Arxiv:Quant-Ph/0009026 V3 22 Feb 2001 E,Orshm Losfradrc Eto Elsieult in Inequality Bell’S of Shall Test Direct We a As for Allows Scheme Our Wires)

Home , QIP (complexity)

View metadata,citationandsimilarpapersatcore.ac.uk arXiv:quant-ph/0009026 v3 22 Feb 2001 e,orshm losfradrc eto elsieult in inequality Bell’s of shall test direct we a As for allows scheme our wires). see, (quantum waveguides semicon- quantum coupled measure ductor in and electrons ballistic manipulate using generate, states to entangled how show shall we specifi-cally, More physics. condensed-matter in violation equality implementations. experimental no are few there A date r solid-state been in [10]. states have measured entangled physics of also generation the been for has a proposals (GHZ) recently state only [9]; entangled ions trapped [8] photons and using prepared been two- have Experimentally, states entangled algorithms. particle quantum swap- many entanglement cryp- of quantum and coding, of ping heart dense the teleportation, at tography, are [7] states (GHZ) Greenberger-Horne- three-particle Zeilinger Einstein-Podolsky- and [6] pairs entanglement. (EPR) is Rosen QIP in speed-up tional [5]. characterization and fabrication nanostructure single-electron in in as progress recent the from synergistically benefit decoherence, can strong QIP relatively of implementation the solid-state of a spite in contrast, technol- In does microelectronics ogy. existing which of hardware, scalability the quantum allow cannot such not gener- QIP with however of realized is application easily It large-scale be any atomic [3]. that in optics believed quantum ally originated in devices and [2] QIP ex- physics only of the rates, realizations decoherence low perimental of need Indeed, the QIP. to for due fur- systems mainly imposes candidate and on constraints scenario strong ideal ther above de- the de- Thus, modifies to coherence superpositions. tends quantum environment between coherence its stroy with system quantum realistic quantum- ( precise detection perform ( manipulations to quantum able coherent synthesis, be state should one end, this To aiyrltdt h blt fprocessing of ability the to related pri- marily are capabilities communication/computation to novel able mechanics. devices Such quantum by QIP provided realize power stimu- additional to the has exploit how intellec- this about informa- thinking unquestionable hand, of new other to lated concepts the hand, on basic theory; understanding one tion/computation in the progress on tual led, has [1] nti etrw rps neprmn ots elsin- Bell’s test to experiment an propose we Letter this In computa- for ingredient key the mentioned, already As h nrdcino unu nomto rcsig(QIP) Processing Information Quantum of introduction The measurement ASnmes 36.z 53.,0.7L,85.30.V 03.67.Lx, 85.30.S, 03.65.Bz, numbers: PACS oevr t ai nrdet a lyarl oad ag-cl unu-nomto rcsigi solid- in processing quantum-information large-scale towards Bell’s role test a to devices. play state and states may entangled ingredients prepare Coulomb-c basic to its in required Moreover, are electrons gates pr five ballistic the (only simplicity using inequality), its states to Due entangled wires. detect quantum and manipulate generate, epooea xeiett etBl’ inequa Bell’s test to experiment an propose We 2 nttt o cetfi Interc Scientific for Institute 3 iatmnod iia oienc iTrn,CroDc el buz 4 02 oio Italy Torino, 10129 24, Abruzzi degli Duca Corso Torino, di Politecnico Fisica, di Dipartimento cnl u owr [11 forward put ecently e tn elsIeult ihBlitcEetosi Semiconductors in Electrons Ballistic with Inequality Bell’s Testing .Teuaodbeitrcino any of interaction unavoidable The ). psdsmcnutrbsdshm a eipeetdwt urnl vial technology. available currently with implemented be can scheme semiconductor-based oposed auIonicioiu, Radu 1 htncthree-particle photonic siuoNazi Istituto nage states entangled hsc 4 swell as [4] physics ag II,V (ISI), hange -1] u pto up but ]-[18], 1,2 gating nl e aFsc el aei (INFM) Materia della Fisica la per onale al Zanardi, Paolo laGaio il etmoSvr 5 I- 65, Severo Settimio Viale Gualino, illa )and [1]. iyvoaini c in violation lity i ¼ inr uis(..eeto pn nqatmdt)ti snot is this dots) quantum sta- in spins of electron case transfer (e.g. the qubits to In can tionary need qubits. we the mobile qubits without into another, stationary interconvert flying to place using one from qubits: entanglement our of character h oetQReeto ubn wt eta iei energy kinetic central (with subband electron QWR lowest de- the within well wave-packet quasi-monoenergetic is a wire the of localized terms along motion in is free scribed wave-function its its (ii) and (i) QWR that that: on mean we wire given a (the one other the in tron the E j optto ihblitcelectrons ballistic for with proposal computation earlier the of hardware quantum ductor gating simple relatively identified. a is end, only) gates this (five sequence To system. solid-state a w-iesse.Mr rcsl,w hl s h so-called the use shall we this dual-rail precisely, across More electron the system. of two-wire state quantum-mechanical the to ing oa,wihhsbe eetyaaye n aiae thr validated Bertoni and by simulations analyzed numerical pro- recently this been of has features which main posal, the following the in summarize We daetQRsrcue.The structures. QWR adjacent QIP. any for prerequisite regime basic coherent the fully is Such experiments [21],[22]. interferometric electrons ballistic of number with a for arena natural beenthe has and [5] nanotechnology semiconductor existing compatible with fully is so-called regime coherent-transport the This preserved. in therefore, are microns; electrons regime few tic scale a nanometric of the values on reach can length co- their herence conditions ex- such may In temperature mobility. low high extremely at perience subband good QWR the lowest the to within as interf well semiconductor/semiconductor of as quality splittings relatively energy the intersubband to large Due one- quasi waveguides. electron as (1D) behave the dimensional QWRs of state-of-the-art fabrication view current [5], by In technology reached confinement (QWRs). carrier nanometric wires quantum semiconductor in ,2 1, napaigfaueo h rpsdshm stemobile the is scheme proposed the of feature appealing An h rpsdeprmna eu sbsdo h semicon- the on based is setup experimental proposed The h uligboko u unu adaei arof pair a is hardware quantum our of block building The as electrons ballistic use to is idea main The n eta wave-vector central and ytepeec fteeeto noeo h ie (called wires the of one in electron the of presence the by 0 ri)adtebssstate basis the and -rail) n asoRossi Fausto and nesdmte hsc.W hwhwto how show We physics. ondensed-matter ersnainfrteqbt edfietebssstate basis the define we qubit: the for representation n h hs oeec fterwv ucin is functions wave their of coherence phase the and Ô ,,3 1,2, = 1 h ri) aigta h lcrni in is electron the that Saying -rail). k 03 oio Italy Torino, 10133 uldsemiconductor oupled i ½ E= j £ ui tt sdefined is state qubit Ñ ytepeec fteelec- the of presence the by ¾ tal. et yIonicioiu by [20]. provided byCERNDocumentServer ). brought toyouby cs electrons aces, yn qubits flying tal. et quantum accord- ballis- ough [19]. CORE 2

easily done. body Coulomb interaction. In contrast, if at least one qubit is

¼i Any QIP device can be built using only single- and two- in the j state, then nothing happens.

qubit gates [23]. We choose the following set of universal The proposed quantum hardware has some advantages.

½ Firstly, the QIP device needs not to be “programmed” at the

À = À; È ;È

quantum gates: ³ ,where is

½ ½ hardware level (by burning off the gates), as it may appear.

i³ Programming is done by switching on/off the gates and this

È = ´½;e µ

a Hadamard gate, ³ diag is a single-qubit phase

C way any quantum algorithm can be implemented [28]. Sec-

shift, and È is a controlled sign ﬂip. We shall use the more

i³

C ondly, we use cold programming, i.e., we set all the gates be-

= ´½; ½; ½;e µ

general two-qubit gate È diag . ³ We now brieﬂy describe the physical implementation of the fore “launching” the electrons, so we do not need ultrafast (i.e. universal quantum gates in terms of the previously introduced subdecoherent) electronics for gate operations. This property dual-rail representation. The Hadamard gate can be imple- is essential and is a distinct advantage of the proposed quan- mented using an electronic beam-splitter, also called waveg- tum architecture over other solid-state proposals [29]. There- uide coupler [24, 25, 26]. The idea is to design the two- fore, the gating sequence needed for the proposed experiment wire system in such a way to spatially control the inter-wire can be pre-programmed using static electric ﬁelds only. electron tunneling. For a given inter-wire distance, a proper One important requirement of our quantum hardware is that modulation (along the QWR direction) of the inter-wire po- electrons within different wires need to be synchronized at tential barrier can produce a linear superposition of the ba- all times in order to properly perform two-qubit gating (the

two electron wave-packets should reach simultaneously the

¼i j½i sis states j and . More speciﬁcally, let us consider a

Coulomb-couplingwindow). It is thus essential to have highly Ä

coupling window, i.e. a tunneling-active region, of length c

monoenergetic electrons launched simultaneously. This can

=¾= characterized by an inter-wire tunneling rate ! .As it propagates, the electron wave-packet oscillates back and be accomplished by properly tailored energy ﬁlters and synchronized single-electron injectors at the preparation stage. forth between the two waveguides with frequency ! .Let

£ We now turn to the proposed experimental setup for testing

=hk =Ñ Ú be the group velocity of the electron wave-packet

Bell’s inequality. Two-particle entangled states (Bell states)

¼i along the wire; then, the state j goes into the superposition

Ä can be generated using three Hadamard gates and a controlled-

cÓ× «j¼i · ×iÒ «j½i « = !Ø =¾ Ä

with .Let Ø be the length necessary for the complete transfer ofÚ the electron from one sign shift (see dashed box in Fig. 1; the controlled sign shift

plus the lower two Hadamards form a CNOT gate). Consider

« = ; Ä = Ú =¾

wire to the other, Ø . For a transfer length

´a; bµ=

the correlation function for two (pseudo)spins È

Ä = Ä =¾ Ø

c the device is equivalent to a beam-splitter and

´½µ ´¾µ

h i = a

i i

hence, up to a phase shift, to a Hadamard gate. By a proper (here, a is the pseudo-spin projection b

modulation of the inter-wire potential barrier we can vary the along the unit vector a) [30]. Any local, realistic hidden- « tunneling rate ! and therefore the rotation angle .Asare- variable theory obeys the Bell-CHSH [31, 32] inequality:

sult, this structure can operate as a NOT gate by adjusting the

¼ ¼ ¼ ¼

jÈ á; bµ· È á ; bµ· È á ; b µ È á; b µj ¾

= Ä

Ä (1) Ø inter-wire potential barrier such that c ( -rotation). Similarly, the gate can be turned off by an appropriate po-

tential barrier for which the electron wave-packet undergoes This inequality is violated in quantum mechanics. For the sin-

a full oscillation period, returning back to its original state glet j , a standard calculation gives the result

Ä =¾Ä ¾ À Ø

( c , -rotation). Another way of turning the gate

´¾µ

´½µ

È ´a; bµ h© j © i = a:b

j (2)

off is to suppress inter-wire tunneling by applying a strong a b

potential bias to the coupled QWR structure.

¼ ¼ ¼ ¼

a:b = b:a = a :b = b :a = ¾=¾ È

The phase shifter ³ can be implemented using either a Choosing , we obtain

¾ ¾

potential step (with height smaller than the electron energy ¾ , violating thus Bell inequality (1).

´a; bµ ) or a potential well along the wire direction; the well Let us now focus on the correlation function È .In is preferred since the phase-shift induced is more stable un- the EPR-Bohm gedankenexperiment we need to measure the

der voltage ﬂuctuations. In order to have no reﬂection from spin component of one particle along a direction Ò.How-

the potential barrier, the width Ä of the barrier should be an ever, in our setup this is not possible directly, since we can

integer multiple of the half wavelength of the electron in the measure only Þ , i.e., whether the electron is in the 0-or

= Ò=¾;Ò¾ ÁÆ

step/well region, Ä . in the 1-rail. The solution is to do a unitary transformation

j i!j i = Í j i

We ﬁnally describe the two-qubit gate. In our scheme the , such that the operator Ò is diagonalized

¼ ¼

h j j i = h j j i

Ò Þ

controlled phase shifter is implemented using a Coulomb to Þ , . We are looking for a uni-

Í Í Í = Ò

coupler [27]. This quantum gate exploits the Coulomb inter- tary transformation which satisﬁes Þ , with

= ´×iÒ cÓ× ³; ×iÒ ×iÒ ³; cÓ× µ

action between two single electrons in different QWR pairs Ò a unit vector. In terms of

Í ´; ³µ=ÀÈ ÀÈ

³ =¾ (representing the two qubits). The gate is similar in construc- our elementary gates we obtain . tion to the beam splitter previously introduced. In this case the Thus, measuring the spin (in the EPR-Bohm setup) along

multi-wire structure (see Fig. 4) needs to be tailored in such a a direction Ò is equivalent to performing the unitary trans-

Í ´; ³µ

way (i) to obtain a signiﬁcant Coulomb coupling between the formation followed by a measurement of Þ . Going

two 1-rails only and (ii) to prevent any single-particle inter- back to our entangled pair, we now apply on each qubit a local

j½i Í ´ ;³ µ Í ´ ;³ µ

½ ¾ ¾

wire tunneling. Therefore, only if both qubits are in the transformation ½ and , respectively. Here,

a =´ ;³ µ b =´ ;³ µ

½ ¾ ¾ state they both experience a phase shift induced by the two- ½ and are the two directions dis- 3

Bell states

π −π/2−ϕ1 −θ1 H H H 1 0.8 0.6 0.4 HH H H 0.2 −π/2−ϕ −θ 0 2 2 ±0.2 ±0.4 ±0.6 ±0.8 FIG. 1: Quantum network for the measurement of Bell’s inequality. ±1 0 0

Bell states are prepared in the dashed boxed; then the ﬁrst qubit is 1 1

=´ ;³ µ

a 2 2 ½ measured along the direction ½ and the second qubit

3 3

b =´ ;³ µ ¾ along the direction ¾ . α 4 4 θ 5 5 6 6 π

|1> H

´ µ Ë ´«; µ

FIG. 3: Correlation functions Ë and for the “ideal” and

´; µ= Ë ´ µ “realistic” singlet, respectively; note that Ë . |1> HH

j© i

perfect singlet « we obtain

« «

´¾µ

j© i

FIG. 2: Testing Bell’s inequality for the singlet state .The ´½µ

Ë ´«; µ h© j j© i = ×iÒ ×iÒ · «

« (4)

³ = ³ =

¾ ¾ ¾

quantum network is obtained from Fig.1 by setting ½

=¾ =¼ = ¾

, ½ and relabelling .

= « = ½

with ¾ .For we recover the correlation

´ µ Ë ´; µ= cÓ×

function of the singlet, Ë .Thetwo

´ µ

functions are plotted in Fig.3; Ë can be identiﬁed by noting

cussed above; at the very end, we measure Þ (i.e., electron in

that there is no « dependence.

0-or1-rail; see Fig.1). È

Experimentally, since the one-qubit gate is easier to con-

© i For the singlet j the correlation function depends only trol, we can measure the coupling « of the Coulomb coupler on the scalar product of the two directions (2), and hence only C È by measuring the dependence of the correlation function

on the angle between them. Without loss of generality we «

´«; µ

Ë on the phase shift (which can be accurately deter-

³ = ³ = =¾; =¼ =

¾ ½

can choose ½ and relabel mined). This procedure can be used to determine the purity of

À =½Ð ¾ .Since , the gating sequence simpliﬁes to only the singlet, and hence to test/calibrate the Coulomb coupler. ﬁve gates, as shown in Fig. 2. With this simple network we We are now interested to see how small the coupling « can can measure the correlation function (2) which violates Bell’s be in order to still have a violation of Bell’s inequality. The inequality (1). To perform an Aspect-type experiment [8], we question we ask is: For what values of « the correlation func-

have to choose independently the directions of measurement

´«; µ tion Ë in eq.(4) violates Bell’s inequality in (1)? To this

for each qubit after the electrons are entangled. In this case end, we have found a numerical solution: the inequality (1) is

È ÀÈ À

we need three more gates (after the gate) on the

¾ ´=¾; ¿=¾µ violated for « . upper qubit in Fig. 2. A schematic representation of the proposed experimental In practice the situation is more complex. The essential setup for measuring Bell’s inequality violation is presented in

ingredient for producing entanglement is the controlled-sign Fig.4. It is possible to reduce the number of gates on the ½-

¼ i

È ´e ; ½µ

shift gate which involves an interaction between the two ¼

rail by using a phase shifter on the -rail diag

qubits. Experimentally this requires a good timing of the two ½

È ´½;e µ

instead of the ½-rail one used so far diag ,since

½ ¼

electrons (they should reach simultaneously the two-qubit gat- i

È e È

the two are equivalent (up to an overall phase) = .

C ing region). Suppose that instead of having an ideal È gate In our setup there are two different ways of producing a

preparing an ideal singlet (dashed box in Fig. 1), in practice È

phase shift : (i) electrically, with a potential applied on top

C «

we realize a È gate (possible with unknown phase ). In this «

of the ¼-rail (the quantum well described above); (ii) mag-

following state: a magnetic ﬁeld on the area between the lower two beam-

È È

splitters (this can be done since the and gates com-

« j½i´j¼i j½iµ

i«=¾

mute). The second method has the advantage of avoiding the

« (3) ¾ ¾ no-reﬂection condition for the potential well (the length of the gate should be a half integer multiple of the electron wave- which is a superposition of the singlet and of a separable state. length). Either way can be used experimentally.

Let us now consider the experimental setup discussed We stress that Aharonov-Bohm rings and quantum in-

a =´¼; ×iÒ ; cÓ× µ b =´¼; ×iÒ ; cÓ× µ

½ ¾ ¾ above, with ½ and , terference experiments with ballistic electrons are standard

both in the ÇÝÞ plane. For the correlation function of the im- tools in mesoscopic physics. A two-slit experiment with an 4

BS Aharonov-Bohm ring having a quantum dot embedded in one 0- SEP SET arm has been reported in [21],[22]. This experiment is similar qubit a to the layout of the lower qubit in Fig.4, but the authors do 1- SEP SET not use beam splitters and Coulomb couplers. In the experi- C C mental setup presented here the more difﬁcult part will be to 1- SEP SET implement the Coulomb coupler (CC) and to perform the ex- qubit b B periment at the single electron level. In our case preparation SEP SET 0- and measurement of the states are done using single electron BS BS V pumps (SEPs) and single electron transistors (SETs) [33], re- State State spectively. preparation GATING measurement In conclusion, we have proposed the ﬁrst measurement of Bell’s inequality violation in coupled semiconductor nanos-

FIG. 4: Experimental setup to test the Bell-CHSH inequality; the ¼- tructures using ballistic electrons. Due to the relative sim-

rails of each qubit are dashed for clarity. A potential Î applied on plicity of the proposed experimental setup (only ﬁve gates are

top of the ¼-rail (dashed box in the ﬁgure) is used to produce a phase needed to produce entanglement and to test Bell’s inequality)

shift on the second qubit; alternatively, the same effect can be this measurement scheme is potentially feasible in terms of ~ achieved with a magnetic ﬁeld B (via the Aharonov-Bohm effect). current semiconductor nanotechnology. R.I. is grateful to G. Amaratunga, F. Udrea and A. Popescu for stimulating discussions and ﬁnancial support.

[1] D.P. DiVincenzo and C. Bennett, Nature 404, 247 (2000); Driven Quantum Dots, quant-ph/9911123. A. Steane, Rep. Proc. Phys. 61, 117 (1998). [18] C. Barnes, J. Shilton and A. Robinson, Quantum computa- [2] J.I. Cirac and P. Zoller, Phys. Rev. Lett. 74, 4091 (1995); tion using electrons trapped by surface acoustic waves, cond- A. Sorensen and K. Molmer, Phys. Rev. Lett. 82, 1971 (1999). mat/0006037. [3] Q.A. Turchette et al., Phys.Rev.Lett.75, 4710 (1995); [19] R. Ionicioiu, G. Amaratunga and F. Udrea, Ballistic Single- A. Imamoglu et al.,Phys.Rev.Lett.83, 4204 (1999). Electron Quputer, quant-ph/9907043. [4] K. Likharev, Proc. IEEE 87, 606 (1999). [20] A. Bertoni et al., Phys. Rev. Lett. 84, 5912 (2000). [5] L. Pfeiffer et al., Microelectronics J. 28, 817 (1997). [21] A. Yacoby et al.,Phys.Rev.Lett.74, 4047 (1995). [6] A. Einstein, B. Podolsky and N. Rosen, Phys. Rev. 47, 777 [22] R. Schuster et al.,Nature385, 417 (1997). (1935). [23] A. Barenco et al.,Phys.Rev.A52, 3457 (1995). [7] D.M. Greenberger, M.A. Horne, A. Shimony and A. Zeilinger, [24] J.A. del Alamo and C.C. Eugster, Appl. Phys. Lett. 56,78 Am. J. Phys. 58, 1131 (1990). (1990). [8] A. Aspect, J. Dalibard, and G. Roger, Phys. Rev. Lett. 49, 1804 [25] N. Tsukada, A.D. Wieck and K. Ploog, Appl. Phys. Lett. 56, (1982). 2527 (1990). [9] C. Monroe et al., Science 272, 1131 (1996); Proc. R. Soc. Lon- [26] M.J. Liang, G.G. Siu and K.S. Chan , J.Phys.D. 27, 1513 don A454, 411 (1998). (1994).

[10] D. Bouwmeester et al.,Phys.Rev.Lett.82, 1345 (1999). [27] M. Kitagawa and M. Ueda, Phys. Rev. Lett. 67, 1852 (1991). È

[11] D. DiVincenzo and D. Loss, Quantum Computers and Quantum [28] The phase shifter ³ can be switch on/off by simply switching

=½Ð È

Coherence, cond-mat/9901137. on/off the electric bias applied to the gate. Since ¾ ,this [12] G. Burkard, D. Loss and E.V. Sukhorukov, Noise of Entangled gate can also be turned on/off by means of external biases and Electrons: Bunching and Antibunching, cond-mat/9906071. appropriate design. [13] D. Loss and E.V. Sukhorukov, Probing Entanglement and Non- [29] Any quantum hardware using “stationary qubits” needs ultra- locality of Electrons in a Double-Dot via Transport and Noise, fast electronics/optical pulses to perform (sub-decoherent) gate cond-mat/9907129. operations on picosecond time-scale. [14] D. DiVincenzo et al., Quantum Computation and Spin Elec- [30] We stress that we are dealing with charge degrees of freedom tronics, cond-mat/9911245. and the present pseudo-spin notation has nothing to do with the [15] D. DiVincenzo, The Physical Implementation of Quantum physical spin of the electrons. Computation, quant-ph/0002077. [31] J.S. Bell, Physics 1, 195 (1964). [16] L. Quiroga and N.F. Johnson, Phys. Rev. Lett. 83, 2270 (1999), [32] J.F. Clauser, M.A. Horne, A. Shimony and R.A. Holt, Phys. cond-mat/9901201. Rev. Lett. 23, 880 (1969). [17] J.H. Reina, L. Quiroga and N.F. Johnson, Quantum Entangle- [33] A. Shnirman and G. Sch¨on, Phys. Rev. B 57, 15400 (1998). ment and Information Processing via Excitons in Optically-