Appl. Phys.

A 75, 229–235 (2002) Applied Physics A

DOI: 10.1007/s003390201334 Materials Science & Processing

a.m. song1,2,3,✉ Electron ratchet effect in semiconductor

devices and artificial materials with broken
centrosymmetry
1 Sektion Physik der Ludwig-Maximilians-Universität, Geschwister-Scholl Platz 1,
80539 München, Germany
2 Division of Solid State Physics and Nanometer Structure Consortium, Lund University,
221-00 Lund, Sweden
3 Department of Electrical Engineering and Electronics, UMIST, Manchester M60 1QD, UK

Received: 16 January 2002/Accepted: 11 February 2002 structed artificially. It was discovered more than 30 years ago
Published online: 22 April 2002 • © Springer-Verlag 2002 that some natural crystals can generate direct electron currents
under uniform illumination [15–17]. Referred to as the pho-
ABSTRACT Studies on nonlinear electron transport in nano-
tovoltaic effect, the phenomenon has been identified as being
meter-sized semiconductor devices with broken centrosymme-
a macroscopic manifestation of the lack of centrosymmetry
try are reviewed. In these devices, an applied alternating (rock-
ing) electric field induces a net flow of electrons in the direction in the microscopic elemental structure, despite the macro-
perpendicular to that of the applied field. Such an electron scopic homogeneity of these crystals. In this case, light serves
ratchet effect has been observed in a number of differently as a source of external fluctuation by imposing an unbiased,
designed devices, fabricated from two types of semiconduc- time-periodic electric field on the crystals, and the generation
tor material systems. The functionality is interpreted with an of direct electron current can be viewed as extremely fast
extended Büttiker–Landauer formula. We show that the de- rectification, or a ratchet effect, at the light frequency. One
vices operate at both cryogenic and room temperatures and at difference compared with a one-dimensional ratchet is that
frequencies up to at least 50 GHz. Based on a similar micro- the generated electron current may flow in a direction differ-
scopic mechanism, we have also constructed, to the best of ent from, or even perpendicular to, that of the driving electric
our knowledge, the first artificial electronic nanomaterial that field of the light. This is due to the broken symmetry of the
operates at room temperature. The promising possibilities for microscopic potential in two or three dimensions. With the
practical applications, such as rectification, microwave detec- high spatial resolution of modern semiconductor fabrication
tion, second-harmonic generation, etc., are also discussed. technologies, it is possible to make different types of artificial
PACS 73.23.Ad; 73.40.Ei; 73.50.Fq; 81.05.Zx; 07.57.Kp electron ratchets in well-controlled ways. Direct voltages or
currents induced by rocking electric fields have been observed
at THz frequencies [18], in electron billiard structures [19],
1 Introduction and in the quantum transport regime [20, 21].
When a system with a broken spatial symmetry is Here, our recent work on the ratchet effect, realized in
taken out of equilibrium by, for example, introducing an ex- nanometer-sized electron billiard structures, is reviewed. In
ternal fluctuation (i.e. the time average is zero), a net flow of our devices, instead of introducing asymmetric, sawtooth-like
particles along a certain direction may occur [1–9]. A sim- ratchet potentials, only the device geometries (or shapes) are
ple way to observe this ratchet effect is to construct a one- tailored to break the spatial reversion symmetry. The elec-
dimensional, periodic, asymmetric (typically sawtooth-like) tron transport is ballistic, i.e. the sizes of the devices are
potential. In operation, both the external fluctuation force and smaller than the electron scattering length, le , and the elec-
the generated net flow of particles are along the potential- trons in the devices are mainly scattered from the designed
modulation direction. The ratchet effect has been suggested device boundary rather than by randomly distributed impu-
to be an underlying mechanism of some biological phenom- rities [22]. The observed nonlinear effect is so strong that
ena, such as molecular motors [1, 2, 7, 10]. Recently, sev- the devices are capable of rectification applications and are
eral groups have invented devices consisting of arrays of hence called ballistic rectifiers. The devices generally have
symmetry-breaking microscopic obstacles and used them to four terminals or electric contact leads. When a regular ac
sort biomolecules [11–14]. This shows that asymmetric ge- voltage or a random signal is applied via two of the elec-
ometries may provide the same functionality as asymmetric trical contact leads, a direct voltage or current is generated
potentials in creating a guided flow of particles. and can be measured through the other two terminals. This
There are also ratchets in which the mobile particles are is similar to the operation of a bridge rectifier, but our de-
electrons. Electron ratchets may exist naturally or can be con- vices have a completely different working principle. Not
only do they operate at cryogenic temperatures, our smallest
✉ Current and permanent address: Department of Electronical Engin- devices have also been shown to work at room tempera-
eering and Electronics, UMIST, PO Box 88, Manchester M60 1QD, UK. ture, with a rectification efficiency reaching 14%. Our recent
Fax: +44-161/2004770, E-mail: aimin.song@ftf.lth.se experiments have also demonstrated that the ballistic rec-
230 Applied Physics A – Materials Science & Processing

tifiers function at frequencies up to at least 50 GHz and affected. This gives rise to a finite negative voltage between L
have a sensitivity to microwaves roughly as high as that of and U, VLU .
a commercial microwave-detection diode. Based on a simi- The mirror symmetry along the central L–U axis results in
lar microscopic mechanism, we have also constructed, to the
best of our knowledge, the first artificial electronic nano-
VLU (ISD ) = +VLU (−ISD ) . (1)
material that operates at room temperature. Some promising
possibilities for practical applications, such as rectification,
microwave detection, higher harmonic generation, etc., will Correspondingly, RSD,LU (ISD ) = −RSD,LU (−ISD ), where the
be discussed. four-terminal resistance RSD,LU = VLU /ISD . Such a picture is
almost perfectly supported by the experimental VLU vs. ISD
curve at T = 4.2 K, shown in Fig. 2. The slight deviations
2 Experiments on GaAs–AlGaAs-based ballistic
from (1) are attributed to unintentional breaking of the desired
rectifiers
symmetry along the L–U axis by the imperfection of the fab-
Figure 1a shows an atomic force microscope image rication. Any imperfection is expected to contribute a linear
of the central part of one of the first ballistic rectifiers. Device term to the VLU vs. ISD curve. This is similar to a change of
fabrication starts with a modulation-doped GaAs–AlGaAs the value of one of the four resistors in an otherwise balanced
heterostructure with a two-dimensional electron gas located resistor bridge.
37 nm below the wafer surface. The electron density is about In the I –V characteristic of a two-terminal nonlinear de-
5 × 1011 cm−2 and the mobility about 5 × 105 cm2 V−1 s−1 at vice, such as a diode, the nonlinearity always comes on top
a temperature of T = 4.2 K, giving le = 5.8 µm. Following of a (often very large) linear term. For the ballistic rectifier
the pattern definition by electron-beam lithography, the dark with the four-terminal geometry, there can, however, be no lin-
areas in Fig. 1a are etched and become nonconductive for ear term present in the relation between output voltage and
electrons. As a result, the triangular antidot is defined in the applied current, as shown by (1). The striking nonlinearity is
cross-junction formed by the two narrow channels (labeled a result of the mirror symmetry along the U–L axis and the
“source” S and “drain” D) and the two wide channels (labeled broken symmetry along the S–D axis. As will be shown in
“upper” U and “lower” L). Sects. 4 and 5, this has quite a few advantages in practical
Since le is larger than the central part of the device, spec- applications.
ular electron scattering from the etched boundaries dominates Based on a similar working principle, other types of bal-
the transport properties. It is therefore possible to alter the de- listic rectifiers can be designed. The inset in Fig. 3 shows
vice properties in the ballistic electron transport regime by schematically a device containing channels defined by gates
simply changing the shape of the device (as will be shown be- rather than by chemical etching. Instead of introducing
low). This is in contrast to the case of the diffusive electron a symmetry-breaking triangular scatterer as in Fig. 1a, the
transport in devices much larger than le , where an electron mirror symmetry along the S–D axis is broken by the horn-
current consists of electrons undergoing a large number of shaped U and L channels. The measured VLU vs. ISD curves at
scattering events with, for example, randomly distributed im- different gate voltages. VGU , applied between the gates and the
purities, and the electron transport is not sensitive to a change upper lead, are plotted in Fig. 3. As expected, the same result
in device shape. is obtained when the gate voltages are applied using the lower
In our experiments, we apply current to leads S and D, and lead as the ground. Because of the imperfection of the fabrica-
we detect the output voltage via L and U. The typical electron tion, the curves are not perfectly symmetric along the ISD = 0
trajectories, illustrated by the arrows in Fig. 1a, suggest an ac- axis. Nevertheless, it is obvious that by increasing the negative
cumulation of electrons in the lower lead and hence a negative gate voltage from −1.2 to −1.5 V, the nonlinear component of
voltage between L and U. Obviously, this cannot happen when I –V curves increases, which can be explained by the model
the applied voltage or current is zero, since the same number (6) in Sect. 3.
of electrons travel along the opposite directions of the arrows.
For a nonzero applied current, ISD = 0, however, as shown
by a detailed model [23, 24], the electron transmission along 0
the arrows in Fig. 1a is changed by the applied electric field,
T=4.2 K
while the transmission in the reversed direction is virtually un-
VLU (mV)

-0.1

-0.2

-40 -20 0 20 40
ISD (µA)
FIGURE 2 The output voltage between the lower and upper leads, VLU ,
FIGURE 1 Atomic force microscope image of the central part of the device as a function of the input current through the source and drain leads, ISD ,
(a), which has a similar functionality to a bridge rectifier (b) measured at 4.2 K
SONG Electron ratchet effect in semiconductor devices and artificial materials with broken centrosymmetry 231

0.01 For a mesoscopic conductor that is connected via perfect

VGU=-1.2 V leads to a number of carrier reservoirs, the scattering approach
T=4.2 K yields the current through lead α:

0 VGU=-1.4V 2e
 
Iα = f(E − µα ) − f(E − µβ )
VLU (mV)

h β=α
VGU=-1.5V × Tβ←α (E, B)dE . (2)
-0.01 1 µm
U
Here, µα is the chemical
 potential of reservoir α and f(E −
−1
S D gate

-0.02 µα ) = exp kB T + 1
E−µα
is the Fermi–Dirac distribution
function. Tβ←α (E, B) is the transmission coefficient for carri-
ers from lead α to lead β at energy E and magnetic field B .
e L e At kB T = 0 and B = 0, (2) becomes
-0.03
2e

-2 -1 0 1 2
ISD (µA) Iα = T [β,α](µα − µβ ) . (3)
FIGURE 3 VLU vs. ISD curves at different gate voltages, VGU , measured h β=α
in a ballistic rectifier fabricated with gates rather than chemical etching, as
µ
schematically shown in the inset. The dark areas in the inset are gates, to
Here, T [β,α] equals µβα Tβ←α (E, B)dE/(µα − µβ ) if µα >
which the same gate voltage is applied. The arrows in the inset illustrate µ
typical electron trajectories µβ , and µαβ Tα←β (E, B)dE/(µβ − µα ) otherwise. This shows
that only the transmissions from reservoirs with higher chem-
If an ac voltage is applied to leads S and D, as expected ical potentials to reservoirs with lower chemical potentials
from the characteristics shown in Figs. 2 and 3, a nega- need to be considered.
tive average (dc) voltage can be observed between L and U. As in the model for ballistic electron transport in a cross-
This will be demonstrated below using InP–InGaAs-based de- junction by Beenakker and van Houten [48], the angular dis-
vices. Actually, a direct voltage output has been observed, tribution of the ballistic electrons ejected from S and D, P(θ),
even when an external noise signal is applied [25], showing largely determines the electron transmission probabilities into
a ratchet effect induced by the broken geometric symme- the L or U lead. In principle, the transmission coefficients at
try and the guidance of ballistic electrons by the triangular a finite value of ISD should be obtained by first calculating the
antidot. self-consistent potential in the device. Such a calculation is
Because the device works similarly to a bridge rectifier very complex. Instead, we consider there to be voltage drops
(see Fig. 1b), it is called a ballistic rectifier. However, only at the entrances and exits of the S and D channels [46]. This
a single device is used here, rather than four diodes as in only results in a change in the velocity component of an elec-
the bridge rectifier. The mechanism is also entirely different, tron along the channel direction, vx , by an amount of the mean
since no doping junction or barrier structure along the current velocity or drift velocity, ∆v, while the velocity component
direction is used in our devices. The pronounced nonlinear in the perpendicular direction, v y , is essentially not affected.
effect comes from the broken symmetry and the ballistic elec- The ejection angle of the electron changes from arctan(v y /vx )
tron transport nature. As a result, as will be shown in Sect. 3, to arctan[v y /(vx ± ∆v)] at ISD = 0, meaning that the trans-
the ballistic rectifier has no intrinsic voltage or current thresh- mission probabilities of electrons into the L or U leads are
old, and it can be used for the detection of very weak signals directly related to the electron drift velocity, and hence ISD , as
without the need of an external bias. is calculated and shown in Fig. 4. This is a particularly good
approximation for this specific device geometry because the
3 Model based on classical ballistic electron electron transmission probabilities are very sensitive to the
scattering ejection angles of the electrons. This allows (3) to be written
as
The Büttiker–Landauer formalism [26, 27], which 2e

treats transport in mesoscopic conductors as a transmis- Iα ≈ T [β,α](I[β,α] )(µα − µβ ) , (4)
sion problem for carriers at the Fermi level, is widely h β=α
used to describe the linear transport behavior of these
conductors [28–34]. Recently, a rapidly increasing number of where T [β,α](I{β,α} ) is equal to T β←α (Iα ) if µα > µβ and
experimental [18, 20, 35–40] and theoretical [41–47] studies T α←β (Iβ ) otherwise.
on nonlinear ballistic transport have been published. Non- The relation between VLU and ISD is obtained via the four-
linearity is particularly important in mesoscopic conductors, terminal resistance
because of the small device sizes and the fact that, in prin- VLU 3h sin θe − sin θ0
ciple, nonlinearity can be induced by any non-zero applied RSD,LU ≡ = , (5)
ISD 2 e2 2NLU − 3NSD (1 − sin θ0 )2
current or voltage. By extending the Büttiker–Landauer for-
mula to the nonlinear regime and the framework of classical where θe = θ0 + arcsin [(∆v/vF ) sin θ0 ], θ0 ≈ π/4 is the mini-
ballistic electron scattering, we have obtained a model which mum ejection angle for an ejected electron to be scattered by
provides analytical descriptions of the ballistic rectification the triangular antidot to the lower lead, vF is the Fermi vel-
effect [23]. ocity, and NSD and NLU are the numbers of propagating modes
232 Applied Physics A – Materials Science & Processing

21 4 gate voltage, as shown in Fig. 3. Based on the above picture,

the nonlinear ballistic electron transport at finite magnetic
20.5 3.5
fields can also be well described [50].
TL<-S

TU<-S
20 3 4 InGaAs–InP-based room-temperature ballistic
rectifiers
19.5 2.5 As the miniaturization of conventional semicon-
ductor devices is approaching physical limits, great effort has
19 2 been made to develop new types of nanodevices, which, from
-40 -30 -20 -10 0 the application point of view, should operate at room tempera-
ISD (µA) ture and GHz frequencies. The ballistic rectifier in Fig. 1a has
FIGURE 4 The calculated TL←S and TU←S as functions of the negative
already been demonstrated to work up to T = 77 K. To func-
source–drain current tion at room temperature, a device size smaller than le (about
100 nm) at T = 300 K is required. This is, however, very dif-
in channels S and D and channels L and U, respectively. The ficult using a GaAs/AlGaAs heterostructure, due to the large
VLU vs. ISD curve, calculated analytically, is shown by the depletion length (up to hundreds nm) close to etched device
solid line in Fig. 5. It is in very good agreement with the ex- boundaries. The much shorter depletion length (below 30 nm)
perimental result (dashed line) especially in the low current of the InGaAs–InP material system makes it possible to fabri-
regime, although there are no adjustable parameters at all in cate much smaller devices [51].
(5). At high currents, electron heating might take place and re- The inset of Fig. 6 shows an atomic force microscope
duce the rectification effect, which is not included in the above image of one of the smallest devices that we have fabricated
model. Moreover, the length of the L and U channels (5 µm) from a modulation-doped In0.75 Ga0.25 As–InP quantum-well
is comparable to the electron mean free path, which results in structure [52]. At room temperature, the density of the two-
some electrons changing direction before they pass through dimensional electron gas is 4.7 × 1015 m−2 and the mobility
the channels and enter the electrical contact regions (carrier 1.2 m2 V−1 s−1 , corresponding to le = 140 nm. Since le is only
reservoirs). This certainly causes a reduced output. slightly shorter than the distance from the left or right chan-
It is interesting to note that in the limit of ISD | → 0, VLU is nel to the triangular antidot, most electrons can still travel this
shown to have a quadratic response to ISD : distance without being scattered. As shown by Hirayama et
al. [53], if the electron transport is partially ballistic, the ballis-
3πh 2 (sin 2θ0 )ISD 2 tic effect may still be observed, although it will be weaker than
VLU = − . (6)
2e3 E F NSD 2NLU − 3NSD (1 − sin θ0 )2 at low temperatures. In our experiment, instead of a dc cur-
rent, we apply an ac voltage (1 kHz) to the device and measure
Although the calculation is performed for the condition of the dc (or average) voltage between the lower and upper termi-
T = 0, we can show that (6) is also valid for sufficiently small nals. Indeed, a rectification efficiency of about 14%, roughly
currents at finite temperature [49], meaning that there is no half of that at 100 K, is observed, as shown in Fig. 6. This is
threshold voltage for the ballistic rectifier. There are a few actually one of the very few types of novel nanodevices shown
advantages associated with the quadratic response, as will to work at room temperature so far.
be discussed in Sect. 4. Equation (6) indicates that |VLU | in-
creases with decreasing E F , NSD , and NLU , and this explains
the enhancement of the nonlinearity with increasing negative
0 Experiment
Quadratic fit
0 T=300 K
dc output (mV)

-20

-0.2
VL U (mV)

-40
-0.4 Experiment
Theory

-0.6 100 nm
-60
0 100 200 300 400
-40 -20 0 20 40 ac input (mV)
ISD (µA) FIGURE 6 Room-temperature operation of one of the smallest ballistic rec-
FIGURE 5 Theoretical (solid line) and experimental (dashed line, the same tifiers. The inset shows a scanning electron microscope image of the device.
as that in Fig. 2) VLU vs. ISD curves for the device shown in Fig. 1a The frequency of the input ac signal is 1 kHz
SONG Electron ratchet effect in semiconductor devices and artificial materials with broken centrosymmetry 233

Interestingly, at sufficiently low temperatures (not shown The fit y = ax 2 (where y represents the dc output voltage,
here), we have also observed that the output of smaller de- x the ac input voltage and a a constant) in Fig. 6 also sup-
vices, such as the one in Fig. 6, changes sign, while the output ports the theoretical prediction of the quadratic response to
of larger devices does not [54]. If a top gate is fabricated, one input voltage (6). This allows second-harmonic signals to be
can actually switch the sign of the output by tuning the gate generated without producing third or higher harmonics. The
voltage. polarization of the generated second-harmonic signal (along
We also perform high-frequency experiments up to the vertical direction) is perpendicular to that of the input sig-
50 GHz. Because of the in-plane nature of the ballistic rec- nal (horizontal direction), which makes separation from the
tifier, i.e. the electrical contacts are laterally separated rather input signal much easier. Because of the quadratic nature, one
than placed on the front surface and the back (substrate), the would also expect the dc output to be a linear function of the
parasitic capacitance between contacts is substantially lower applied microwave power, as is in fact demonstrated in Fig. 7.
than in a conventional vertical device of the same size. Fur-
thermore, the new working mechanism does not rely on any 5 Artificial nonlinear nanomaterials
minority carrier diffusion or barrier structure, two factors
that also often limit the speed of conventional semiconductor Based on a working principle similar to that of
diodes. The ballistic rectifier is therefore expected to function the ballistic rectifier, we have designed a nanostructured ma-
at very high frequencies. Figure 7 shows the dc output of the terial, constructed by the arrangement of nanometer-sized,
device versus the power of a 50-GHz signal at room tempera- symmetry-breaking elements into a two-dimensional lattice.
ture. Although we have not been able to test the devices at Figure 8 shows an atomic force microscope image of the
frequencies higher than 50 GHz so far, we expect, from the nanomaterial, fabricated from the same modulation-doped
working principle, the cut-off frequency to be much higher, In0.75 Ga0.25 As–InP quantum-well structure, using electron-
possibly up to hundreds of GHz or even in the THz regime. beam lithograph and chemical wet etching. Noticeably, the
It should be noted that the power (horizontal axis) in material is not a simple assembly of individual ballistic rec-
Fig. 7 is the output power from the signal source. Due to the tifiers. The offset introduced in neighboring columns of tri-
impedance mismatch between the signal source (50 Ω) and angular antidots is crucial. The triangular antidots are placed
the device (a few kΩ), as well as some power loss due to in such a coherent way that neighboring antidots support
the cables at 50 GHz, effectively only a small fraction of the each other by simultaneously scattering electrons and form-
power from the signal source is actually applied to the device. ing channels through which electrons are ejected. The typical
It is estimated that the real sensitivity of the device is a few electron trajectories, indicated by the arrows, suggest the ac-
hundred mV output per mW input, which is virtually as high cumulation of electrons in the upward direction. Indeed, the
as that of a commercial microwave detection diode. We should material generates a dc voltage between the lower and upper
point out that our devices have, however, not yet been specif- contacts when an ac voltage or electric field is applied to the
ically optimized. Much improvement can be expected in the left and right sides of the material, demonstrating an intrinsic
future, by, for example, further reducing the device size. For nonlinear functionality [55].
a device about half the size of the one shown in Fig. 6, one As the nanomaterial is intrinsically nonlinear, individual
can expect not only a much higher efficiency, but also stable devices can be made by simply cutting pieces from the mate-
performance that is insensitive to temperature changes up to rial according to the requirements for different applications.
room temperature. Moreover, it is also possible to achieve room-temperature op-

dc output
0

-1
50 GHz
- - -
e e
Room temperature
dc output (mV)

-2

-3
e e
-4

-5
+ + +
-6
0 0.5 1 1.5 2
Power (mW)
dc output
~ 150 nm

V
FIGURE 7 The output dc voltage of the device versus the power of the FIGURE 8 An atomic force microscope image of the nanomaterial. The
applied 50-GHz signal at room temperature arrows represent typical electron trajectories
234 Applied Physics A – Materials Science & Processing

0 6 Conclusions

We have demonstrated a new type of nonlinear

-10 nano-rectifier, which generates a strong ratchet effect. Based
on a similar microscopic mechanism, we have also con-
dc output (µV)

-20 50 GHz structed, to the best of our knowledge, the first artificial
Room temperature electronic nanomaterial that operates at room temperature.
The facts that they can be reproducibly fabricated and can
-30 function at room temperature as well as at 50 GHz suggest that
the ballistic rectifier and the artificial nanomaterial might be
-40 very close to practical applications.

ACKNOWLEDGEMENTS The work performed at Munich was

-50 supported through the Deutsche Forschungsgemeinschaft (SFB 348). Col-
-20 -15 -10 -5 0 5 laboration with A. Lorke, J.P. Kotthaus, A. Kriele, S. Manus, M. Streibl,
W. Wegscheider and M. Bichler is gratefully acknowledged. The author
Power (dBm) also thanks the Alexander von Humboldt Foundation for a Research Fel-
FIGURE 9 The output dc voltage of the nanomaterial versus the power of lowship. A. Löfgren, I. Maximov, P. Omling, L. Samuelson, W. Seifert, I.
the applied 50-GHz signal at room temperature. The rf signal is applied to the Shorubalko and H.Q. Xu at Lund University and H. Zirath at the Department
material via coplanar probes of Microwave Technology, Chalmers University of Technology, Göteborg,
Sweden, also contributed greatly to the work. The experiments at Lund were
supported by the Swedish Natural Science Research Council, the Swedish
eration by fabricating nanomaterials with “lattice constants” Research Council for Engineering Science, the “Quantum Materials” pro-
gram of the Swedish Foundation for Strategic Research and the European
comparable to, or shorter than, the electron mean free path. Commission through the LTR research project Q-SWITCH.
Similar to a ballistic rectifier, the in-plane nature of the nano-
material ensures that the parasitic capacitance between con- REFERENCES
tacts is much lower than in a conventional vertical device of
the same size. The nanomaterial thus has an extremely high 1 M.O. Magnasco: Phys. Rev. Lett. 71, 1477 (1993)
2 J. Prost, J.-F. Chauwin, L. Peliti, A. Ajdari: Phys. Rev. Lett. 72, 2652
working speed, as is evident in high-frequency experiments. (1994)
The curve in Fig. 9 shows the output dc voltage between the 3 C.R. Doering, W. Horsthemke, J. Riordan: Phys. Rev. Lett. 72, 2984
upper and lower contacts as a function of the rf power of (1994)
a 50-GHz signal, which is transmitted to the nanomaterial via 4 P. Hänggi, R. Bartussek: In Nonlinear Physics of Complex Systems, ed.
by J. Parisi, S.C. Müller, W. Zimmermann, Lecture Notes in Physics,
coplanar probes. The measured material is a 30 µm × 30 µm Vol. 476 (Springer, Berlin, Heidelberg 1996)
piece, with lattice constants of ax = 800 nm and a y = 300 nm 5 F. Jülicher, A. Ajdari, J. Prost: Rev. Mod. Phys. 69, 1269 (1997)
in the lateral and vertical directions, respectively, and both the 6 P. Reimann, M. Grifoni, P. Hänggi: Phys. Rev. Lett. 79, 10 (1997)
base and the height of the triangular antidots are 150 nm. The 7 R.D. Astumian: Science 276, 917 (1997)
8 R.D. Astumian, F. Moss (Eds.): Chaos 8, 533 (1998)
experiment is performed at room temperature. Similar results 9 P. Reimann: Phys. Rep. 361, 57 (2002)
are obtained in other materials with different structural param- 10 K. Kitamura, M. Tokunaga, A.H. Iwane, T. Yanagida: Nature 397, 129
eters. (1999)
Although we have not tested the material at frequencies 11 D. Erta: Phys. Rev. Lett. 80, 1548 (1998)
12 T.A.J. Duke, R.H. Austin: Phys. Rev. Lett. 80, 1552 (1998)
higher than 50 GHz, from the working principle we expect 13 C.F. Chou, O. Bakajin, S.W.P. Turner, T.A.J. Duke, S.S. Chan, E.C. Cox,
the cut-off frequency to be much higher, possibly up to hun- H.G. Craighead, R.H. Austin: Proc. Nat. Acad. Sci. USA 96, 13 762
dreds of GHz or even in the THz regime. The large area, in (1999)
combination with the in-plane layout of the material, enables 14 A. van Oudenaarden, S.G. Boxer: Science 285, 1046 (1999)
15 A.M. Glas, D. von der Linde, T.J. Negran: Appl. Phys. Lett. 25, 233
straightforward and efficient coupling to high-frequency mi- (1974)
crowaves. Moreover, microwaves travelling along the normal 16 V.I. Belinicher, B.I. Sturman: Sov. Phys. Usp. 23, 199 (1980)
direction of the nanomaterial surface yield the most efficient 17 B.I. Sturman, V.M. Fridkin: The Photovoltaic and Photorefractive Ef-
coupling, while coupling has to be achieved in specific ways fects in Noncentrosymmetric Materials (Gordon and Breach, Philadel-
phia 1992)
(including using antennae) in a conventional vertical device. 18 A. Lorke, S. Wimmer, B. Jager, J.P. Kotthaus, W. Wegscheider, M. Bich-
From the material point of view, to the best of our knowledge, ler: Physica B 249, 312 (1998)
this is the first artificial electronic nanomaterial that operates 19 A.M. Song, A. Lorke, A. Kriele, J.P. Kotthaus, W. Wegscheider,
at room temperature. M. Bichler: Phys. Rev. Lett. 80, 3831 (1998)
20 H. Linke, W. Sheng, A. Löfgren, H.Q. Xu, P. Omling, P.E. Lindelof:
An external noise signal can also induce a guided net flow Europhys. Lett. 44, 341 (1998)
of electrons in the upward direction and generate a direct 21 H. Linke, T.E. Humphrey, A. Lofgren, A.O. Sushkov, R. Newbury,
voltage [25]. In many aspects, the ratchet effect in the artifi- R.P. Taylor, P. Omling: Science 286, 2314 (1999)
cial nanomaterial resembles the photovoltaic effect in some 22 J. Imry: In Directions in Condensed Matter Physics, ed. by G. Grinstein,
G. Mazenko (World Scientific, Singapore 1986) pp. 101; C.W.J. Been-
natural crystals [15–17]. In our man-made artificial crystals, akker, H. Van Houten: Solid State Phys. 44, 1 (1991); S. Datta: Elec-
however, we are able to design (adjust) the size and shape of tronic Transport in Mesoscopic Systems (Cambridge University Press,
the symmetry-breaking elemental units, as well as the lattice Cambridge 1995)
23 A.M. Song: Phys. Rev. B 59, 9806 (1999)
constants. Such flexibility may also provide a new route in 24 A.M. Song, A. Lorke, A. Kriele, J.P. Kotthaus, W. Wegscheider,
further studies of the photovoltaic effect, and possibly some M. Bichler: In Proc. 24th Int. Conf. Phys. Semicond., ed. by D. Gershoni
biological phenomena as well. (World Scientific, Singapore 1998)
SONG Electron ratchet effect in semiconductor devices and artificial materials with broken centrosymmetry 235

25 A. Löfgren, I. Shorobalko, P. Omling, A.M. Song: to be published 38 D.R.S. Cumming, J.H. Davies: Appl. Phys. Lett. 69, 3363 (1996)
26 R. Landauer: IBM J. Res. Dev. 1, 223 (1957); ibid.: Philos. Mag. 21, 863 39 A. Messica, A. Soibel, U. Meirav, A. Stern, H. Shtrikman, V. Umansky,
(1970) D. Mahalu: Phys. Rev. Lett. 78, 705 (1997)
27 M. Büttiker: Phys. Rev. Lett. 57, 1761 (1986); ibid.: IBM J. Res. Dev. 32, 40 K. Hieke, M. Ulfward: Phys. Rev. B 62, 16 727 (2000)
317 (1988) 41 P.F. Bagwell, T.P. Orlando: Phys. Rev. B 40, 1456 (1989)
28 M.L. Roukes, A. Scherer, S.J. Allen Jr., H.G. Craighead, R.M. Ruthen, 42 L.I. Glazman, A.V. Khaetskii: Europhys. Lett. 9, 263 (1989)
E.D. Beebe, J.P. Harbison: Phys. Rev. Lett. 59, 3011 (1987) 43 O. Heinonen, M.D. Johnson: Phys. Rev. Lett. 71, 1447 (1993);
29 G. Timp, H.U. Baranger, P. deVegvar, J.E. Cunningham, R.E. Howard, M.D. Johnson, O. Heinonen: Phys. Rev. B 51, 14 421 (1995)
R. Behringer, P.M. Mankiewich: Phys. Rev. Lett. 60, 2081 (1988) 44 M. Büttiker: J. Phys. Condens. Matter 5, 9361 (1993)
30 H. van Houten, C.W.J. Beenakker, P.H.M. van Loosdrecht, T.J. Thorn- 45 S. Komiyama, H. Hirai: Phys. Rev. B 54, 2067 (1996)
ton, H. Ahmed, M. Pepper, C.T. Foxon, J.J. Harris: Phys. Rev. B 37, 8534 46 S. Ulreich, W. Zwerger: Superlattices Microstruct. 23, 719 (1998)
(1988) 47 H.Q. Xu: Appl. Phys. Lett. 78, 2064 (2001)
31 C.J.B. Ford, S. Washburn, M. Büttiker, C.M. Knoedler, J.M. Hong: Phys. 48 C.W.J. Beenakker, H. van Houten: Phys. Rev. Lett. 63, 1857 (1989)
Rev. Lett. 62, 2724 (1989) 49 A.M. Song, S. Manus, M. Streibl, A. Lorke, J.P. Kotthaus, W. Wegschei-
32 C.W.J. Beenakker, H. van Houten: Phys. Rev. B 39, 10 445 (1989) der, M. Bichler: Superlattices Microstruct. 25, 269 (1999)
33 L.W. Molenkamp, A.A.M. Staring, C.W.J. Beenakker, R. Eppenga, 50 A.M. Song, A. Lorke, J.P. Kotthaus, W. Wegscheider, M. Bichler: Super-
C.E. Timmering, J.G. Williamson, C.J.P.M. Harmans, C.T. Foxon: Phys. lattices Microstruct. 25, 149 (1999)
Rev. B 41, 1274 (1990) 51 A.M. Song, P. Omling, L. Samuelson, W. Seifert, I. Shorubalko, H. Zi-
34 J. Spector, H.L. Stormer, K.W. Baldwin, L.N. Pfeiffer, K.W. West: Appl. rath: Jpn. J. Appl. Phys. 40, L909 (2001)
Phys. Lett. 56, 1290 (1990) 52 P. Ramvall, N. Carlsson, P. Omling, L. Samuelson, W. Seifert, M. Stolze,
35 A. Palevski, C.P. Umbach, M. Heiblum: Appl. Phys. Lett. 55, 1421 (1989) Q. Wang: Appl. Phys. Lett. 68, 1111 (1996)
36 R. Taboryski, A.K. Geim, M. Persson, P.E. Lindelof: Phys. Rev. B 49, 53 Y. Hirayama, S. Tarucha: Appl. Phys. Lett. 63, 2366 (1996)
7813 (1994) 54 I. Shorubalko, A. Löfgren, P. Omling, A.M. Song: to be published
37 R.I. Hornsey, A.M. Marsh, J.R.A. Cleaver, H. Ahmed: Phys. Rev. B 51, 55 A.M. Song, P. Omling, L. Samuelson, W. Seifert, I. Shorubalko, H. Zi-
7010 (1995) rath: Appl. Phys. Lett. 79, 1357 (2001)