scattering resonances in 1d coherent transport through a correlated quantum dot: an application of...
TRANSCRIPT
J Comput Electron (2006) 5:247–250
DOI 10.1007/s10825-006-8852-z
Scattering resonances in 1D coherent transport through acorrelated quantum dot: An application of the few-particlequantum transmitting boundary methodAndrea Bertoni · Guido Goldoni
C© Springer Science + Business Media, LLC 2006
Abstract We present a method for the calculation of the
scattering states of a N + 1th-particle coherently interact-
ing with N correlated particles confined in a nanostructure
and placed within an open domain. The method is based on a
generalization of the quantum transmitting boundary method[C. Lent and D. Kirkner, J. App. Phys. 67, 6353 (1990)].
The antisymmetry conditions of the N + 1-identical parti-
cles current-carrying state results from a proper choice of
the boundary conditions. As an example which is relevant to
coherent electronics, we apply the method to compute the ex-
act transmission functions and phases of an electron crossing
a 1D quantum dot with zero, one or two bound electrons.
Keywords Scattering states · Fano resonances · Quantum
dot · Open boundaries · 1D coherent transport
Introduction
The process of quantum scattering of a charge by a system of
bound particles has attracted much attention since the early
days of quantum mechanics. Inelastic scattering of electrons
by atoms served as a validation for theoretical approaches
beyond the single-particle approximation [1, 2]. The above
methods, especially those based on phase-shift analysis of
scattered wavefunctions, were also applied to the study of
carrier scattering by a localized impurity in bulk metals and
A. Bertoni (�)INFM-S3 National Research Center on nanoStructuresand bioSystems at Surfaces,Via Campi 213/A, 41100 Modena, Italye-mail: [email protected]
G. GoldoniINFM-S3 National Research Center on nanoStructuresand bioSystems at Surfaces, Modena, Italy; Dipartimento diFisica, Universita di Modena e Reggio Emilia, Italy
semiconductors, thanks to the similarities of the impurity
states to atomic orbitals.
An electrostatically-defined semiconductor quantum dot
(QD) is an ideal laboratory to test and quantify quantum
coherence and correlation effects among carriers in scatter-
ing experiments, with a single electron transmitted through
or reflected by electrostatically charged QDs [3]. Indeed,
QDs posses many desirable features from this point of
view: an atomic-like structure that can be fully controlled
by electrostatically tuning the confining potential; tunable
coupling to source and drain leads; precise determination
of conductance down to the quantized transmission limit;
tunable number of electrons within the same sample; and last
but not least, transmission phases can be directly measured
by experimental setups exploiting quantum-interference. In
this context, Yacoby et al. [4] and Schuster et al. [5] recently
measured the transmission phase of an electron scattered
by a QD. This stimulated a novel interest on transmission
phases; indeed, not only these studies have a fundamental
interest, but may directly play a role in the next-generation
of quantum nanoelectronic devices and, possibly, of future
quantum computing architectures [6].
In this work we generalize the quantum transmittingboundary method (QTBM) [7], which will be briefly re-
viewed below, to the solution of the few-particle spinless
Schrodinger equation. The new method, which we termed
few-particle QTBM (FP-QTBM), allows us to include proper
open boundary conditions and simulate the scattering of one
electron by a few charges confined in a QD. Below we dis-
cuss its application to a 1D double-barrier resonant-tunneling
(DBRT) device, with an electron incoming from one lead
and a number of other electrons occupying a bound state of
the central dot (see Fig. 1). We found that, in spite of the
single dimensionality of the structure, the coupling between
the continuum of the single-particle scattering states and the
Springer
248 J Comput Electron (2006) 5:247–250
Fig. 1 Potential profile V (x) of the simulated system (see Eq. (1)).Four bound single-particle levels χm are represented. The QD electronis initialized in the ground state and the scattered electron, arriving fromthe left lead, is given a kinetic energy of 10 meV. Two 50 meV tunnelbarriers mimic the quantum point contacts that connect the QD to theleads. The bottom potential of the QD is varied in the simulations
discrete spectrum of the bound states, gives rise to Fano reso-
nances, where the transmission phase shows an abrupt jump
of π at the zeros of the transmission coefficient.
Few-particle quantum transmitting boundarymethod and scattering resonances
The mechanism of charge transport through a QD operating
in single-electron regime (i.e., when only the scattered elec-
tron is localized in the dot) is well understood: Breit-Wigner
resonances are found in the transmission spectra, stemming
from the coupling of a quasibound state to the scattering states
in the leads. Such transmission resonances show a Lorentzian
line shape of the amplitude t = C(�/2)/(E − Eqb + i�/2),
where C is a complex constant, � the width of the resonance,
inversely proportional to the life time of the quasibound state
with energy Eqb, and E is the energy of the incoming particle.
Below we briefly discuss the FP-QTBM which allows us to
investigate scattering experiments beyond this single-particle
picture.
Let us consider the envelope function of two spinless iden-
tical particles in 1D, ψ(x1, x2), describing the stationary state
with one electron in a QD bound state (we take here the
ground state without loss of generality) χ0(x), with energy
E0, and a second electron incoming from the left boundary
(Fig. 1) as a plane wave of energy (E − E0), where E is
the total energy of the two-particle system. If one supposes
that a change in the energy of the quasibound state does not
alter significantly its life time, t can be equivalently seen
as a function of E at a fixed QD potential or as a function
of Eqb at a fixed incoming-electron energy. We chose this
second perspective considering a fixed Fermi level in the
leads and a variable QD potential, and we suppose that the
applied source-drain bias is sufficiently small that either zero
or one quasibound state lies inside the transport window. The
Hamiltonian of the system reads
H = − h2
2m
(∂2
∂x1
+ ∂2
∂x2
)+V (x1)+V (x2)+ e2
4πε|x1 − x2| ,
with m, e and ε effective mass, elementary charge and dielec-
tric constant, respectively and V (x) the given single-particle
potential (see Fig. 1). We aim at solving Hψ(x1, x2) =Eψ(x1, x2) in the square domain x1, x2 ∈ [0, L]. Due to
the Pauli exclusion principle the two-particle wave function
must be antisymmetric ψ(x1, x2) = −ψ(x2, x1). Therefore
the computational domain can be reduced to the black re-
gion in Fig. 2. Along the diagonal boundary, the condition
ψ(x, x) = 0 is imposed by the wave function antisymmetry
(this would also apply to charged bosons due to the diverging
Coulomb energy). To include the other two boundaries we
follow the QTBM. Owing to our choice of boundary condi-
tions (one particle coming from the left lead as a plane wave
and the other in the ground state of the QD) the total energy
E belongs to a continuous region of the energy spectrum.
Furthermore we limit ourself to energies E that are lower
than the double ionization energy, i.e., the incoming electron
does not posses enough energy to unbind the other particle
without being captured in the QD. This condition means that
the probability to find both electrons in the leads is zero, i.e.
ψ(x1, x2) = 0 when both x1 and x2 are outside [0, L] (regions
F in Fig. 2). Following the QTBM and taking into account
Fig. 2 Computational domain and boundaries. The Schrodinger equa-tion is numerically solved in the black triangular region. Due to thechoice of the system parameters, the wave function is zero in fourforbidden regions marked by F that represent the possibility of bothelectrons being in the leads. The two-particle wave function is incidentand reflected through boundary A while it is transmitted through bound-ary B. On the diagonal boundary, i.e. when x1 = x2, ψ(x1, x2) = 0 dueantisymmetrization. In the calculation performed we took L = 50 nm
Springer
J Comput Electron (2006) 5:247–250 249
the antisymmetry of the wave function, the form of the wave
function in the two leads A ≡ (0 < x1 < L , x2 < 0) and B ≡(x1 > L , 0 < x2 < L) is
ψ A(x1, x2) = −[χ0(x1)eik A
0 x2 +MA∑
m=0
bAmχm(x1)e−ik A
m x2
+∞∑
m=MA+1
bAmχm(x1)ek A
m x2
](1a)
ψ B(x1, x2) =[
MB∑m=0
bBmχm(x2)e−ik B
m x1
+∞∑
m=MB+1
bBmχm(x2)ek B
m x1
], (1b)
respectively, where χm(x) is the m-th bound state, with en-
ergy Em , of the QD; k Am =
√(2m/h2)|E − Em − V (0)| and
k Bm =
√(2m/h2)|E − Em − V (L)|; bA
m and bBm are the m-th
channel reflection and transmission coefficients, respectively
(and are unknowns of the problem); MA and MB represent
the number of bound QD states that can be excited when
the incoming electron is either reflected back in lead A or
transmitted to lead B. Equations (1a) represent the general
form of a two-particle wave-function of energy E when one
particle is on a constant-potential lead and the other is bound
by the confining potential. Note that also the higher energy
channels must be included in the calculation as evanescent
waves, in perfect analogy with the QTBM, where the χm’s
represent the transverse modes of the leads.
The boundary Eqs. (1a) and (1b) are coupled to the
Schrodinger equation for the internal points and discretized
by finite difference method. The resulting linear system is
solved by means of the parallel direct sparse solver library
Pardiso [8, 9]. Explicit calculations are performed using
GaAs material parameters. Below we show results in a en-
ergy range for which the bound electron is left in the ground
state after the scattering, i.e., MA = MB = 0. In particular,
we consider a 30 nm potential well of variable negative depth,
with two 50 meV barriers, each 3 nm wide. The simulation
domain is L = 50 nm and the energy of the injected electron
is 10 meV. The leads are at zero potential. Once the linear sys-
tem has been solved, the transmission probability and phase
are easily obtained from the bB0 complex coefficients.
A generalization of the procedure described above can
be derived in order to calculate the scattering states of an
impinging electron when N electrons are bound in the QD.
In this case the first step is the computation of the correlated
N -electron eigenstates of the QD χm(x1, x2, ..., xN ); then
the (N + 1)-electron Schrodinger equation is solved together
with (N + 1) open boundary conditions, similar to Eq. (1a),
in a (N + 1)-dimensional domain [10].
Results and conclusions
Figure 3 shows the transmission probability (solid line) and
the transmission phase (dotted line) for one (A), two (B)
and three (C) particles, as a function of the QD potential.
Single-particle Breit-Wigner resonances are present in the
three cases (see, as an example, Fig. 3(D)): they appear as
large Lorentzian peaks centered around the eigenenergies
of the closed system. As the number of electrons bound in
the QD increases, the above resonances shift toward stronger
confining energies: this is due to the Coulomb blockade, orig-
inated by the larger negative charge of the carriers bound in
the QD. Note that in our approach the Coulomb blockade
effect is automatically obtained without the need of an elec-
trostatic model. The transmission phase changes smoothly
by π on each Breit-Wigner resonance, as contemplated by the
Fig. 3 Transmission spectra for one electron (A), two electrons (B)and three electrons (C). The transmission probability (solid curve) andthe transmission phase (dotted curve) are reported as a function of thebottom potential of the QD. While in the single-particle case only Breit-Wigner resonances are present, in the other cases, with electron-electroncorrelation, also Fano resonances appear. Lower graphs show two par-ticular resonances, of different types (D: Breit-Wigner, E: Fano) of thethree-electron spectrum
Springer
250 J Comput Electron (2006) 5:247–250
single-particle approach. In addition to the above resonances,
a number of extremely sharp peaks with typical asymmetric
Fano lineshape (see, as an example, Fig. 3(E)) is present in
the transmission spectra of the two- and three-particle scat-
tering states. The transmission phase increases smoothly by
π on the resonance peak but shows an abrupt drop of π in
correspondence of the zero of the transmission probability.
This behavior of the phase, already demonstrated in the liter-
ature for single-channel single-particle transport through 2D
QDs [11,12], and experimentally measured in single-electron
transistors [13], is not obtainable for 1D single-particle sys-
tems: Fano resonance in 1D system are a genuine few-particle
phenomenon brought about by electron-electron correlation.
In conclusion, we have developed a generalization of the
QTBM which allowed us to simulate the scattering of one
electron through a nanostructure with confined carriers cou-
pled to the leads with full inclusion of the Coulomb corre-
lation. Application to scattering from a charged 1D QD has
been shown. As a final remark we note that related experi-
ments on charge transport through QDs [14,15] demonstrated
also that the transmitted current include a coherent compo-
nent also in the case of a multi-electron occupancy of the dot.
With the few-particle QTBM it should be possible to quantify
this coherent part [16], since it corresponds to the probabil-
ity that the electron is transmitted in the same channel from
which is injected.
Acknowledgments This work has been partially supported by projectsMIUR-FIRB n.RBAU01ZEML, MIUR-COFIN n.2003020984, INFMCalcolo Parallelo 2004, and MAE, Dir.Gen. Promozione CooperazioneCulturale.
References
1. Goldberger, M.L., Watson, K.M.: Collision Theory. Courier DoverPublications (2004)
2. Joachain, C.J.: Quantum Collision Theory. North-Holland Publish-ing, Amsterdam (1975)
3. Beenakker, C.W.J., van Houten, H.: J. Solid State Phys. 44, 1 (1991)4. Yacoby, A., Heiblum, M., Mahalu, D., Shtrikman, H.: Phys. Rev.
Lett. 74, 4047 (1995)5. Schuster, R., Buks, E., Heiblum, M., Mahalu, D., Umansky, V.,
Shtrikman, H.: Nature 385, 417 (1997)6. lloyd, S.: Science 261, 1569 (1993)7. Lent, C., Kirkner, D.: J. App. Phys. 97, 6353 (1990)8. Schenk, O., Gartner, K.: J. Future Generation Comp. Syst. 20, 475
(2004)9. Schenk, O., Gartner, K., Fichtner, W.: BIT 40, 158 (2000)
10. Bertoni, A., Goldoni, G., Molinari, E.: in preparation (2006)11. Lee, H.-W.: Phys. Rev. Lett. 82, 2358 (1999)12. Taniguchi, T., Buttiker, M.: Phys. Rev. B 60, 13814 (1999)13. Gores, J., Goldhaber-Gordon, D., Heemeyer, S., Kastner, M.A.,
Shtrikman, H., Mahalu, D., Meirav, U.: Phys. Rev. B 62, 2188(2000)
14. Aikawa, H., Kobayashi, K., Sano, A., Katsumoto, S., Iye, Y.: Phys.Rev. Lett. 92, 176802 (2004)
15. Konig, J., Gefen, Y.: Phys. Rev. Lett. 86, 3855 (2001)16. Bertoni, A.: J. Comp. Elect. 2, 291 (2003)
Springer