Quantum teleportation in high dimensions

Yi-Han Luo, Han-Sen Zhong, Manuel Erhard, Xi-Lin Wang, Li-Chao Peng, Mario Krenn,
Xiao Jiang, Li Li, Nai-Le Liu, Chao-Yang Lu, Anton Zeilinger, and Jian-Wei Pan

Hefei National Laboratory for Physical Sciences at Microscale and Department of Modern Physics,
University of Science and Technology of China, Hefei, 230026, China,
CAS Centre for Excellence in Quantum Information and Quantum Physics, Hefei, 230026, China,
Austrian Academy of Sciences, Institute for Quantum Optics and Quantum Information (IQOQI),
Boltzmanngasse 3, A-1090 Vienna, Austria,
Faculty of Physics, University of Vienna, Boltzmanngasse 5, 1090 Vienna, Austria
(Dated 12 May 2019; email: [email protected], [email protected])

Precise measurement or perfect cloning of unknown quantum states is forbidden1 by the

laws of quantum mechanics. Yet, quantum teleportation2 in principle allows for a faithful

and “disembodied” transmission of unknown quantum states between distant quantum

systems using entanglement. There have been numerous experiments on teleportation of

quantum states of single photons3-8, atoms9, trapped ions10-11, defects in solid states12, and

superconducting circuits13. However, all demonstrations to date were limited to a two-

dimensional subspace—so-called qubit—of the quantized multiple levels of the quantum

systems. In general, a quantum particle can naturally possess not only multiple degrees of

freedom, but also, many degrees of freedom can have high quantum number14 beyond the

simplified two-level subspace. Here, making use of multiport beam-splitters and ancillary

single photons, we propose a resource-efficient and extendable scheme for teleportation

of arbitrarily high-dimensional photonic quantum states. We report the first experimental

teleportation of a qutrit, which is equivalent to a spin-1 system. Measurements over a

complete set of 12 states in mutually unbiased bases15 yields a teleportation fidelity of

0.75(1), well above the optimal single-copy qutrit-state-estimation limit of 1/2. The fidelity

also exceeds the limit of 2/3, the maximum possible for explanation through qubits only.

Thus, we strictly prove a genuine three-dimensional, universal, and highly non-classical

1
quantum teleportation. Combining previous methods of teleportation of two-particle

composite states6 and multiple degrees of freedom7, our work provides a complete toolbox

for teleporting a quantum particle intact. We expect that our results will pave the way for

quantum technology applications in high dimensions, since teleportation plays a central

role in quantum repeaters and quantum networks16.

The ability of coherent control of high-dimensional (HD) quantum states is important for

developing advanced quantum technologies. Compared to the conventional two-level systems,

HD states can offer extended possibilities such as both higher capacity and noise resilience in

quantum communications17,18, more efficient quantum simulation19 and computation20, as well

as larger violation of Bell inequality21. Recent years have witnessed an increasing capability to

generate and measure HD entangled states21-26. However, the previous work is predominantly

limited in the coherent control of single particle HD state. A joint projection of two independent

particles with unknown states into maximally entangled HD states, which requires some forms

of controlled interactions and will play a crucial role in the HD teleportation, dense coding, and

quantum computing, is more challenging and remains largely unexplored experimentally.

We start by describing our protocol of HD quantum teleportation. For the sake of simplicity,

here we explain it using the example of a three-level system, where the underlying physics can

be generalized to arbitrary N-level systems, as elaborated in detail in the Methods. Suppose

Alice wishes to teleport to Bob the quantum state

|   a   0 | 0 a  1 |1 a   2 | 2 a [1]

of a single photon, where | 0 , |1 , and | 2 are encoded by three different paths of the photon

(c.f. Fig. 1), their subscripts label the photon, and their coefficients are complex numbers that

fulfil  0  1   2  1 . The quantum resource required for teleporting this state is a HD

2 2 2

entangled state previously shared between Alice and Bob, for example,

2
 | 00  bc  (| 00 bc  |11 bc  | 22 bc ) / 3 . [2]

This is one of the three-dimensional (3D) Bell states which, together with other eight orthogonal

ones (listed in Methods), forms a complete orthonormal basis of the bipartite 3D Hilbert space.

Conceptually within the theoretical framework of Bennett et al.2, the most crucial step for

the HD teleportation is performing a joint measurement of photon a and b, a process called 3D

Bell-state measurement (BSM). With equal probabilities of 1/9, the 3D-BSM projects photon a

and b into one of the nine 3D Bell states randomly. Alice can then broadcast the 3D-BSM result

classically, which allows Bob to accordingly apply a unitary single-particle 3D transform27 to

reconstruct the original quantum states (eqn.1) at his location (Methods). In general, for an N-

level bipartite system, there exists N 2 HD Bell states. An unambiguous HD-BSM poses a new

challenge both theoretically and experimentally.

Recall that in the teleportation of qubits, the four Bell states can be grouped into three

symmetric and one anti-symmetric state under particle exchange, which facilitate the

discrimination using linear optics28. In the 3D case already, however, the situation becomes

fundamentally more complicated. There are three Bell states that are symmetric and the other

six are neither symmetric nor anti-symmetric. In theory, it was shown29 that it is impossible to

discriminate two-photon HD Bell states with linear optics only when the dimensions N  3.

To overcome such a linear optical limitation29, here we utilize N-2 additional single photons,

so-called ancillary photons, and a multiport beam splitter (BS) with N-input-N-output all-to-all

connected ports30, which is a generalisation of the quantum Fourier transform. A detailed

derivation of the HD-BSM procedure is shown in Methods.

To get a deeper insight on how the high-dimensional Bell-state measurement works, we use

the fact that in reverse a Bell state is generated. This simplifies the analysis because in this case

we can focus on one specific "click-pattern" as an example, and send single photons backwards

3
from these detectors. We choose to propagate three indistinguishable single photons from the

output ports {a0' , a1' , a2' } backwards through the multi-port, as shown in Fig. 1. Then we

condition onto cases where in each input port one and only one photon exists. Due to the all-to-

all connection in the multi-port, the resulting state contains all length-2 permutations of the 3D

states (Methods). Detection of the ancillary photon in the superposition state

(| 0 x  |1 x  | 2 x ) / 3 , results in the obtained state (normalization omitted):

| 0 a (|1 b  | 2 b ) |1 a (| 0 b  | 2 b ) | 2 a (| 0 b  |1 b ) [3]

The unitary transformation of this state to a target 3D Bell state (Eqn.2) requires an

expanded Hilbert space of four dimensions. The extra fourth level | 3 , is added to assist the

° 31 in Fig. 1), and erased afterwards,

physical realization of the unitary transformation (see U

which leads to the target 3D Bell state (Eqn. 2) |  00  ab  (| 00 ab  |11 ab  | 22 ab ) / 3

(Methods). The analysis holds exactly the same if the three indistinguishable photons are

“incident” from ports {b0' , b1' , b2' } or {x0' , x1' , x2' } .

Thus, in the experiment, a simultaneous click of the three detectors in the ports {a0' , a1' , a2' } ,

{b0' , b1' , b2' } or {x0' , x1' , x2' } indicates an unambiguous projection of the input photons a and b to

the 3D Bell state |  00  ab , which projects Bob’s photon c onto |   c   0 | 0 c  1 |1 c   2 | 2 c .

This state is already identical to the original state of photon a without the need of any additional

unitary corrections. The success probability of the HD-BSM using this scheme is 1/81.

Combining with active feed-forward techniques increases the success probability to 1/9 for

linear optics.

Figure 2 shows the experimental set-up for the 3D quantum teleportation. A femtosecond

pulsed laser beam is split into two parts to simultaneously create two photon pairs. The first

part of the pump beam is divided into three paths by two beam-displacers which are then

4
focused on the same β-borate-borate (BBO1) crystal. We select the case where in total one

photon pair is produced by type-II beam-like spontaneous parametric down-conversion31 but

without knowing at which one of the three paths, which generates the desired entangled state

|  00  bc used as the quantum channel for the 3D teleportation. To ensure long-term phase

stability between the three paths, we specifically design and fabricate interferometers with small

(4 mm) separation between the three paths. Hence, air fluctuations and disturbances act

collectively on all paths such that the qutrits are effectively protected in a decoherence-free

space.

Before being sent to the HD-BSM, photon b from the entangled qutrits first undergoes the

° 31 , which is experimentally realized using a network of polarising

unitary transformation U

beam splitters and half-wave plates (HWPs). The details are shown in Methods. Another pump

beam from the same laser passes through BBO2 and creates the second photon pair. One of

them is used for the preparation of an arbitrary superposition of the three paths as the input state

a to be teleported. The other one is used as the ancillary qutrit x in the HD-BSM.

The experimental realization of the HD-BSM puts stringent technological requirements on

phase stability, efficiency, and precision. To meet these demands, the HD-BSM is operated in

hybrid polarisation-path encoding and employs a fully connected three-input and three-output

ultra-low-loss multi-port interferometer. As shown in Fig. 2, in the input, single photons a and

b are initialised in horizontal polarization, while photon x is in vertical polarization. Photon a

is first combined with photon x using a PBS. The combined beams pass through a HWP set at

22.50, and are superposed with photon b on a partially-polarisation-dependent beam splitter

(pPDBS) which totally reflects vertically polarised photons and partially—with a ratio of 1/3—

reflects horizontally polarised photons. One of the output ports of the pPDBS is detected by

three detectors directly, while the other port is further sent through a quarter-wave plate set at

5
450 and then split by a polarising beam splitter and detected by six detectors. It is

straightforward to check that all the three photons from the inputs a, b, and x are evenly

distributed to each of the output with a ratio of 1/3, realizing the most important function of the

multiport.

The all-to-all multiport in Fig. 2 involves three-photon nine-path Hong-Ou-Mandel

interferences at the polarising beam splitter and the pPDBS. All nine paths are synchronized to

arrive within ~10 fs of each other, a delay much smaller than the coherence time of the

narrowband (3 nm) filtered single photons (~450 fs). The use of compact and precisely aligned

beam-displacers ensures a good spatial overlap in all multi-path interferences simultaneously.

Verifications of all two-photon Hong-Ou-Mandel interference combinations at the pPDBS with

an average visibility of 0.82(1) are presented in Methods. These visibilities in combination with

the entanglement source (with a measured fidelity of 0.94(1)) quantify the quality of the three-

dimensional teleportation experiment, see Methods for details.

It is necessary to prove that the teleportation experiment works universally for all possible

superposition states in the general form of |   a (eqn. [1]) and has a performance exceeding that

using only classical methods. Classically, the optimal single-copy state-estimation fidelity of a

three-level quantum system32 is 0.5 when averaging over the whole Hilbert space. Sampling

only over partial state space in biased bases, however, would allow the classical strategy to

make use of the biased information to obtain an average state estimation fidelity higher than

0.5. It is therefore important to carefully choose a minimal set of input states such that the

random sampling of which leads to the same classical limit as sampling over the whole state

space. Such a minimal set of states lies in mutually unbiased bases15. For a three-dimensional

system, we need to measure 12 states from four mutually unbiased bases:

6
 (1)
B{1,2,3} : (1, 0, 0), (0,1, 0), (0, 0,1)
B{1,2,3} : (1,1,1), (1,  ,  2 ), (1,  2 ,  )
(2)

. [4]
B{1,2,3} : ( ,1,1), (1,  ,1), (1,1,  )
( 3)

B{1,2,3} : ( 2 ,1,1), (1,  2 ,1), (1,1,  2 )

(4)

Here the vectors (  0 , 1 ,  2 ) denote the state  0 | 0 c  1 |1 c   2 | 2 c ,   exp(i 2 / 3) ,

where the normalising constant is omitted. We measure fidelities of the final teleported states,

defined as the overlap of the experimentally measured density matrix c with the ideal input

state |   a , which can be written as Tr(|     ) . Conditioned on the detection of a specific

“click-pattern” within the HD-BSM (see Fig. 1), we register the counts of Bob’s photon and

analyse its properties. The verifications of the teleportation results are based on four-fold

coincidence detections which in our experiment occur with a rate of ~0.11 Hz. In each setting,

the typical data accumulation time is 20-40 minutes, which allows to sufficiently suppress

Poisson noise.

(1)
Figure 3a shows the teleportation results of group B{1,2,3} , which can be straightforwardly

measured in the computational basis. The extracted fidelities are 0.76(3), 0.81(3), and 0.78(3)

for the teleported state | 0 , |1 , and | 2 , respectively. However, the measurements for the

other three groups, which involves equal superpositions of all the three levels, are more

complicated. To experimentally access the fidelity of these states in the general form

  ( 0  exp(i1 ) 1  exp(i2 ) 2 ) / 3 , [5]

we decompose the density matrix into three parts,   ( 012   021   120 ) / 3 , where

 ijk | ij  ij  | ij  ij  | k  k ,

| ij   (| i  exp(i j  i ) | j ) / 2 .

The decomposition unitarily transforms the qutrits into two-dimensional superposition states

and one computational state. Our measurement apparatus allows a simultaneous three-outcome

7
readout, directly accessing one of the  ijk (Methods). We show in Fig. 3b, c, and d the

( 2) (3) ( 4)
measurement results for three representative states from the group of B{1,2,3} , B{1,2,3} , and B{1,2,3} ,

respectively. The other six states are presented in Extended Data Figure 6. We note that all

reported measurements are without background or accidental count subtraction. The fidelity

imperfection is mainly from double pair emissions, spatial mode mismatch in the multi-photon

multi-path interferences, and interferometric noise in the state preparation and measurements

(Methods).

The fidelities of all the 12 states are displayed in Fig. 4, which are the minimal set allowing

us to faithfully derive the teleportation fidelity for the three-level quantum system. In the current

experiment, the averaged fidelity is calculated to be 0.75(1), well above the classical limit of

0.5 which can be obtained with the best classical strategy.

Proving the universality and non-classicality is already sufficient for teleporting qubits.

However, for the N-dimensional teleportation, it is important to further verify that all N-

dimensions still can form a coherent superposition and thus survived the teleportation intact.

Hence, a genuine N-dimensional teleportation should be distinguished from lower-dimensional

cases by excluding the hypotheses that the teleported state could be represented with less

dimensions. For our specific 3D teleportation, we can calculate that the maximal overlap

between any two-level superposition and the genuine three-level states is 2/3 (Methods). The

teleportation fidelity measured in our work exceeds this threshold by 9 standard deviations, thus

conclusively establishing a genuine three-dimensional quantum teleportation.

To summarize, we have for the first time demonstrated the possibility to completely teleport

the multiple quantized levels of a quantum system. Our generalized scheme (Methods) can

readily be applied to other degrees of freedom, such as photon’s orbital angular momentum33

and to larger dimensions, which would be suitable to be implemented in an integrated photonics

8
platform24, 26. It would also be interesting to investigate in the future the teleportation of multiple

atomic levels in trapped ions10,11 and cold atoms9.

The ability to perform the HD-BSM developed in this work provides a new possibility for

fundamental test of Bell’s inequalities and advanced quantum information technologies. As an

example, the to-be-teleported HD quantum state can itself be fully undefined, such as being part

of a two-particle HD entanglement. This leads to entanglement swapping34, where heralded by

a HD-BSM click, two remote HD particles can be entangled with no direct interaction. Such a

scheme can distribute entanglement over long distances and can enable an event-ready Bell test.

Remarkably, the created HD entanglement can tolerate a higher detection inefficiency than the

qubit case21, 35, and would provide significant advantages in long-distance Bell test closing both

locality and detection loopholes36-38 and device-independent quantum key distribution39.

Note: After completing our work, we became aware of an independent experiment

arXiv:1904.12249.

References

1. Wootters, W. & Zurek, W. A single quantum cannot be cloned. Nature 299, 802–803
(1982).
2. Bennett, C. H. et al. Teleporting an unknown quantum state via dual classic and Einstein-
Podolsky-Rosen channels. Phys. Rev. Lett. 70, 1895–1899 (1993).
3. Bouwmeester, D. et al. Experimental quantum teleportation. Nature 390, 575–579
(1997).
4. Furusawa, A. et al. Unconditional quantum teleportation. Science 282, 706–709 (1998).
5. Marcikic, I., de Riedmatten, H., Tittel, W., Zbinden, H. & Gisin, N. Long-distance
teleportation of qubits at telecommunication wavelengths. Nature 421, 509–513 (2003).

6. Zhang, Q. et al. Experimental quantum teleportation of a two-qubit composite system.

Nat. Physics 2, 678–682 (2006).

9
7. Wang, X.-L. et al. Quantum teleportation of multiple degrees of freedom of a single
photon. Nature 518, 516–519 (2015).
8. Ren, J.-G. et al. Ground-to-satellite quantum teleportation. Nature 549, 70–73 (2017).
9. Bao, X.-H., et al. Quantum teleportation between remote atomic-ensemble quantum
memories. Proc. Natl. Acad. Sci. 109, 20347-20351 (2012).
10. Riebe, M. et al. Deterministic quantum teleportation with atoms. Nature 429, 734–737
(2004).
11. Barrett, M. et al. Deterministic quantum teleportation of atomic qubits. Nature 429, 737–
739 (2004).
12. Pfaff, W. et al. Unconditional quantum teleportation between distant solid-state quantum
bits. Science 345, 532–535 (2014).
13. Steffen, L. et al. Deterministic quantum teleportation with feed-forward in a solid state
system. Nature 500, 319–322 (2013).
14. Fickler, R. et al. Quantum entanglement of high angular momenta. Science 338, 640–643
(2012).
15. Ivanovic, I. D. Geometrical description of quantal state determination. J. Phys. A: Math.
Gen. 14, 3241–3245 (1981).
16. Kimble, H. J. The quantum internet. Nature 453, 1023-1030 (2008).
17. Cerf, N., Bourennane, M., Karlsson, A. & Gisin, N. Security of quantum key distribution
using d-level systems. Phys. Rev. Lett. 88, 127902 (2002).
18. Ali-Khan, I., Broadbent, C. J. & Howell, J. C. Large-alphabet quantum key distribution
using energy-time entangled bipartite states. Phys. Rev. Lett. 98, 060503 (2007).
19. Neeley, M. et al. Emulation of a quantum spin with a superconducting phase qudit.
Science 325, 722–725 (2009).
20. Lanyon, B. P. et al. Simplifying quantum logic using higher-dimensional Hilbert spaces.
Nat. Physics 5, 134–140 (2009).
21. Collins, D., Gisin, N., Linden, N., Massar, S. & Popescu, S. Bell inequalities for
arbitrarily high-dimensional systems. Phys. Rev. Lett. 88, 040404 (2002).
22. Krenn, M. et al. Generation and confirmation of a (100×100)-dimensional entangled
quantum system. Proc. Natl. Acad. Sci. 111, 6243-6247 (2014).
23. Xie, Z.-D. et al. Harnessing high-dimensional hyperentanglement through a biphoton
frequency comb, Nat. Photonics 9, 536–542 (2015).
24. Kues, M. et al. On-chip generation of high-dimensional entangled quantum states and
their coherent control. Nature 546, 622–626 (2017).

10
25. Erhard, M. et al. Experimental Greenberger–Horne–Zeilinger entanglement beyond
qubits. Nat. Photonics 12, 759 (2018).
26. Wang, J. et al. Multidimensional quantum entanglement with large-scale integrated
optics. Science 360, 285–291 (2018).
27. Reck, M., et al. Experimental realization of any discrete unitary operator. Phys. Rev. Lett.
73, 58-61 (1994).
28. Weinfurter, H. Experimental Bell-state analysis. Europhys. Lett. 25, 559-564 (1994).
29. Calsamiglia, J. Generalized measurements by linear elements. Phys. Rev. A 65,
030301(R) (2002).
30. Żukowski, M., Zeilinger A., & Horne M. Realizable higher-dimensional two-particle
entanglements via multiport beam splitters. Phys. Rev. A 55, 2564-2579 (1997).
31. Wang, X.-L. et al. Experimental ten-photon entanglement. Phys. Rev. Lett. 117, 210502
(2016).
32. Bruß, D. & Macchiavello, C. Optimal state estimation for d-dimensional quantum
systems. Phys. Rev. A 253, 249–251 (1999).
33. Yao, A. M. & Padgett M. J. Orbital angular momentum: origins, behavior and
applications. Adv. Opt. Photonics 3, 161-204 (2011).
34. Żukowski, M., Zeilinger, A., Horne, M. & Ekert, A. Event-ready-detectors bell
experiment via entanglement swapping. Phys. Rev. Lett. 71, 4287-4289 (1993).
35. Vértesi, T., Pironio, S. & Brunner, N. Closing the detection loophole in Bell experiments
using qudits. Phys. Rev. Lett. 104, 060401 (2010).
36. Hensen, B. et al. Loophole-free Bell inequality violation using electron spins separated
by 1.3 kilometres. Nature 526, 682–686 (2015).
37. Shalm, L. K. et al. Strong loophole-free test of local realism. Phys. Rev. Lett. 115, 250402
(2015).
38. Giustina, M. et al. Significant-loophole-free test of Bell’s theorem with entangled
photons. Phys. Rev. Lett. 115, 250401 (2015).
39. Acín, A., Gisin, N. & Masanes, L. From Bell’s theorem to secure quantum key
distribution. Phys. Rev. Lett. 97, 120405 (2006).

Acknowledgements: This work was supported by the National Natural Science Foundation
of China, the Chinese Academy of Sciences, and the National Fundamental Research
Program.

11
Figure captions

Figure 1 | Principle scheme for teleportation of three-dimensional quantum states. Alice

holds a quantum state |   a encoded in three dimensions (depicted by three paths) that she

wishes to teleport to Bob. To do so, they first share a three-dimensional, maximally entangled

state. Then Alice performs a high-dimensional Bell-state measurement (HD-BSM) on her

photons. Conceptually, our approach upon realising a HD-BSM consists of two parts: A

° 31 ), and a multi-port beam splitter that

unitary transformation in an expanded state space ( U

enables collective quantum interference between Alice’s teleportee photon (a), her part of the

entangled state (b) and an additional ancillary photon (x). Specific “click-patterns” of different

detectors indicate successful projections into one of the nine entangled Bell states. Alice can

now transmit the classical-information of her “click-pattern” to Bob, who performs a unitary

transformation (U3) on his photon to recover the original state of Alice’s teleportee photon.

Figure 2 | Experimental set-up to teleport path-encoded qutrits. An ultraviolet pulsed

laser is used to create a three-dimensionally entangled photon-pair (path-encoded) in a non-

linear crystal (BBO1) shared between Alice and Bob. The teleportee and ancillary photon are

produced in a second non-linear crystal (BBO2). All 12 input qutrit states to be teleported and

the ancilla-photon are prepared using polarisation dependent beam-displacers (BDs)

controlled by half- and quarter-waver plates (HWP, QWP). The expanded unitary

° 31 is implemented in a four-dimensional hybrid polarisation-path state

transformation U

space. A polarising beam splitter (PBS) traces out the additionally employed fourth-

dimension. All three photons (a, b, x) enter the three-dimensional multi-port beamsplitter

which consists of nested interferometers implemented in polarisation and path degree-of-

freedom. A specifically designed partially polarising beam splitter (pPDBS) ensures equally

distributed input ports to all output ports. Simultaneous “click-patterns” of detectors

12
{a0' , a1' , a2' } , {b0' , b1' , b2' } or {x0' , x1' , x2' } indicates a successful BSM and heralds a teleported

photon at Bob’s side. Adjusting the HWP and QWP in Bob’s measurement apparatus allows

for a complete analysis of the teleported qutrits.

Figure 3 | Experimental results of qutrit teleportation. Measurement results for 6 out of all
(14)
12 basis states from different mutually unbiased bases groups B{1,2,3} for calculating the

fidelities are displayed. Dashed empty bars indicate ideal measurement result for comparison.
(1)
a All three computational basis states from the group B{1,2,3} and their relative four-photon

occurrences are shown. b-d Measurement result of coherent superposition states from
( 2) (3) ( 4)
mutually unbiased bases groups B{1,2,3} , B{1,2,3} , and B{1,2,3} respectively. The different

measurement outcomes | ij  represent all possible two-dimensional combinations with

phases according to the prepared qutrit state. Error bars are calculated using Monte-Carlo

simulation with an underlying Poissonian count rate distribution.

Figure 4 | Data summary for demonstrating a universal, non-classical and genuine qutrit

teleportation experiment. Here the fidelities of all 12 basis states are listed to demonstrate

universality. Our average achieved fidelity of 0.75(1) significantly overcomes both, the non-

classical bound of 1/2 and the genuine qutrit bound of 2/3. The shaded area is 1 standard

deviation of the average fidelity. Error bars for fidelities are calculated with a Monte-Carlo

simulation and an underlying Poissonian distribution of photon counts. Gaussian error

propagation yields the error for the average fidelity.

13
Figures:

Figure 1

Figure 2

14
Figure 3

Figure 4

15
Supplementary Information

16
17
18
19
20
21
22