0% found this document useful (0 votes)
9 views4 pages

Coherent Branched Flow in A Two-Dimensional Electron Gas. Nature 410, 183-186 (2001)

Uploaded by

3289881606
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
9 views4 pages

Coherent Branched Flow in A Two-Dimensional Electron Gas. Nature 410, 183-186 (2001)

Uploaded by

3289881606
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 4

letters to nature

perovskite-type compounds [(1−x)La, xCa]MnO3. Phys. Rev. 100, 545–563 (1955). of quantum mechanical phase coherence in the electron flow.
19. Matsumoto, G. Study of (La1−xCax)MnO3. I. Magnetic structure of LaMnO3. J. Phys. Soc. Jpn
29, 606–615 (1970).
These findings may have important implications for a better
20. Kaplan, M. D. & Vekhter, B. G. Cooperative Phenomena in Jahn-Teller Crystals (Plenum, New York, understanding of electron transport in 2DEGs and for the
1995). design of future nanostructure devices.
21. Saitoh, T. et al. Electronic structure of La1−xSrxMnO3 studied by photoemission and x-ray absorption Images of electron flow from the quantum point contact (QPC)
spectroscopy. Phys. Rev. B 51, 13942–13951 (1995).
22. Arima, T. & Tokura, Y. Optical study of electronic structure in perovskite-type RMO3 (R ¼ LaY;
are obtained by raster scanning a negatively charged scanning probe
M ¼ Sc, Ti, V, Cr, Mn, Fe, Co, Ni, Cu). J. Phys. Soc. Jpn 64, 2488–2501 (1995). microscope (SPM) tip above the surface of the device and simulta-
23. Iliev, M. N. et al. Raman spectroscopy of orthorhombic perovskite YMnO3 and LaMnO3. Phys. Rev. B neously measuring the position-dependent conductance of the
57, 2872–2877 (1998).
24. Yamamoto, K., Kimura, T., Ishikawa, T., Katsufuji, T. & Tokura, Y. Raman spectroscopy of the charge-
device. The negatively charged tip capacitively couples to the
orbital ordering in layered manganites. Phys. Rev. B 61, 14706–14715 (2000). 2DEG, creating a depletion region that backscatters electron
25. Podobedov, V. B., Weber, A., Romero, D. B., Rice, J. P. & Drew, H. D. Effect of structural and magnetic waves. When the tip is positioned over areas with high electron
transitions in La1−xMxMnO3 (M ¼ Sr, Ca) single crystals in Raman scattering. Phys. Rev. B flow from the QPC, the conductance is decreased, whereas when the
58, 43–46 (1998).
26. Abrashev, M. V. et al. Comparative study of optical phonons in the rhombohedrally distorted
tip is over areas of relatively low electron flow the conductance is
perovskites LaAlO3 and LaMnO3. Phys. Rev. B 59, 4146–4153 (1999). unmodified. By raster scanning the tip over the sample, a two-
27. Inoue, J. et al. Raman scattering by orbital waves in perovskite LaMnO3. Physica B 237–238, 51–53 dimensional image of electron flow can be obtained.
(1997).
The QPC sample is mounted in an atomic force microscope and
28. Schaack, G. in Light Scattering in Solids VII (eds Cardona, M. & Güntherodt, G.) 24–173 (Springer,
Berlin, 2000).
cooled to liquid He temperatures. The QPC is formed in the 2DEG
29. Mitchell, J. F. et al. Structural phase diagram of La1−xSrxMnO3+d: Relationship to magnetic and inside a GaAs/AlGaAs heterostructure by negatively biasing two
transport properties. Phys. Rev. B 54, 6172–6183 (1996). gates on the surface—a negative potential on these gates creates two
30. Hirota, K., Kaneko, N., Nishizawa, A. & Endoh, Y. Two-dimensional planar ferromagnetic coupling in
LaMnO3. J. Phys. Soc. Jpn 65, 3736–3739 (1996).
depletion regions that define a variable-width channel between
them, as shown in Fig. 1a. The heterostructure for the devices
used in this experiment was grown by molecular beam epitaxy on
Acknowledgements
an n-type GaAs substrate. The 2DEG resides 57 nm below
We thank N. Nagaosa, T. Katsufuji, P. Prelovšek and S. E. Barnes for discussions. This work
was supported in part by NEDO Japan, CREST Japan, the Science and Technology Special
the surface with mobility m = 1.0 × 106 cm2 V−1 s−1 and density
Coordination Fund for Promoting Science and Technology, and Grant-in-Aid for n = 4.5 × 1011 cm−2. These values of mobility and density correspond
Scientific Research Priority Area from the Ministry of Education, Science and Culture of to a mean free path l = 11 mm, Fermi wavelength lF = 37 nm, and
Japan. S.O. acknowledges the financial support of JSPS. Part of the numerical calculation
was performed in the supercomputing facilities in IMR, Tohoku University.

Correspondence and requests for materials should be addressed to Y.T.


a
(e-mail: [email protected]).

.................................................................
Coherent branched flow in a
two-dimensional electron gas
M. A. Topinka*, B. J. LeRoy†, R. M. Westervelt*†, S. E. J. Shaw†,
R. Fleischmann‡, E. J. Heller†§, K. D. Maranowskik & A. C. Gossardk

* Division of Engineering and Applied Sciences, † Department of Physics,


§ Department of Chemistry and Chemical Biology, Harvard University, b
Cambridge, Massachusetts 02138, USA
‡ Max-Planck-Institut für Strömungsforschung, Bunsenstraße 10, 8
D-37073 Göttingen, Germany
k Materials Department, University of California, Santa Barbara,
California 93106, USA 6
500 nm
G (e2/h)

.................................. ......................... ......................... ......................... ......................... ........


Semiconductor nanostructures based on two-dimensional elec- 4
tron gases (2DEGs) could form the basis of future devices for
sensing, information processing and quantum computation.
Although electron transport in 2DEG nanostructures has been 2
well studied, and many remarkable phenomena have already been
discovered (for example, weak localization, quantum chaos, uni-
versal conductance fluctuations1,2), fundamental aspects of the 0
–1.0 –0.8 –0.6
electron flow through these structures have so far not been
Vg (V)
clarified. However, it has recently become possible to image
current directly through 2DEG devices using scanning probe Figure 1 Experimental set-up. a, Schematic diagram of the experimental set-up used for
microscope techniques3–13. Here, we use such a technique to imaging electron flow. The tip introduces a movable depletion region which scatters
observe electron flow through a narrow constriction in a electron waves flowing from the quantum point contact (QPC). An image of electron flow is
2DEG—a quantum point contact. The images show that the obtained by measuring the effect the tip has on QPC conductance as a function of tip
electron flow from the point contact forms narrow, branching position. Two ohmic contacts ,1 mm away from the QPC (not shown) allow the
strands instead of smoothly spreading fans. Our theoretical study conductance of the QPC to be measured using an a.c. lock-in amplifier at 11 kHz. The
of this flow indicates that this branching of current flux is due to root-mean-square voltage across the QPC, 0.2 mV, was chosen in order not to heat
focusing of the electron paths by ripples in the background electrons significantly above the lattice temperature of 1.7 K. b, Conductance of the QPC
potential. The strands are decorated by interference fringes used for Fig. 2b versus QPC width controlled by the gate voltage. Steps at integer
separated by half the Fermi wavelength, indicating the persistence multiples of 2e2/h are clearly visible. The inset is a topographic AFM image of the QPC.

NATURE | VOL 410 | 8 MARCH 2001 | www.nature.com © 2001 Macmillan Magazines Ltd 183
letters to nature
Fermi energy EF = 16 meV. The conductance of the QPC, shown in QPC with disordered background potentials. Our model for the
Fig. 1b, increases as the width of the channel is increased (by potential landscape incorporates known properties of 2DEGs in
changing the gate voltage Vg) and shows well defined conductance GaAs/AlGaAs heterostructures14,15 and replicates the physical para-
plateaus at integer multiples of the conductance quantum 2e2/h meters of the experimental system. To model the full in-plane
(refs 1, 2). potential, we consider two contributions. The first contribution is
Figure 2a and b shows images of electron flow from two different from the negatively charged gates that define the QPC, which we
QPCs at the temperature 1.7 K; both QPCs are biased on the G = model using a smooth analytic potential that reproduces the
2e2/h conductance plateau. Figure 2b shows the flow patterns on quantized conductance steps characteristic of the QPC. The Fermi
each side of a QPC and Fig. 2a shows a higher-resolution image of level remains constant inside the 2DEG, so that the propagation
flow from one side of a different QPC. In both these images, the there is primarily affected by the second piece of the potential, the
current leaves the point contact in a central lobe, as expected from disordered background. There are two contributions to this back-
an exact quantum-mechanical calculation of electron flow through ground that we considered: the donor dopant atoms and unwanted
an ideal QPC. Rather than continuing out as a smoothly widening impurities. For the impurities, distributed throughout the crystal
fan, it quickly forks into several different paths and continues to structure, we take a random distribution in three dimensions and
branch off into ever smaller rivulets for the full width of the scan. match the impurity atom density with a reasonable estimate for the
This branching behaviour was observed in all of the 13 QPC exit physical sample (1.25 × 1015 cm−3). The donors are located in a plane
patterns observed so far. Previously, there have been suggestions of displaced from the 2DEG by 22 nm with a sheet density of 8 ×
an unexpected narrowness in observed flow from a QPC6, but until 1012 cm−2; we estimate that half of the donors are ionized. Their
now, high-resolution, detailed images of electron flow from a QPC distribution is random in space except for correlations limiting the
have been difficult to obtain and no observations or predictions for maximum local density14. We use a 1/r3 potential at large distances
this type of strong branching behaviour have been published. The for the impurities and donors, where r is the distance between a
average electron flow reflected by the tip back through the QPC falls location in the plane of the 2DEG and the location of a given
off approximately as 1/r2 with distance r from the QPC. impurity or donor atom in the crystal. This 1/r3 dependence is the
In order to explore the experimentally observed current flow, we key feature of the full screened potential in a 2DEG from a point
numerically calculated a set of conducting wavefunctions through a charge15.

1 µm

1 µm
∆G: 0.00e /h 2
–0.25e2/h

Figure 2 Experimental images of electron flow. a, Image of electron flow from one side of by lF/2, half the Fermi wavelength, are seen to persist across the entire scan. b, Images
a QPC at T = 1.7 K, biased on the G = 2e2/h conductance step. Dark regions correspond of electron flow from both sides of a different QPC, again biased on the G = 2e2/h
to areas where the tip had little effect on QPC conductance, and hence are areas of low conductance step. The gated region in the centre was not scanned. Strong channelling
electron flow. The colour varies and the height in the scan increases with increasing and branching are again clearly visible. The white arrow points out one example of the
electron flow. Narrow branching channels of electron flow are visible, and fringes spaced formation of a cusp downstream from a dip in the potential.

184 © 2001 Macmillan Magazines Ltd NATURE | VOL 410 | 8 MARCH 2001 | www.nature.com
letters to nature
Using the values for donor inhomogeneity and impurity density has not been studied in detail, falling as it does between the well
expected for the 2DEG sample in our model14, we calculated the explored extremes of gaussian white noise and widely spaced point
mobility classically using the momentum relaxation time for an scatterers. A detailed study of this phenomenon will appear else-
appropriate ensemble of trajectories and found it to match the where (S.E.J.S. and E.J.H., manuscript in preparation). Similar
measured value to within 10%. The distribution of energies in the scattering, with the unexpected resulting strands of concentration
disordered potential is very nearly gaussian with a standard devia- of flux, might be expected in widely different contexts, such as
tion of about 8% E F . The correlation length of the disordered sound propagation through the ocean, where it has recently been
potential is about 25 nm, that is, of the order of the wavelength. discussed16.
In Fig. 3a we show a typical potential used in both classical and The experimental images in Fig. 2 show the position-dependent
quantum simulations. Figure 3b and c shows the results of classical effect of the SPM tip on the conductance through the device. The
and quantum-mechanical flux density calculations in that potential charged tip induces an approximately lorentzian bump in the
which clearly show branched flow. The agreement between the potential17 seen by electrons propagating through the system; we
classical and quantum results is very good, leading to the conclusion can simulate the experiments by adding a lorentzian to the potential
that the branched flow is essentially a classical phenomenon. at the tip position. Figure 4a shows the overall computed current
Though there are occasional events where the flow is split by an flow through the device before the addition of a lorentzian. We
impurity near the 2DEG, most of the bumps in the potential are well compared the simulated flux in a small patch (Fig. 4b) with the
below the Fermi energy of the electrons. Locally, potential valleys act conductance as a function of lorentzian position (Fig. 4c). This
like lenses that focus the electron paths, albeit not perfectly, giving comparison confirms the relationship between flux and the images
rise to a near-focal point known as a cusp. One example of such a achieved by the experimental technique.
cusp in the experimental images may be seen in Fig. 2a, as indicated A striking feature of the experimental images (Fig. 2a and b) is the
by the arrow. The stronger branches surviving at large distances are appearance of fringes oriented perpendicularly to electron flow and
the indirect result of passing over many hills and valleys (S.E.J.S. and spaced by lF/2, half the Fermi wavelength. These fringes are caused
E.J.H., manuscript in preparation), as shown in Fig. 3. by coherent constructive and destructive backscattering of electrons
This classical phenomenon is robust, and it is seen in smooth from the tip. In the region close to the point contact the fringes can
potentials of disparate origin with features generally well below the be explained by multiple reflections of electron waves between the
energy of the scattered particles. The length scale for the formation tip and the QPC gates. At farther distances two effects could
of the channels is determined by the autocorrelation length of the suppress the fringes: phase decoherence and thermal broadening.
potential. This regime of classical dynamics in a random potential In our sample the phase coherence length, lf = 18 mm18, is much
larger than our scan area and thus not a concern. The thermal length
lth ¼ ~2p=mlF kB T ¼ 1:4 mm (where T is temperature, kB is Boltz-
mann’s constant, ~ is Planck’s constant divided by 2p, m is the mass
of the electron, and l F is the Fermi wavelength), the distance over
which electrons differing in energy by kBT drift 1 radian out of
phase, is relevant. The fringes in our experimental and theoretical
images, however, persist up to this length and beyond.
The persistence of the fringes can be explained by an unusual
source of coherent backscattering (S.E.J.S., R.F. and E.J.H., manu-

Figure 3 Calculated electron flow. a, Surface plot of the random potential for computed
electron flow, including contributions from impurities, donors, and gates; green areas
are low and white areas are high potential. The ‘shadow’ is cast by classical flux
through the same potential. We note that the branched flux does not follow valleys in Figure 4 Calculated tip scan. a, Quantum-mechanical flux through a random potential.
the potential. b, Classical and c, quantum-mechanical flux of electrons flowing through b, The flux from the boxed area in a. c, A raster scan of conductance as a function of SPM
the potential in a. In the classical case, we followed the dynamics of an appropriate tip position in the same system as a and b. The conductance image in the model
ensemble of classical trajectories and show the classical flux density. The quantum- corresponds to the flux image, confirming our assertion that the experiment images
mechanical results show the flux density of the transmitted wavefunction, coming electron flow. Additionally, the simulation c shows quantum fringes, as seen in the
through the point contact on the left. We note that both results show the same experiment. Though this simulation is at zero temperature, the fringes do survive thermal
branching behaviour. averaging.

NATURE | VOL 410 | 8 MARCH 2001 | www.nature.com © 2001 Macmillan Magazines Ltd 185
letters to nature
script in preparation). The electron waves that are backscattered 12. Yoo, M. J. et al. Scanning single-electron transistor microscopy: imaging individual charges. Science
276, 579–582 (1997).
from the depletion region under the tip (located a distance rtip from 13. Gurevich, L., Canali, L. & Kouwenhoven, L. P. Scanning gate spectroscopy on nanoclusters. Appl. Phys.
the QPC), interfere with a combination of the backscattered waves Lett. 76, 384–386 (2000).
from all impurities in a ring covering the area rtip 6 lth away from the 14. Grill, R. & Döhler, G. H. Effect of charged donor correlation and Wigner liquid formation on the
transport properties of a two-dimensional electron gas in modulation delta-doped heterojunctions.
QPC to produce alternating constructive and destructive total
Phys. Rev. B 59, 10769–10777 (1999).
interference as rtip is changed. As the tip is moved, the phase of 15. Davies, J. H. The Physics of Low-Dimensional Semiconductors: An Introduction (Cambridge Univ. Press,
the waves backscattered from the tip seen at the QPC varies as 2rtipk, New York, 1998).
where k is the wavevector. Importantly, this mechanism is resistant 16. Wolfson, M. A. & Tomsovic, S. On the stability of long-range sound propagation through a structured
ocean. At ηxxx.lanl.gov/abs/nlin.CD/0002030ι (2000).
to thermal broadening, because as k is varied around kF the phases of 17. Eriksson, M. A. et al. Effect of a charged scanned probe microscope tip on a subsurface electron gas.
backscattered waves returning from impurities within the radius Superlattices Microstruct. 20, 435–440 (1996).
rtip 6 lth vary in step with the phase from the tip (S.E.J.S., R.F. and 18. Altshuler, B. L., Aronov, A. G. & Khmelnitsky, D. E. Effects of electron-electron collisions with small
E.J.H., manuscript in preparation) with the result that the fringes energy transfers on quantum localisation. J. Phys. C 15, 7367–7386 (1982).

can easily survive. This mechanism requires phase-coherent trans-


port over the round-trip distance to the most remote fringes, and Acknowledgements
may provide a new direct way to measure electron wave coherence This work was supported in part at Harvard University by the Office of Naval Research/
length. Augmentation Awards for Science and Engineering Research Training (ONR/AASERT),
Another unusual feature of the fringes is that their separations are by ONR and by the National Science Foundation through grants for the Materials
Research Science and Engineering Center and for the Institute for Theoretical Atomic and
smaller than the width of the lorentzian perturbation used in Molecular Physics at Harvard University and Smithsonian Astrophysical Observatory.
making the measurements. The tip perturbation has an estimated Work at the University of California Santa Barbara was supported by the NSF Science and
half-width at half-maximum of ,60 nm, based on electrostatic Technology Center QUEST.
simulations (ref. 17 and M.A.T. et al., manuscript in preparation).
Correspondence and requests for materials should be addressed to R.M.W.
Our numerical work, however, gives evidence that the relevant (e-mail: [email protected]).
feature is the presence of a classically forbidden ‘‘depletion
region’’ beneath the tip. The edge of this depletion region provides
the backscattering needed to direct electrons back through the QPC.
This result is in accord with experimental findings that images of
electron flow are only observed when the voltage on the tip is
sufficient to create a small depletion region in the 2DEG (M.A.T.
.................................................................
et al., manuscript in preparation). Moreover, backscattering reflec- Strongly linked current flow in
tion from a circular depletion region where paths reverse and lead
back to the QPC comes only from a small zone on the depletion polycrystalline forms of
region.
The information about electron flow made available by this the superconductor MgB2
imaging technique has several important implications and possible
future applications. The observation that electrons can flow D. C. Larbalestier*†, L. D. Cooley*, M. O. Rikel*, A. A. Polyanskii*,
through a 2DEG in narrow, branching channels may be important J. Jiang*, S. Patnaik*, X. Y. Cai*, D. M. Feldmann*, A. Gurevich*,
for experiments relying on the assumption of ‘ballistic’ flow of A. A. Squitieri*, M. T. Naus*, C. B. Eom*†, E. E. Hellstrom*†, R. J. Cava‡,
electrons for distances less than the mean free path. Such detailed K. A. Regan‡, N. Rogado‡, M. A. Hayward‡, T. He‡, J. S. Slusky‡,
images of electron flow may also prove valuable in future explora- P. Khalifah‡, K. Inumaru‡ & M. Haas‡
tions of spin transport for spintronics and possibly allow novel
* Applied Superconductivity Center, University of Wisconsin–Madison,
quantum computer implementations. The surprising persistence of
1500 Engineering Drive, Madison, Wisconsin 53706, USA
coherent fringes well past the thermal length may help provide † Department of Materials Science and Engineering, University of Wisconsin–
additional insight into a broad range of coherent phenomena Madison, 1509 University Avenue, Madison, Wisconsin 53706, USA
including universal conductance fluctuations, phase coherence, ‡ Department of Chemistry and Princeton Materials Institute,
and weak localization. This imaging technique can provide much Princeton University, Princeton, New Jersey 08544, USA
detail about coherent electron flow, and is an important tool in
investigating the underlying physics as well as the future design of .......................................... ......................... ......................... ......................... .........................
2DEG nanostructures. M The discovery of superconductivity at 39 K in magnesium
Received 25 October 2000; accepted 2 January 2001.
diboride1, MgB2, raises many issues, a critical one being whether
this material resembles a high-temperature copper oxide super-
1. Beenakker, C. W. J. & van Houten, H. in Solid State Physics (eds Ehrenreich, H. & Turnbull, D.) 44,
1–228 (Academic, New York, 1991).
conductor or a low-temperature metallic superconductor in terms
2. Sohn, L. L., Kouwenhoven, L. P. & Schon, G. Mesoscopic Electron Transport in Semiconductor of its behaviour in strong magnetic fields. Although the copper
Nanostructures (Kluwer Academic, Boston, 1997). oxides exhibit very high transition temperatures, their in-field
3. Topinka, M. A. et al. Imaging coherent electron flow from a quantum point contact. Science 289,
performance2 is compromized by their large anisotropy, the result
2323–2326 (2000).
4. Eriksson, M. A. et al. Cryogenic scanning probe characterization of semiconductor nanostructures. of which is to restrict high bulk current densities to a region much
Appl. Phys. Lett. 69, 671–673 (1996). less than the full magnetic-field–temperature (H–T) space over
5. Crook, R., Smith, C. G., Simmons, M. Y. & Ritchie, D. A. Imaging cyclotron orbits and scattering sites which superconductivity is found. Moreover, the weak coupling
in a high-mobility two-dimensional electron gas. Phys. Rev. B 62, 5174–5178 (2000).
6. Crook, R. et al. Imaging diffraction-limited electronic collimation from a non-equilibrium one-
across grain boundaries makes transport current densities in
dimensional ballistic constriction. J. Phys. Cond. Matter 12, L167–L172 (2000). untextured polycrystalline samples low and strongly sensitive to
7. McCormick, K. L. et al. Scanned potential microscopy of edge and bulk currents in the quantum Hall magnetic field3,4. Here we report that, despite the multiphase,
regime. Phys. Rev. B 59, 4654–4657 (1999). untextured, microscale, subdivided nature of our MgB2 samples,
8. Zhitenev, N. B. et al. Imaging of localized electronic states in the quantum Hall regime. Nature 404,
473–476 (2000).
supercurrents flow throughout the material without exhibiting
9. Finkelstein, G., Glicofridis, P. I., Ashoori, R. C. & Shayegan, M. Topographic mapping of the quantum strong sensitivity to weak magnetic fields3. Our combined mag-
Hall liquid using a few-electron bubble. Science 289, 90–94 (2000). netization, magneto-optical, microscopy and X-ray investigations
10. Manoharan, H. C., Lutz, C. P. & Eigler, D. M. Quantum mirages formed by coherent projection of
show that the supercurrent density is mostly determined by flux
electronic structure. Nature 403, 512–515 (2000).
11. Jamneala, T., Madhavan, V., Chen, W. & Crommie, M. F. Scanning tunneling spectroscopy of pinning, rather than by the grain boundary connectivity. Our
transition-metal impurities at the surface of gold. Phys. Rev. B 61, 9990–9993 (2000). results therefore suggest that this new superconductor class is not

186 © 2001 Macmillan Magazines Ltd NATURE | VOL 410 | 8 MARCH 2001 | www.nature.com

You might also like