EXPERIMENTAL DEMONSTRATION OF QUANTUM KEY

EXPERIMENTAL DEMONSTRATION OF QUANTUM KEY DISTRIBUTION WITH ENTANGLED PHOTONS FOLLOWING THE PING- PONG CODING PROTOCOL Martin Ostermeyer, Nino Walenta University of Potsdam, Institute of Physics, Nonlinear Optics and experimental Quantum Information Am Neuen Palais 10, 14469 Potsdam, Germany Abstract: We demonstrate quantum key distribution (QKD) following the ping-pong coding protocol [1]. No transferred bits have to be discarded or checked via a public communication channel. A key transfer rate of 4250 bit/s is reached. The QKD makes use of polarization entangled photons generated by type II parametric down conversion. Keywords: Quantum cryptography, quantum key distribution, entanglement 1 Introduction The quantum key distribution protocol invented by Bennett and Brassard (BB84) [2] has become a standard in quantum key distribution (QKD). In the practical implementation of this protocol single photons are used as qubit carriers and statistically swapped basis sets e.g. for the photon polarization at the sender called Alice and the receiver called Bob. QKD protocols on the basis of entangled qubits have the potential of improved security and transfer rate of the key transmission. Thus, the ping-pong coding protocol by Boström and Felbinger [1] that we address in this paper provides an improved detection probability of an eavesdropper compared to the BB84 protocol. QKD using entangled photons has been demonstrated in a range of experiments before. Polarization entangled pairs of photons were used in the single particle BB84 protocol to improve the security compared to the operation with weak coherent pulses [3, 4]. They were also applied to protocols like the Ekert protocol [5] where both photons were

used for the key transmission [4] or the six state protocol [6] implemented by Enzer et.al [7]. Another class of experiments uses the time bin entanglement [8, 9] instead polarization entanglement. The SARG protocol [10] based on time bin qubits was recently used to demonstrate key distribution beyond 100 km with error rates of 8.9 % [11] and 5 % [12] respectively. In this paper we present to the best of our knowledge the first implementation of the ping-pong coding protocol for QKD. The inventors of the protocol also claim the capability of the protocol for quasi secure direct communication. However, with non ideal detectors showing quantum efficiencies below 100 % and noisy quantum channels a practical realization of this direct secure communication variant of the proposal is in question but not the QKD. In contrast to other implementations of QKD with entangled photons the pingpong coding is a deterministic secure protocol without the need of post processing of the key. The information gets encoded in the phase of the entangled qubits. Since only one qubit is transmitted from Bob to Alice (ping) and back from Alice to Bob (pong) the encoded information cannot be extracted from this single qubit. Decoding only becomes possible if a Bell state measurement on both qubits is performed to evaluate the correlation of both qubits with each other. The problem of entanglement distribution is resolved by the generic pingpong characteristic of the protocol. Both photons are generated and detected at Bob s site. Only one photon travels to Alice and then back to Bob. The problem of distributing the entanglement is transferred to the generation and storage of one photon at one of the communication partners (Bob). 2 Implementation of ping pong coding using polarization entangled photons For our implementation of the protocol polarization entangled photons are used to distribute the key. The information is deterministically encoded in the correlation of the photons and not the individual property of the single photon. To encode data in a binary alphabet two of the four Bell states are sufficient. We decided to use the states as 1 with [ H V ± V H ] ± 1 = 2 h t h t. as 0 and Here the index h or t denotes the travel or home photon analogous to the travel and home qubit described in the original protocol. Our implementation of the ping-pong protocol is depicted in Fig. 1. The polarization entangled photons are generated by type II spontaneous parametric down conversion (SPDC) [13] in a beta-barium-borate crystal of 1 mm length. For this proof of principle experiment the BBO crystal is pumped by a mode locked Picosecond laser with pulse duration of 8 ps average power of 300 mw, and repetition rates of 85 MHz at 355 nm wavelength. The pump spot in the BBO-crystal had a diameter of 180 µm. The photons were detected with fibre coupled avalanche diode detectors with active quenching (AQR-14 by Perkin Elmer). Their specified quantum efficiency at 710 nm is 72%. A Pockels cell with quarter wave voltage of

210 V was used to switch between the and state. The optical axis of the resulting quarter wave plate was aligned parallel to the vertical polarization direction. Pump laser Trigger diode BBO Travel photon Bob Alice Home photon Delay Eve D 1,B D 2,B R <100% CONTROL BSA PBS HWP BS Eve Pockels cell D V,A D H,A PBS Public channel Fig. 1: Set up for the optical implementation of the ping-pong-coding protocol. PBS denotes polarizing beam splitters, BSA the district for the Bell state analysis, HWP half wave plate, and BBO a beta- Barium-Borate crystal. The part of the intersection of the two SPDC emission cones that is coupled into the single mode fibres of the photon detectors was matched to the pump spot size [14] and corresponds to a bandwidth of 10 nm (FWHM). The single photon counting rate of the photons generated from SPDC was 1.2 10 5 counts/second. The coincidence rate amounts to 17 % of this value. The ratio between the two pair and single pair generation rate is around 4 10-2. Thus there is a reasonable small but limited chance of beam splitter attacks. The SPDC behind the BBO-crystal was corrected for runtime differences by a halfwave plate and one compensation crystal each per output path of the SPDC. The polarization entanglement of the source was characterized by verifying the CHSH inequality with an S parameter of 2.66±0.003. The home photon stays with Bob. It can be stored in a delay line e.g. a fibre loop. Within the proof of principle experiment in our lab Alice and Bob are just separated by about one meter. Thus, we used the multiple reflections between mirrors to store the home photon. The travel photon is send to Alice (ping). There are two different modes, the message mode and the control mode. In message mode (the standard mode) the travel photon bounces back from Alice to Bob (pong) and will be interfered with Bob s home photon at the non polarizing beam splitter BS. At Alice on the way to Bob a bit can be encoded via the Pockels cell. At Bob a special Bell analysis takes place. In case of the message mode travel and home will enter the BS simultaneously. There is either one photon each per output arm of the beamsplitter ( ) or two photons in one or the other arm together ( ). In case of one photon and only one photon has to travel along a longer arm with a half wave plate (HWP) in the path. The other one passes directly on to the polarizing beam splitter PBS. The

resulting signature for the state is two simultaneous clicking detectors D 1,B and D 2,B. The signature for is the clicking of both detectors D 1,B and D 2,B but with one delayed detector click because of the longer path for one of the photon ports between BS and PBS. Using this setup we can expect unique signatures for the two different Bell states that both click on two detectors. Because of the detector efficiency below 100 % and the dark counts of the detectors using always two klicking detectors serves for a higher degree of reliability in the evaluation of the Bell state. The back reflecting mirror at Bob is only partially reflecting with reflectivity R control. Thus, with a given probability R control the travel photon is transmitted through Alice mirror and a control run starts. In this case the polarization of the single travel photon gets characterized by Alice. And at Bob automatically the polarization of the single home photon gets characterized. The home photon simply enters the Bell state analyzer and propagates along the direct or detour path. Because of the half wave plate (HWP) a photon with vertical polarization will always click at D 1,B and a photon with horizontal polarization will always click at D 2,B. By comparing their results of this control run online Alice and Bob have the chance to detect an eavesdropper with 50 % probability [1]. The entire control mode sequence happens and runs automatically without any additional active switching. To increase the robustness of the entire scheme a trigger signal provided by the pump pulse of the SPDC is used. This trigger signal marks the time slot when photons are to be expected and serves as trigger to gate the single photon detectors. 3 Photon interference Within the operation of the setup presented in the last section the crucial point is the interference of the photons at the beam splitter that will enable Bob to distinguish the two Bell states and. If the photon pair of a -state hits the beam splitter with one photon each per input mode (BS, see Fig. 1) one photon each per beam splitter output will result [15]. On the contrary the -state yields two photons in one output with 50 % probability in each output path and no photon in the corresponding complementary output path. To test this expected interference we varied the length of the home photon s path via a translation stage (see path entitled Delay in Fig. 2 ) and measured the coincidence rate of the detectors D 1,B and D 2,B at equal detection times and delayed detection times. This delay means the specific delay time of the two detectors D 1,B and D 2,B defined by the photons flight time along the extra path length delay between BS and PBS which passes the half wave plate (HWV, see Fig. 2). This path via the half wave plate has a length of 1.74 m whereas the other output of BS that directly enters the polarizing beam splitter (PBS) is just 50 mm long. The difference of these two output paths of the beam splitter of 1.69 mm amounts to a delay flight time of 5.66 ns. Because of this extra length in one of the outputs of the beam splitter there is a dip at delay 0 s to be expected for the coincidences at equal detection times for a - state. And

there is a maximum to be expected for the state at equal detection times since both photons travel along the same output path of BS with necessarily equal flight times. The coincidences for detector clicks delayed by the extra 5.66 ns of the longer BS-output path produce the mirrored interference coincidence signal, meaning they yield a dip for the - state and a maximum for the - state. This means there are typical and easy to distinguish signatures for the two different Bell-states to be expected. D 1 Astigmatismus compensation JenLas 355-p5 L 1 Trigger Diode BD Fibre Coupler 2 C 2 1 L 9 L 7 L 6 L 8 Pockelscell L 2 PBS BS L 3 HWP 2 L 4 L 5 C 1 D 2 D 3 BBO HWP 1 Delay Fig. 2: Modified setup to test for the photon-photon interference at the beam splitter (BS). PBS denotes a polarizing beam splitter, BSA the elements for the Bell state analysis, HWP half wave plate, and BBO a beta-barium-borate crystal. 0.20 no Pockels cell voltage applied 0.20 Pockels cell voltage applied Correlation Coefficient 0.15 0.10 0.05 Correlation Coefficient 0.15 0.10 0.05 0.00-25 -20-15 -10-5 0 5 10 15 20 25 Stage Position (µm) 0.00-25 -20-15 -10-5 0 5 10 15 20 25 Stage Position (µm) Fig. 3: Measured photon interference at beam splitter BS with Pockels cell without applied voltage (left) and with applied voltage (right). The black dotted curves show the coincidences for equal detection times the red crossed curves mark the coincidence of 5.66 ns delayed detection times.

For the QKD-operation the entangled photon source is adjusted to produce - states. The Pockels cell with quarter wave voltage applied should convert them to - states. Two different coincidence signals are measured for each Pockels cell status. With and without applied quarter wave voltage coincidences with equal detection times and coincidences delayed by 5.66 ns are evaluated. Fig. 3 shows these four measured coincidence signals as a function of the translation stage position (see Fig. 2). The width of these signals corresponds roughly to the coherence length of the biphoton of 240 fs. The shape of the coincidences signals with a maximum or minimum corresponds to the above set forth expectation. The dips of the curves do not reach zero mainly because of the imperfection of the BS and PBS. The BS has an asymmetric ratio of transmission and reflectance of 0.37:0.57. As these measurements demonstrate the two Bell states are easily distinguishable at the stage position 0 µm (for equal path length of the home and travel photon) and can be switched as expected via the Pockels cell. No voltage applied to the Pockels cell yields the expected signature for the -state. Quarter wave voltage applied to the Pockels cell yields the expected signature for the -state. 4 Demonstration of the protocol To demonstrate the protocol we generated a random 10000 bit long binary key. The key was transmitted from Alice to Bob in the above explained way. The simultaneous and delayed coincidences are evaluated using a multichannel interface (Becker&Hickel SPC 134) in time tag modus operated by a PC. For this proof of principle demonstration 20,000 pulses per transmitted bit where used. At our single photon counting rate of 1.2 *10 5 this means that 6 detected photon pairs were used to transmit a single bit. The error rate was 3.8 %. The error rate is mostly limited to the non perfect synchronization of the different channels of the detector readout interface. This technical problem is not solved yet but is solely a technical problem. If the synchronization errors are excluded the error rate amounted to 1.8 %. The key was transmitted with 4250 bits/s. After the generation and transmission of the key a message can be transmitted via a public channel. The key is used by Alice to publicly submit the 10000 bit large logo of the University of Potsdam following the one-time pad protocol. An XOR-operation is applied bit by bit during the encoding at Alice and the decoding procedure at Bob (see Fig. 4).

Alice random key XOR public channel coded message Bob quantum channel XOR Fig. 4: Generated key and transmitted logo of the University of Potsdam. The key is transmitted from Alice to Bob via the quantum channel using entangled photons, whereas the logo is transferred via a public channel using the secret key. The transmitted logo is shown with the resulting corrected error rate of 1.8 %. In this Letter we demonstrated the first implementation of the ping-pong coding protocol as quantum key distribution protocol. The key is encoded in the polarization entangled states of photon pairs. Only one photon travels from Bob to Alice and back to Bob. The practicability of the implemented scheme in the current lab version gets evaluated. Replacement of the Bell analysis with the interference at the beam splitter might be desirable and will be examined although the length difference of home and travel photon is not critical in sub wavelength range. Currently a key transmission rate of 4250 bit/s are reached. The quantum bit error rate of 3.8 % can be reduced by a better synchronization of the detector read out. 5 References [1] K. Boström, T. Felbinger Phys. Rev. Lett. 89 (18) (2002) 187902-1-187902-4 "Deterministic Secure Direct Communication Using Entanglement [2] C.H. Bennett and G. Brassard Proc. IEEE Int. Conference on Computers, Systems, and Signal Processing, Bangalore (IEEE, New York), 175-179 (1984) [3] G. Brassard, N. Lütkenhaus, T. Mor, B. C. Sanders Phys. Rev. Lett. 85 (6) (2000) 1330-1333 Limitations on Practical Quantum Cryptography [4] T. Jennewein, C. Simon, G. Weihs, H. Weinfurter, A. Zeilinger Phys. Rev. Lett. 84 (20) (2000) 4729-4732 Quantum Cryptography with Entangled Photons

[5] A. K. Ekert Phys. Rev. Lett. 67 (6) (1991) 661-663 Quantum Cryptography Based on Bell s Theorem [6] D. Bruss PRL 81 3018 (1998) [7] D. G. Enzer, P. G. Hadley, R. J. Hughes, C. G. Peterson, P. G. Kwiat New Journal of Physics 4 (2002) 45.1-45.8 Entangled-photon six-state quantum cryptography [8] D. Stucki, N. Gisin, O. Guinnard, G. Ribordy, H. Zbinden New Journal of Physics 4 (2002) 41.1-41.8 Quantum key distribution over 67 km with a plug&play system [9] S. Fasel, N. Gisin, G. Ribordy, H. Zbinden Eur. Phys. J. D 30 (2004) 143-148 Quantum key distribution over 30 km of standard fiber using energy-time entangled photon pairs: a comparison of two chromatic dispersion reduction methods [10] Valerio Scarani, Antonio Acín, Grégoire Ribordy, and Nicolas Gisin, Quantum cryptography protocols robust against photon number splitting attacks for weak laser pulse implementations. Phys. Rev. Lett. 92 057901 (2004) [11] C. Gobby, Z. L. Yuan, and A. J. Shields. Quantum key distribution over 122 km of standard telecom fiber. Applied Physics Letters 84 3762 (2004) [12] P. A. Hiskett, D. Rosenberg, C. G. Peterson, R. J. Hughes, S. Nam, A. E. Lita, A. J. Miller, and J. E. Nordholt, Long-distance quantum key distribution in optical fiber, New Journal of Physics, 8 193 (2006) [13] P.G. Kwiat, K. Mattle, H. Weinfurter und A. Zeilinger, New high-intensity source of polarization-entangled photon pairs, Physical Review Letters 75 4337 (1995) [14] C. Kurtsiefer, M. Oberparleiter und H. Weinfurter, High efficiency entangled photon pair collection in type ii parametric fluorescence, Physical Review A 64 023802 (2001) [15] D. Bouwmeester, A. Ekert, A. Zeilinger (Eds.), The Physics of Quantum Information, Springer (2000), Chapter 3.6