66}, 155302 (2002)
370: \bibitem{Jain-Mandal} C.C. Chang, S.S. Mandal and J.K. Jain, Phys. Rev. B {\bf67}, 121305(R) (2003)
371: \bibitem{Lopez-Fradkin} A. Lopez and E. Fradkin, Phys. Rev. B {\bf 69}, 155322 (2004)
372: \bibitem{Goerbig1} M.O. Goerbig, P. Lederer and C.M. Smith, Phys. Rev. B {\bf 69}, 155324 (2004)
373: \bibitem{Goerbig2} M.O. Goerbig, P. Lederer and C.M. Smith, Europhys. Lett. {\bf 68}, 72 (2004)
374: \bibitem{Jain-chang} C.C. Chang and J.K. Jain, Phys. Rev. Lett. {\bf 92}, 196806 (2004)
375: \bibitem{Wojs-Quinn2} A. W\'ojs, K.-S Yi and J.J. Quinn, Phys. Rev. B {\bf 69}, 205322 (2004)
376: \bibitem{Irene1} I. Dujovne et al., Phys. Rev. Lett. {\bf 90}, 036803 (2003)
377: \bibitem{Murthy} G. Murthy, Phys. Rev. B {\bf 60}, 13702 (1999)
378: \bibitem{Mandal} S.S. Mandal and J.K. Jain, Phys. Rev. B {\bf 64}, 125310 (2001)
379: \bibitem{Irene2} I. Dujovne et al., Solid State Comm. {\bf 127}, 109 (2003)
380: \bibitem{Cyrus-thesis} C.F. Hirjibehedin, PhD Thesis, Columbia University (2004)
381: \bibitem{Gallais} Y. Gallais et al., Physica E, to appear (2006)
382: \bibitem{Mandal-Jain-2} S.S. Mandal and J.K. Jain, Phys. Rev. B {\bf 63}, 201310(R) (2001)
383: \bibitem{Irene3} I. Dujovne et al., Phys. Rev. Lett. {\bf 95}, 056808 (2005)
384: \bibitem{Cyrus} C.F. Hirjibehedin, A. Pinczuk, B.S. Dennis, L.N. Pfeiffer and K.W. West, Phys. Rev. Lett. {\bf 91}, 186802 (2003)
385: \bibitem{Byszewski} M. Byszewski et al., Nature Phys. {\bf 2}, 239-243 (2006)
386: \end{thebibliography}
387:
388:
389: \end{document}
390: