Professional Documents
Culture Documents
Alessandro Braggio
Dipartimento di Fisica, LAMIA-INFM-CNR, Università di Genova, Via Dodecaneso 33, I-16146
Genova, Italy
Matteo Merlo
Dipartimento di Fisica, INFN, LAMIA-INFM-CNR, Università di Genova, Via Dodecaneso 33, I-16146
Genova, Italy
Nicodemo Magnoli
Dipartimento di Fisica, INFN, Università di Genova, Via Dodecaneso 33, I-16146 Genova, Italy
Maura Sassetti
Dipartimento di Fisica, LAMIA-INFM-CNR, Università di Genova, Via Dodecaneso 33, I-16146
Genova, Italy
Current statistics of an antidot in the fractional quantum Hall regime is studied for Laughlin’s
series. The chiral Luttinger liquid picture of edge states with a renormalized interaction
exponent g is adopted. Universal sub-poissonian transport regimes are identified through an
analysis of current cumulants in the sequential tunneling regime. A comparison between noise
and skewness is proposed in order to clearly distinguish the charge of the carriers, regardless
of possible non-universal interaction renormalizations. Super-poissonian statistics is obtained
in the shot limit for g < 1, and plasmonic effects due to the finite-size antidot are tracked.
1 Introduction
The peculiar properties of quasiparticles (QPs) in the fractional quantum Hall effect (FQHE)
have received great attention especially for the states at filling factor ν = 1/p, p odd integer, in
which gapped bulk excitations were predicted to exist and to possess fractional charge e∗ = νe
(e < 0 electron charge) and statistics.1 A boundary restriction of this theory was subsequently
put forward in terms of edge states by Wen.2 This theory recovered the fractional numbers of QPs
in the framework of chiral Luttinger Liquids (χLL), and indicated tunneling as an accessible tool
to probe them.3 Accordingly, QPs with charge e/p were measured in shot noise experiments with
point contact geometries and edge-edge backscattering.4 A key prediction of χLL theories is that
the interaction parameter should be universal and equal to ν. As a consequence, the quasiparticle
(electron) local tunneling density of states obeys a power law in energy D(E) ∝ E α with α = ν−1
(α = 1/ν − 1). Several geometries have been set up in experiments to see this power laws
through measurements of tunneling current I or conductance G.5 The observations confirm the
power-law behaviour but a discrepancy with theory is found in the values of the exponent. This
disagreements of χLL predictions with observed exponents are still not completely understood,
although several theoretical mechanisms have been put forward to reproduce a renormalized
Luttinger parameter, including coupling to phonons or dissipative environments,6,7 effects of
interaction range,8 edge reconstruction with smooth confining potentials.9
Our purpose is to discuss fractional Hall edges in an enriched χLL theory where the pos-
sibility of a renormalized interaction parameter g #= ν is assumed a priori, analysing different
transport regimes and clearly distinguishing signatures of charge e∗ from effects due to the QP
propagators governed by g. To do so, we choose a geometry where two fractional quantum
Hall edges, at fractional filling factor ν = 1/p (p odd integer), are connected via weak tunnel-
ing through an edge state encircling an antidot as in Fig. 1a.10,11,12,13 We compare Fano factor
and skewness for not poissonian transport regimes. We describe transport regions where the
Fano factor is sensitive to the power laws of the QP propagators and presents super-poissonian
correlations.
(AB) coupling with the external magnetic field is described by H AB and the tunneling between
the circular antidot and the edges is given by HR/L T . The standard Wen’s hydrodynamical
where ρl (x) aisthe electron excess density and v is the edge magnetoplasmon velocity (! = 1).
The theory is bosonized with the prescription ρl (x) = ∂x φl (x)/2π where φl (x) = φ0l (x) + φpl (x)
with the scalar fields comprising both a charged φ0l (x) and a neutral sector φpl (x).2 Here we choose
for convenience to set all fields with the same chirality (right movers) and we consider the lateral
right/left leads infinitely extended. The neutral edge modes for the antidot are defined on a
ring of length L with periodic boundary conditions.10,11,12 The antidot charged zero mode reads
φ0l (x) = 2πL νnx − χ with n the excess number of antidot QPs and χ the Hermitian conjugate
"∞
operator of n. The antidot Hamiltonian, due to finite size, reduces to HA = Ec n + s=1 s (a †s as
0 2
where Ec = πνv/L is the topological charge excitation energy and ( = 2πv/ L is the plasmonic
excitation quantized energy of the neutral sector. This Hamiltonian is expected to describe
the antidot anywhere on a ν = 1/p plateau due to the FQHE incompressibility. Nevertheless,
encircling a finite area, it is sensitive to the actual position in the plateau through a coupling to
the Aharonov-Bohm (AB) flux.10 By a gauge transformation, it is easy to see that this amounts
simply to a shift of n in HA 0 according to E n2 → E (n − Φ/Φ )2 , where Φ = hc/|e| is the
c c 0 0
flux quantum. The canonical commutation relation [χ, n] = i together with [φ0j (x), φpl (x# )] = 0
10
ensure that the field on each edge satisfies [φl (x), φl (x# )] = iπνsgn(x−x# ). Charge ν QP fields for
the l-leads are now defined through exponentiation ψl (x) ∝ eiφl (x) . The commutation relation
guarantees that the fields ψl create fractional charge excitations with fractional statistics.2
Each χLL supports several excitations other than the single QP, given in general by ψlm (x) ∝
exp[imφl (x)].3 The electron corresponds to m = 1/ν, while the single QP fields ψl (x) are obtained
by setting m = 1. One should therefore consider all possibilities for tunneling, i.e. all terms like
t(m) ψAm † (x)ψ m (x), j = L, R. Renormalization group (RG) flow equations have been set up for
j
the antidot geometry by Geller and Loss. 10 One can show that the single-qp tunneling amplitude
dominates at intermediate temperature due to antidot finite size. 12,13 This appears to be the
case in most experimental observations where single QP tunneling is clearly observed.14 We
a) ϕ b) c)
1
L 0
xL tL n+3
k2 0.75
A n+2 I III V ν
II IV 0.50
xR k3
tR
R
0 n+1
ν2
En
−
0.25
E+
n
n
0 2 4 6 8 e∗ V /Ec 0.47 0.50 0.53 ϕ
Figure 1: a) Geometry of the system: in white, the Hall fluid; in black the depleted areas defining tunneling
points (dashed lines); solid lines define the chiral edge states left (L), right (R) and the antidot (A). b) Scheme
of transport regions in the (V, ϕ) plane: roman numbers indicate the number of charge states involved in the
n
transport; thin lines signal the onset of transitions, where energies E± = 0. c) Current cumulants as a function
∗
of ϕ in the two state regime (II) for e V = 0.1Ec , kB T = 0.004Ec : Fano factor k2 /ν (top lines) and skewness
k3 /ν 2 (bottom lines) where ν = (e∗ /e); comparison between different values of g: g = 1/5 (solid line), g = 1/3
(dot-dashed line), g = 1/2 (dotted line). Horizontal thin lines indicate the universal limits at 1/2 at 1/4.
†
therefore only retain the dominant term HjT = v(tj ψA (xj )ψj (0) + h.c., j = L, R, that represents
the single-QP tunneling between the infinite edges and the antidot.
A finite source-drain voltage V is applied between the left and right edges, producing a backscat-
tered tunneling current I of QPs through the antidot. In the DC limit the sequential tunneling
current is obtained through the solution of a standard Master Equation with the states labeled
by the number of fractional QPs in the antidot.10 In this approximation the transport depends
only on the incoherent tunneling rates ΓL/R (E). Their expression is well known within the
Luttinger description of edge " states with fully relaxed plasmonic excitations of the antidot,15
Γj (E) = |tj |2 Γ(E) = |tj |2 l wl γ(E − l() where E is the energy associated to the QP tunneling
event and γ(x) = (βωc /2π)1−g |Γ(g/2+iβx/2π)|2 eβx/2 with Γ(x) the Euler Gamma function and
β = 1/kB T . Note that in the standard χLL theory g = ν. Here, we will assume g = νF where
F describes possible renormalization effects due to coupling of the infinite edges with additional
modes (e.g. phonons, or to edge reconstruction). Its the explicit value will depend on the de-
tails of interaction.6,9 Renormalizations of the tunneling matrix elements ti → t̃i are in principle
present as well, but do not affect the results as we will discuss. Note finally that the fractional
charge e∗ is solely determined by ν.a We estimated the temperature regimes where our sequen-
tial tunneling picture holds and we found them consistent with experimental possibilities.13 The
sequential tunneling rates, in the Master Equation, depend only on the energy change of single
QP jumps (transitions n → n + 1, n + 1 → n for the number of antidot QPs). Processes in the
forward direction define the energies E± n = e∗ V /2 ± 2E (ϕ − n − 1/2), with ϕ = Φ/Φ + e∗ V /2E .
c 0 c
The conditions E± n = 0 grid the parameter space (V, ϕ) into diamonds, see Fig. 1(b), where the
transport is dominated by fixed number of QP states, indicated in Fig. 1b) with roman numbers.
a
It is thus separated from the dynamical behaviour governed by g.
3 Results
where Fano factor and normalized skewness correspond to k2,3 respectively. From Eq. 2 we have
that the normalized current cumulant are expressed as a product of two contributions: one com-
ing from the charge of the carrier e∗ and the other from the statistics of the transport process,
espressed in terms of particle number irreducible cumulants. Our task will be to find conditions
where %%N &&p /%N & assumes universal values, independently from any non-universal parameter
(e.g. the tunnelling asymmetry η = |tR |2 /|tL |2 ). We underline the fact that such universality
must hold independently of the Luttinger parameter g, since renormalization processes are al-
most invariably present and not controllable. We define therefore special the conditions, in the
parameter space, where the statistics of the particle transport is fixed and %%N &&p /%N & takes uni-
versal values. In the antidot geometry, different conditions can be found where this measurement
is special in the sense discussed above. According to Esq. (2), kp are the natural observables
to look at in order to measure the QP charge e∗ if the conditions are special. Hereafter we will
mainly concentrate on noise and skewness because they are more easily experimentally accessi-
ble. We remind that for a poissonian process, as the transport of weak backscattering current
in a point contact,b k2 = e∗ /e = ν, and k3 = (e∗ /e)2 = ν 2 .
# ∗$# # ∗ $ $
e βe V Γ0+ Γ0− f− (e∗ V )
k2 = coth − 2η , (3)
e 2 Γ2tot
% n&
with Γn± = Γ E± , Γtot = Γ0+ f+ (E+0 ) + ηΓ0 f (E 0 ) and f (x) = 1 ± e−βx . For the skewness we
− + − ±
find # ∗ $2 # $
Γ0+ Γ0− f+ (e∗ V ) 02 02 2 ∗
e 2 Γ+ Γ− f− (e V )
k3 = 1 − 6η 12η . (4)
e Γ2tot Γ4tot
We analyze now the behaviours (3) and (4) in different regimes looking for special conditions.
Thermal limit: e∗ V ( kB T . The Fano factor is independent of the charge fractionalization,
k2 = 2kB T /eV , reflecting the fluctuation-dissipation theorem. On the contrary, the normalized
skewness that measures the fluctuations time-reversal asymmetry induced by the current depends
b
This is the case of fractional charge measurements in noise experiments.4
I II III IV V VI
k2 2
ν
1
0
0 2 4 6 8 e∗V /Ec
Figure 2: Fano factor vs V for g = 1/3 (bold lines), 1 (thin lines); barrier asymmetry η = 1 (solid lines), 10
(dashed lines). Parameter: ϕ = −0.25, kB T = 0.01Ec and ν = 1/3.
on the carrier charge ratio (e∗ /e) = ν. Indeed, for low voltages V → 0+ one has k3 = (e∗ /e)2 [1 −
3η Sech2 (βEc (ϕ − 1/2))/(1 + η)2 ]. Note that the ϕ dependence can be used to extract the charge
ratio (e∗ /e) and η.
Shot limit: kB T ( e∗ V . In the blockade regions I with |βE± 0 | * 1, one has k = (e∗ /e)
2
and k3 = (e /e) . In this case the statistics of the transport process is poissonian: the trans-
∗ 2
port through the antidot is almost completely suppressed, %I& ≈ 0, and the residual current is
generated only by a thermally activated tunneling that is completely uncorrelated. So regions I
constitute an example of special regime. We consider now the two-state regime (II) for βE± 0 * 1.
For fractional edges g < 1, k2,3 have a particular functional dependence on ϕ (see Fig. 1c ). We
find that they both develop a minimum and that for not too strong asymmetries the absolute
values of the minima are knmin = (e∗ /e)n−1 /2n−1 . These minimal values do not depend neither
on g, as the comparison of solid (g = 1/5), dotted (g = 1/3) and dashed (g = 1/2) curves in
Fig. 1(c) confirms,
' nor on η. 12 For Fermi liquid
( edges g = 1, we have k2 = (e∗ /e)(1+ η 2 )/(1+ η)2
and k3 = (e∗ /e)2 1 − 6η(1 + η 2 )/(1 + η)4 , independently from ϕ. Here, k2 and k3 assume their
minimal values (e∗ /e)/2 and (e∗ /e)2 /4 in the symmetric case η = 1. In this conditions we have
the strongest anticorrelation that is signalled by a marked sub-poissonian statistics. We can con-
clude that in the two-state regime, in the shot limit, the values of the minima for k2,3 obtained
varying η, ϕ correspond to a special condition where the system shows the same universal sub-
poissonian statistics for any g ≤ 1. This represents a means of testing fractional charge outside
poissonian conditions and insensitive to renormalizations of the Luttinger parameter.
dashed line). An investigation of this effect demonstrates that the rate enhancement due to the
onset of the plasmonic collective excitations is responsible of the super-poissonian behaviour at
small asymmetries.
Acknowledgments
Financial support by the EU via Contract No. MCRTN-CT2003-504574 and by the Italian
MIUR via PRIN05 is gratefully acknowledged.
References
1. R. B. Laughlin, Phys. Rev. Lett. 50, 1395 (1983); B. I. Halperin, ibid. 52, 1583 (1984).
2. X. G. Wen, Phys. Rev. Lett. 64, 2206 (1990); Phys. Rev. B 43, 11025 (1991).
3. C. L. Kane and M. P. A. Fisher, Phys. Rev. B 46, 15233 (1992).
4. R. de-Picciotto et al Nature (London) 389, 162 (1997).
5. A. M. Chang et al, Phys. Rev. Lett. 77, 2538 (1996); M. Grayson et al, ibid. 80, 1062
(1998); S. Roddaro et al, ibid. 93, 046801 (2004).
6. B. Rosenow and B. I. Halperin, Phys. Rev. Lett. 88, 096404 (2002); O. Heinonen and S.
Eggert, ibid. 77, 358 (1996).
7. S. Khlebnikov, Phys. Rev. B 73, 045331 (2006); L. S. Levitov et al, ibid. 64, 075322
(2003).
8. E. Papa and A. H. MacDonald, Phys. Rev. Lett. 93, 126801 (2004); Phis. Rev. B 72,
045324 (2005); X. Wan et al, Phys. Rev. Lett. 94, 166804 (2005).
9. K. Yang, Phys. Rev. Lett. 91, 036802 (2003); X. Wan et al, Phys. Rev. B 68, 125307
(2003); Y. N. Joglekar et al, ibid. 68, 035332 (2003).
10. M. R. Geller and D. Loss, Phys. Rev. B 56, 9692 (1997); M. R. Geller and D. Loss, ibid.
62, 16298(R) (2000).
11. S. Kettemann, Phys. Rev. B 55, 2512 (1997).
12. A. Braggio et al, Phys. Rev. B 74, 041304(R) (2006).
13. M. Merlo et al, condmat/0701438 Phys. Rev. B in press.
14. V. J. Goldman et al, ibid. 64, 085319 (2001).
15. A. Braggio et al, Europhys. Lett. 50, 236 (2000).
16. L. S. Levitov and M. Reznikov, Phys. Rev. B 70, 115305 (2004).
17. N. Byers and C. N. Yang, Phys. Rev. Lett. 7, 46 (1961).
18. D. A. Bagrets and Yu. V. Nazarov, Phys. Rev. B 67, 085316 (2003).
19. Y. M. Blanter and M. Büttiker, Phys. Rep. 336, 1 (2000).
20. A. Braggio et al, Phys. Rev. B 67, 233308 (2003).