You are on page 1of 12

Signatures of multiple Majorana zero-modes in 1D nanowires

Arnab Barman Ray,1 Jay D. Sau,2 and Ipsita Mandal3


1
Institute of Optics, University of Rochester, Rochester, 14623, USA
2
Department of Physics, Condensed Matter Theory Center and Joint Quantum Institute,
University of Maryland, College Park, MD 2074
3
Laboratory ofAtomic And Solid State Physics, Cornell University, Ithaca, NY 14853, USA
(Dated: September 23, 2018)
abstract
2

CONTENTS

I. Introduction 2

II. Differential conductance with normal leads 3


A. Quasi-1D Rashba nanowire 3
B. Quasi-1D p-wave topological superconductors 4

III. Josephson effect for a multichain Kitaev ring coupled to a microwave cavity cavity 8
A. Analytical expression for cavity-response 9
B. Numerical results for cavity-response 10

IV. Discussion and concluding remarks 11

V. Acknowledgement 11

References 11

I. INTRODUCTION

It is well-known that mutiples MZMs can occur at one end either due to chiral protection in BDI class,1,2 or as
Kramers pairs for DIII systems. The analytical expression for tunneling conductance for a superconducting wire
2
hosting multiple MZMs is given by the zero-bias peak value of Gn (0) = 2he |Q|, for |Q| = N, N − 1, where Q is the
topological quantum number of the phase.3 For the BDI symmetry class, Q ∈ (0, ±1, ±2, . . . ± N ). For |Q| < N − 1,
a Q-dependent upper and lower bound on Gn (0) exists. For a system which can support at most one MZM, due
to perfect Andreev reflection at the MZM, the zero-temperature, zero-voltage limit of the differential conductance,
2
measured by a normal tip, is quantized to Gn (0) = 2he in the topological phase. The same scenario, but with a
superconducting probe, presence of an MZM should be signalled by symmetric differential conductance peaks at
2
eV = ±∆ with a universal magnitude Gsc (±∆) = (4 − π) 2he , due to resonant Andreev reflection from the MZM,
for a wide range of tunneling strengths and temperatures. This result was derived analytically by Peng et al. in
Ref.4 in the tunneling limit (i.e. in the limit of small junction transparency). Since the effect of thermal broadening
in a superconductor is exponentially suppressed by the superconducting gap, a conventional superconducting lead
as the probe in an MZM tunneling experiment might be more promising. The spatial profile of MZMs using an
STM tip, which can be either normal or superconducting, was numerically investigated by Chevallier et al. in Ref..5
Their system could host at most one MZM at one end of the wire. In this work, we consider systems which can
host multiple MZMs at each end. The differential conductance, detected by the normal or superconducting leads,
should depend on the number of MZMs existing in a particular topological phase, and as we change the parameters
of the Hamiltonian to scan among different phases, there should be a corresponding change in the signals.
Other proposed detection techniques of MZMs include a superconducting ring interrupted by a weak link and
pierced by a magnetic flux, while being coupled inductively to a microwave cavity. Such a set-up should exhibit
a 4π (2π) variation of the cavity response with respect to the external magnetic flux in the topological (trivial)
phase, and related such dependence to the fractional (normal) Josephson effect. As opposed to previous works this
set-up, the treatment of the 1D Kitaev chain in Ref. 6 takes into account the low-energy MZMs, the gapped bulk
states, and the interplay between these states in the presence of the cavity on equal footing. The advantage of such
a description is that it allows one to analyze the topological transition, where the crossover from the fractional to
the normal Josephson effect takes place.
In the first part of this paper in Sec. II, we numerically compute the differential conductance for the normal
and superconducting leads for different temperatures, when the system under probe is a superconducting quasi-1D
(Q1D) nanowire able to host multiple MZMs at each end. In the second part of the paper in Sec. III, we apply the
treatment of Ref. 6 to the ladder Kitaev model of Ref. 7. This simple multichain generalization of the 1D Kitaev
model gives a system with topological phases with multiple MZMs at each end. Hence, it allows us to analyze the
cavity responses for multiple topological transitions.
3

II. DIFFERENTIAL CONDUCTANCE WITH NORMAL LEADS

In this section, we analyze the behaviour of differential conductance that can be detected using normal leads.
Our numerical results agree with the analytical expressions derived in Ref. 3. The generic system we treat here is
shown in Fig. 1. The number of sites in each chain is Nx , while the number of chains is Ny .

FIG. 1: The cartoon depicts the generic system treated in this section. The tunelling barrier is fixed at τ = 0.1 tx ,
where tx is the hopping along the chain direction. The leads, as well the transverse and longitudinal hopping
terms, are quasi-1D in nature.

A. Quasi-1D Rashba nanowire

We consider a one-dimensional Rashba nanowire aligned along the x-axis brought into contact with an s-wave
superconductor in presence of an external magnetic field of strength Vx applied in the x-direction. This can be used
to construct a toy model for a quasi-one-dimensional (Q1D) system, where we stack an array of these 1D chains
coupled by a weak hopping amplitude, as shown below:

Ny
Nx X NX Ny n
x −1 X o
† †
X
H= ψi,j (−µ τz + ∆s τx + Vx σx ) ψi,j + ψi+1,j (−tx − i α σy ) τz ψi,j + h.c.
i=1 j=1 i=1 j=1
y −1 n
Nx NX o

X
− ψi,j+1 ty τz ψi,j+1 + h.c. , (2.1)
i=1 j=1

ψi,j =(c†i,j,↑ , c†i,j,↓ , ci,j,↓ , −ci,j,↑ ) . (2.2)

where σm and τm (m = (x, y, z)) are Pauli matrices which act respectively in the spin and particle-hole spaces,
µ is the chemical potential, α the spin-orbit coupling, ∆s is the magnitude of the s-wave superconducting gap.
Furthermore, tx is the hopping strength between neighboring sites on the same chain, while ty is the hopping
strength between sites in neighboring chains ( ty  tx ).The site-indices (i, j) label the fermions in (x, y) strip, such

that i ∈ [1, Nx ] and j ∈ [1, Ny ]. The fermion creation operator is denoted by the row matrix ψi,j . We will now
discuss the results for differential conductance for this system, when we use normal leads.
In Fig. 2(a), we have shown how the values of conductance reflect the actual phase diagram of the system for
Nx = 100, Ny = 3, ∆s = tx , Vx = 2 tx , α = 0.25 tx , ty = 0.3 tx as a function of µ (in units of tx ). The zero-
2
temperature conductance, in units of eh is exactly equal to the number of zero modes in the system. From the
phase diagram, we can see the presence of Andreev Bound States(ABS) at certain parts of the spectrum. These,
however, do not significantly affect the zero-bias differential conductance. With a rise in temperature, we see that
conductance values are lower than the saturation values. We also find that the differential conductance dies off
quickly at nonzero values of voltage. Fig. 2(b).
4

(a) (b)

FIG. 2: For quasi-1D Rashba nanowire Eq. (2.1), with parameters Nx = 100, Ny = 3, ∆s = tx , Vx = 2 tx ,
α = 0.25 tx , ty = 0.3 tx : (a) correspondence between the spectrum and the zero bias differential conductance at
the first site (energy and µ in units of tx ), and (b) differential conductance at the first site as a function of voltage
(in units of tx ) for µ = 1.2 tx , at different temperatures.

B. Quasi-1D p-wave topological superconductors

A TR-invariant topological superconductor8 can be realized by a spin-triplet, equal spin pairing (ESP), p-wave
superconductivity. These properties are conjectured to be present in the Q1D transition metal oxide Lithium
molybdenum purple bronze Li0.9 Mo6 O17 (LiMO) and some organic superconductors.9–14 The hopping integrals
along the crystallographic directions of these materials vary as, tx  ty  tz , making them Q1D conductors.
Due to the anisotropic electrical conductivity, LiMO may be modeled as an array of parallel one dimensional
systems coupled by weak transverse hopping.15 Furthermore, such weakly coupled array of parallel one dimensional
topological superfluids can also be realized in cold fermion systems.16
We consider the one dimensional spin-triplet topological superconductor by a lattice Hamiltonian including
nearest-neighbor hopping (tx ), on-site chemical potential (µ) and a generic p-wave superconducting order parameter,
given by:
X h i  
− tx c†i+1,σ ci,σ − µ c†i,σ ci,σ αR c†i+1,σ ci,σ0 (i σy )σσ0 + h.c. ∆σσ0 c†i+1,σ c†i,σ0 + h.c. . (2.3)
X
H 1D =

+
i,σ,σ 0 i,σ,σ 0

Here i ∈ [1, Nx ] is the lattice index, σ =↑, ↓ represents the spin index, and αR is the strength of spin-orbt (SO)
coupling which is taken along the y-axis. In the momentum space, the superconducting gap function is given by:
∆αβ (k) = hcα (k)cβ (−k)i, where cα (k) is the annihilation operator of a single electron with spin α and momentum
k along x-direction. The spin symmetry of Cooper pairing can be either singlet (S = 0) or triplet (S = 1), such
that generically ∆αβ (k) = ([∆s (k) + d(k) · σ])αβ , with the singlet component ∆s (k) = ∆s (−k), and the triplet
component represented by the vector d satisfying d(k) = −d(−k). Throughout this study, we choose d(k) =
d(k) (0, 0, 1) along ẑ. The superconducting term in real space is then of the form: i ∆(c†i+1,↑ c†i,↑ + c†i+1,↓ c†i,↓ ) + h.c.
- which can be easily seen to be TR-invariant. Thus we get a TR-invariant superconductor containing the ESP
spin-triplet p-wave state proposed to be realized in LiMO, which belongs to the DIII class.
In fact for αR = 0, the model has a chiral as well as a mirror symmetry both of which allow an integer (Z)
invariant. This chiral symmetry explains the persistence of the zero modes to TR-breaking terms including stray
Zeeman fields, perpendicular to the d(k)-vector. However, V k d(k) breaks the chiral symmetry and the overlapping
MZMs hybridize into ordinary gapped states. These Zeeman terms can be captured by the Hamiltonian:

HZ = V · σ τ0 . (2.4)
5

Furthermore, a generic SO coupling term aligned in an arbitrary direction in spin space can be written as:
HSO = αR sin k (a · σ) τz . (2.5)
One can show that although a k d(k) preserves chiral symmetry, an SO term in the plane perpendicular to d(k)
breaks chiral symmetry.
In the Nambu basis, defined with Ψk = (ck,↑ , ck,↓ , c†−k,↓ , −c†−k,↑ )T , the total bulk Hamiltonian now takes the
form:
X †
H 1D = Ψk Hk1D Ψk , Hk1D = [ (k) − µ ] σ0 τz + ∆ sin k σz τx + HZ + HSO . (2.6)
k

Here k = kx is the 1D crystal-momentum, (k) = −2 tx cos k is the single-particle kinetic energy, ∆ sin k is the
p-wave order parameter. For simplicity, we will consider a pure triplet order parameter with ∆s (k) = 0.
Now, a realistic Q1D spin triplet superconductor such as LiMO (or Q1D TRI systems in cold fermions) may be
modeled as an array of 1D chains coupled by a weak hopping amplitude ty  tx . A truly 3D system can be realised
by stacking 2D arrays and coupling them through a third hopping integral tz  ty  tx . We take Ny parallel
chains, indexed by j ∈ [1, Ny ], coupled only by transverse hopping ty . The Q1D Hamiltonian is a generalization of
Eq. (2.6) given by:
X † y y
H Q1D = Ψk,j (Hk1D δj,j 0 + Hj,j 0 )Ψk,j 0 , Hj,j 0 = −ty σ0 τz (δj,j 0 +1 + δj,j 0 −1 ) . (2.7)
k j j0

where Ψk,j = (ck,j,↑ , ck,j,↓ , c†−k,j,↓ , −c†−k,j,↑ )T .

(a) (b)

FIG. 3: For quasi-1D p-wave topological superconductors, with parameters Nx = 100, Ny = 3, ∆ = tx , ty = 0.3 tx ,
αR = 0: (a) correspondence between the spectrum and the zero bias differential conductance(at the first site) at
different temperatures(energy and µ in units of tx ), where the third panel shows the sum of the invariants
associated with the mirror symmetry operator iσz τ0 ; (b) variation with applied voltage(in eV) at µ = 0.

In the absence of Rashba spin-orbit coupling and Zeeman fields, this model has a chiral as well as a mirror
symmetry (just like the one-chain case), both of which allow an integer (Z) invariant, which means Majorana
Kramers Pairs (MKPs) exist at the edges. For numerical simulations, we take Ny = 3. In Fig. 3, we see the
existence of MKPs for different values of µ, the maximum number of which is Ny at each end of the system. In
fact, the Q1D Hamiltonian can be decomposed
 into 3 separate chains each of which satisfies the mirror symmetry,
ˆ
meaning, [M, Hk ] = 0, where M = i d(k) · σ τ0 = i σz τ0 . In the Fourier space, the Hamiltonain takes the form
can be written as:
X †
H Q1D = Ψk Hk1D Ψk , Hk1D = [ (k) − µ0 ] σ0 τz + ∆ sin k σz τx , (k) = −2 tx cos k , (2.8)
k
6

(a) (b)

FIG. 4: For quasi-1D p-wave topological superconductors, with parameters Nx = 100, Ny = 3, ∆ = tx , ty = 0.3 tx ,
αR = 0.5 tx : (a) correspondence between the spectrum and the zero bias differential conductance at different
temperatures (energy and µ in units of tx ) with a = (1, 1, 0) and Zeeman field strength V = 4 tx , where the third
panel shows the BDI invariant associated with the chiral symmetry operator σz τy ; (b) variation with applied
voltage(in eV) at µ = 3 tx .

(a) (b)

FIG. 5: For quasi-1D p-wave topological superconductors, with parameters Nx = 100, Ny = 3, ∆ = tx , ty = 0.3 tx ,
αR = 0.5 tx correspondence between the spectrum and the zero bias differential conductance at different
temperatures (energy and µ in units of tx ): (a) with a = (1, 0, 0) and V = 0, (b) with a = (1, 1, 0) and V = 4 tx

where Hk1D
q is the effective Hamiltonian for each chain. The three chains have modified chemical potentials of
µ0 = µ ± 2 t2y , µ. The integer invariant associated with the mirror symmetry of each of these chains is given by:

π
d ln {∆ sin(k) + i(−2 tx cos k − µ0 } d ln {−∆ sin(k) + i(−2tx cos (k) − µ0 }
Z  
1
Z= dk − , (2.9)
2iπ −π dk dk

which equals the number of MZMs at each end of the system. This is obtained by block-diagonalizing Hk1D in
the eigenbasis of M and taking the difference in the number of times the complex determinant of the two blocks
encircle the origin in the complex plane.15 The sum of the three invariants is plotted in the third panel of Fig. 3(a).
Fig. 3(a) also shows how the differential conductance at the first site of the chain varies with the spectrum, at
7

(a) (b)

FIG. 6: Normal lead zero-bias differential conductance for quasi-1D p-wave topological superconductors with
broken“generalized” reflection invariance in y-direction, with Nx = 100, Ny = 3, ∆ = tx , ty = 0.3 tx . The
Hamiltonian has three chains having chemical potentials µ, 0, −µ. We have considered (a) αR = 0 (the third
panel shows the mirror invariant); and (b) αR = 0.5 tx , a = (1, 0, 0) . Energy and µ are in units of tx .

different temperatures. Fig. 3(b) shows how the differential conductance varies with applied voltage, and we see
that it quickly dies off at non-zero voltages.
A chiral symmetry, represented by the operator S = (d · σ) τy , explains the persistence of the MZMs in presence
of TR-breaking terms such as Zeeman fields (HZ = V · σ τ0 ) perpendicular to the d-vector, which would break M.
However, V k d breaks the chiral symmetry and the overlapping MZMs hybridize into ordinary gapped states.15
Fig. 4 (a) shows the existence of MZMs in the presence of a Zeeman field (V ) along x̂. The introduction of a
TR-breaking Zeeman field lifts the Kramers degeneracy, but the MZMs still persist, which maximum number is Ny
at each end. At finite temperatures, we find lower conductances for all cases, as is expected. The third panel in
Fig. 4 (a) shows the sum of the three invariants associated with each of the 3 sectors. In this case, each invariant
is simply equal to the number of times the complex determinant of each block of the off-block-diagonalized version
of Hk (in the eigenbasis of S) encirles the origin in the complex plane. This equals the number of MZMs at each
end of the system. Fig. 4 (b) shows how the conductance dies off at non-zero values of voltage.
Fig. 5 shows the case for non-zero Rashba SOI in the events of zero and non-zero magnetic fields. When we
switch on a small Rashba SOI with a = (1, 0, 0) (generating the term αR sin k σx τz in Eq. (2.6)), it does not have
a significant effect on the conductance and the presence of MZMs, as shown in Fig. 5(a). This is expected. The
spectrum looks identical to that in Fig. 3, and hence will give the same conductance. Now, the introduction of
TR-breaking Zeeman field lifts the Kramers degeneracy. However, the Ny MZMs at each end still persist, as seen
in in Fig. 5(b), where we have both a Zeeman field V = 4 tx and a Rashba SOI with a = (1, 1, 0).
We consider a special case with a chain-dependent chemical potential. Considering a 3-chain system with different
chemical potentials of µ, 0 and −µ, we get the plots of Fig. 6, which show that MKPs persist for the case with
Rashba SOI but no Zeeman field. Such a system might be effected with the application of an transverse electric
field. For both the cases of zero and non-zero Rashba SOI, we get multiple MKPs in the system. Following the
treatment of Ref. 15, we can show the decoupling of the system into independent channels with modified chemical
potentials. To see this, we write the Hamiltonian in the chain index basis as:
 
H1 −ty σ0 τz 0
Hchain =  −ty σ0 τ3 H2 −ty σ0 τ3  , (2.10)
0 −ty σ0 τ3 H3

where H1 , H2 and H3 are essentially the same Hamiltonian but with different chemical potentials as described
before. It is to be noted that each of the elements in the matrix above are themselves 4Nx × 4Nx matrices. The
8

Hamiltonian can be rewritten as:


 −µ σ 
0 τz −ty σ0 τz 0
Hchain = H0 + −ty σ0 τz 0 −ty σ0 τz , (2.11)
0 −ty σ0 τz µσ0 τz

where H0 is the common part of the three chains. When µ = 0, this reduces to the original generalized reflection
symmetry described in Ref.
 15. Thus, we realize
√ 2 that after rotating
√ the Hamiltonian
 by unitary transformation
2 2 2
−tym −µm −µ 2 2 2
2tym +µm 2 −tym −µm +µm 2tym +µm
−1 − t2
− t2
the unitary matrix U = 
 √ym ym
√ , where U is constructed using the

µm + 2t2 2
ym +µm µm − 2t2 2
ym +µm
− tµym
m − tym − tym
1 1 1
eigenvectors of the
q second term in Eq. (2.11), it gets decoupled into three class DIII sectors with modified chemical
potentials of (± 2t2y + µ2 , 0). In the matrix, tym ≡ ty σ0 τz and µm ≡ µσ0 τz . In fact, we note that such general
reductions can always be found whenever the matrix additions to a system of single chain class DIII Hamiltonians
are Hermitian and have the same internal structure (for example here, the internal structure of the terms −ty σ0 τz
and −µ σ0 τz are similar).

III. JOSEPHSON EFFECT FOR A MULTICHAIN KITAEV RING COUPLED TO A MICROWAVE


CAVITY CAVITY

The s-wave superconductor is interrupted underneath the weak link in the quantum wire. Current can flow
around the loop only through the semiconductor weak link.17
We consider the Hamiltonian describing a ladder of the Kitaev chains,as shown below:
X † X h † i
tx ci,j ci+1,j + ∆x c†i,j c†i+1,j + ty c†i,j ci,j+1 + ∆y c†i,j c†i,j+1 + h.c. ,

H =−µ ci,j ci,j − (3.1)
i,j i,j

where i and j are the lattice coordinates for the x and y axes. The intrachain transfer integral tx and the intrachain
p-wave superconducting pairing amplitude ∆x are present in the Kitaev model, while we have newly introduced
the effective interchain transfer integral ty and the superconducting pairing amplitude ∆y .
By making a phase transformation of ci,j , we can change the overall phase of the superconducting pairings.
Hence, only the phase difference between ∆x and ∆y is physical and without loss of generality we can set ∆x to be
real, keeping ∆y complex in general. However, to get phases with multiple MZMs, we need the system to be in the
BDI class for which ∆y has to be real. Let us now consider a ring made up of the this Kiatev ladder and threaded
by a cavity flux, with the two ends of the wire coupled with hopping strength t0 .

(a) (b)

FIG. 7: Energy spectrum as a function of chemical potential µ in a 3-ladder ring, with (a) open t0 = 0, and (b)
closed µ = 0, t0 = 0.2 t configurations. The system is in BDI class, with tx = ty ≡ t and ∆x = ∆y = 0.2 t. Energy
and µ are in units of t.

Fig. 7 corresponds to a 3-ladder Kitaev ring, with symmetric tx = ty ≡ t and ∆x = ∆y = 0.1 t. As a function
of the Fermi energy µ, the spectrum of an open ring (t0 = 0) shows a varying number of MZMs between zero and
three (panel a). As we connect the ring (t0 /t = 0.1), these MZMs hybridize into finite energy states, as long as φ
9

(a) (b)

FIG. 8: The same system as in Fig. 7, with its energy spectrum presented at (a) µ = 0, and (b) µ = 0 t , as a
function of phase difference φ, when it is connected into a ring. The junction coupling is t0 = 0.1 t.

is not tuned to the special parity crossing points φpm = π/4 + n π (for integer n). Fig. 8 shows the dependence
with φ at µ = 0 (three MZMs) and µ = 2.2 t (one MZM). Note that two of the three MZMs in the former case are
degenerate due to y → −y symmetry in the system.

A. Analytical expression for cavity-response

First we review the formalism necessary to get the expression for cavity-response, as carried out in Ref. 6.
By a unitary transformation, we can eliminate the DC flux as:
X † hX 2 π i Φ̂ac

† † †
Hel → U Hel U = − µ ci,j ci,j − tx e N Φ0
ci,j ci+1,j + ty ci,j ci,j+1
i,j i,j
X 4 π i Φtot j
 i
− ∆x e N Φ0
c†i,j c†i+1,j + ∆y c†i,j c†i,j+1 + h.c. ,
i,j
2 π i ΦDC 2 π i Φ̂ac
c†N,j c1,j + h.c. .
X
HT → U † HT U = − t0 e Φ0
e N Φ0
(3.2)
j

It is important to note that only the DC flux can be dealt with by this process (eliminates the DC phase from the
pair potential), since a similar procedure with the ac part would be inconsistent as Φ̂ac does not commute with
the photonic Hamiltonian Hph . However, assuming that λ  1, we expand the exponentials containing Φ̂ac upto
second order in λ. With these approximations, the final Hamiltonian can be written as:

(Φ̂ac )2 D
H ≈ Hel + HT + Hph − Φ̂ac I + ,
2
N −1
2πi X Xh † 2 π i ΦDC i
I= tx ci0 ,j ci0 +1,j + 2 i0 ∆x ci0 ,j ci0 +1,j + t0 e Φ0 c†N,j c1,j − h.c. , (3.3)
N Φ0 0 j
i =1
 −1 X h
2 NX
2π 2 π ΦDC i
D= tx c†i0 ,j ci0 +1,j + (2 i0 )2 ∆x ci0 ,j ci0 +1,j + t0 e Φ0 c†N,j c1,j + h.c. . (3.4)
N Φ0
i0 =1 j

The transmission τ of the cavity can be written as follows:


κ X
τ= , Πtot (ωc ) = Π(ωc ) − hDi , hDi = fn hn|D|ni ,
−i (ω − ωc ) + κ + i λ2 Πtot (ωc ) n
X fn − fm
Π(ω) = |hn|I|mi|2 , (3.5)
−(n − m ) + ω + i δ
n6=m
10

(a) (b)

∂Js 2 π i ΦDC
FIG. 9: The diagonal susceptibility ΠJ = ∂Φ DC
as a function of φ ≡ Φ0 for the same system as in Fig. 8,
evaluated at four values of µ with increasing number of MZMs.

where fn (ΦDC ) is the occupation of the electronic energy state n (ΦDC ), and δ is a small real number in order to
account for the dissipation. Using the sum rule (see Ref. 18),
X |hn|I|mi|2 ∂in ∂n
− hn|D|ni = 2 + , in = − , (3.6)
n − m ∂ΦDC ∂ΦDC
m6=n

where in is the current carried by the nth Andreev level of energy n (ΦDC ). The parameter δ is used to account
for the effects of dissipation in a non-ideal setup. Hence, we obtain
X fn − fm X ∂ 2 n X X fn |hn|I|mi|2
Πtot (ω) = |hn|I|mi|2 + fn 2 −
−(n − m ) + ω + iη n
∂ΦDC n
n − m
n6=m m6=n

∂Js X ∂fn ∂n X fn − fm |hn|I|mi|2


= + −ω ≡ ΠJ (ω) + ΠD (ω) + ΠN D (ω) , (3.7)
∂ΦDC n
∂ΦDC ∂ΦDC n − m (n − m ) − ω − iδ
n6=m
P
where Js = fn in is the persistent (or Josephson) current flowing through the ring in the presence of a DC
n
flux ΦDC . There are three contributions to the cavity response: (i) the persistent current constituting the non-
dissipative part (ΠJ (ω)), (ii) the diagonal term involving the decay of the levels (ΠD ), and (iii) the non-diagonal
part arising from the usual Kubo contribution (ΠN D (ω)). In the following, we assume the zero temperature limit
(T → 0), which in turn implies that ΠD (ω) vanishes. The first term is independent of the frequency ω, and thus
any dependence on this parameter will be due to ΠN D (ω).

B. Numerical results for cavity-response

Here, we perform numerical simulations for a 3-chain system with 60 sites on each chain. All the MZMs have a
parity such that the respective electron modes (formed by superposition of one MZM at each end) at zero energy are
unoccupied. This system can be represented as the 3-tuple, (0, 0, 0), where each entry represents the parity of the
occupied MZM. The diagonal and non-diagonal parts of the zero-temperature susceptibility is computed assuming
that the parity for all MZMs is conserved at the level crossings. We confirm the fractional hall effect for MZMs in
Fig. 9, and note the increase of the diagonal susceptibility with the number of MZMs in the system.
The non-diagonal contribution ΠN D can be dissociated into three parts, one that involves transitions between
bulk states: ΠBB , one that involves transitions between the MZMs and the bulk, ΠBM and one that involves
transitions between the multiple Majorana modes themselves, ΠM M . Even with multiple Majorana modes, due to
the requirement of parity conservation under a transition, we see that ΠM M disappear because MZM transitions
that are allowed by parity ca
ot happen owing to the equality of their occupation numbers (f ) at zero temperature. We expect that at non-
zero temperatures this could lead to divergences in their susceptibilities near the level crossings as well at another
point dictated by the ω of the microcavity, which would be manifested as strong absorption. We note that such
divergences might also arise in systems with parities such as (0, 1, 0), where transitions between MZMs of the same
11

(a) (b) (c)

(d) (e) (f)

FIG. 10: The non-diagonal susceptibility at low frequencies, (a), (b) at ω = 0.2∆x , (c), (d) at ω = 0.02∆x , and
(e), (f) at ω = 0.002∆x as a function of φ for the same system as in Fig. 8, evaluated at four values of µ with
increasing number of MZMs: µ/t = −4, −2.7, −1.3, −0.05 for black, red, green and blue. We have taken δ = 0.001.
φ ≡ 2 π iΦΦ0 DC .

parity give non-zero contributions. A more comprehensive study at non-zero temperatures and at different parity
configurations is required to fully justify these claims.
We plot the non-diagonal contribution to the susceptibility in Fig. 11, which shows the real and imaginary parts
of the susceptibilities at three different values of ω. We find, as expected a higher ω makes for a larger contribution
to the overall susceptibility. The magnitude of the non-diagonal contribution at ω = 0.2 ∆x is comparable to that
from the Josephson current. The imaginary part, which arise entirely due to the presence of dissipation(δ) shows
similar behavior as different ω. The real part on the other hand changes significantly.
We plot the variation of the real and imaginary non-diagonal susceptibility at the phase point φ =
pi with the chemical potential µ in Fig. . We find in this case that while the behavior near the 3 → 2 MZM
transition is highly oscillatory, the susceptibility cleanly diverges for the other phase transition points. This is due
an increase in the contribution from the ΠBM part as the gap closes and the bulk states move close to the MZMs.

IV. DISCUSSION AND CONCLUDING REMARKS

V. ACKNOWLEDGEMENT

We thank Pablo San-Jose for for collaboration in the initial stages of the project. We also thank Denis Chevallier
for useful discussions.

1
Y. Niu, S. B. Chung, C.-H. Hsu, I. Mandal, S. Raghu, and S. Chakravarty, Phys. Rev. B 85, 035110 (2012).
2
I. Mandal, Europhysics Letters 110, 67005 (2015); I. Mandal and S. Tewari, Physica E 79, 180 (2016); I. Mandal,
Condensed Matter Physics 19, 33703 (2016).
3
M. Diez, J. P. Dahlhaus, M. Wimmer, and C. W. J. Beenakker, Phys. Rev. B 86, 094501 (2012).
4
Y. Peng, F. Pientka, Y. Vinkler-Aviv, L. I. Glazman, and F. von Oppen, Phys. Rev. Lett. 115, 266804 (2015).
5
D. Chevallier and J. Klinovaja, Phys. Rev. B 94, 035417 (2016).
6
O. Dmytruk, M. Trif, and P. Simon, Phys. Rev. B 94, 115423 (2016).
12

(a) (b) (c)

(d) (e) (f)

FIG. 11: The non-diagonal susceptibility at low frequencies, (a), (b) at ω = 0.2∆x , (c), (d) at ω = 0.02∆x , and
(e), (f) at ω = 0.002∆x as a function of µ for the same system as in Fig. 8, evaluated at φ = π. We have taken
δ = 0.001. φ ≡ 2 π iΦΦ0 DC .

7
R. Wakatsuki, M. Ezawa, and N. Nagaosa, Phys. Rev. B 89, 174514 (2014).
8
E. Dumitrescu and S. Tewari, Phys. Rev. B 88, 220505 (2013).
9
A. G. Lebed, K. Machida, and M. Ozaki, Phys. Rev. B 62, R795 (2000).
10
J.-F. Mercure, A. F. Bangura, X. Xu, N. Wakeham, A. Carrington, P. Walmsley, M. Greenblatt, and N. E. Hussey, Phys.
Rev. Lett. 108, 187003 (2012).
11
A. G. Lebed and O. Sepper, Phys. Rev. B 87, 100511 (2013).
12
I. J. Lee, S. E. Brown, W. G. Clark, M. J. Strouse, M. J. Naughton, W. Kang, and P. M. Chaikin, Phys. Rev. Lett. 88,
017004 (2001).
13
I. J. Lee, D. S. Chow, W. G. Clark, M. J. Strouse, M. J. Naughton, P. M. Chaikin, and S. E. Brown, Phys. Rev. B 68,
092510 (2003).
14
J. Shinagawa, Y. Kurosaki, F. Zhang, C. Parker, S. E. Brown, D. Jérome, J. B. Christensen, and K. Bechgaard, Phys.
Rev. Lett. 98, 147002 (2007).
15
E. Dumitrescu, T. D. Stanescu, and S. Tewari, Phys. Rev. B 91, 121413 (2015).
16
C. Qu, M. Gong, Y. Xu, S. Tewari, and C. Zhang, Phys. Rev. A 92, 023621 (2015).
17
F. Pientka, A. Romito, M. Duckheim, Y. Oreg, and F. von Oppen, New Journal of Physics 15, 025001 (2013),
arXiv:1210.3237 [cond-mat.mes-hall].
18
N. Trivedi and D. A. Browne, Phys. Rev. B 38, 9581 (1988).

You might also like