1.Key Laboratory of Quantum Information, University of Science and Technology of China, Hefei 230026, China 2.Synergetic Innovation Center of Quantum Information & Quantum Physics, University of Science and Technology of China, Hefei 230026, China
Fund Project:Project supported by the National Natural Science Foundation of China (Grant Nos. 61435011, 61275115, 61525504) and Anhui Initiative in Quantum Information Technologies, China (Grant No. AHY020200).
Received Date:17 December 2018
Accepted Date:06 January 2019
Available Online:01 February 2019
Published Online:05 February 2019
Abstract:Quantum memories are indispensable for quantum communication and quantum computation, which are able to realize the storage and retrieval of a quantum state such as a single photon, entanglement, or a squeezed state on demand. Among those memories realized by different protocols, the Raman quantum memory has advantages in its broadband and high-speed properties, resulting in huge potential applications in quantum network and quantum computation. However, the realization of Raman quantum memory for a true single photon and photonic entanglement is a challenging job. In this review, after briefly introducing the main benchmarks for quantum memories, showing the state of the art, we focus on the review of the experimental progress recently achieved in storing the quantum state by Raman scheme in our group. We believe that all achievements reviewed are very hopeful in building up a high-speed quantum network. Keywords:quantum storage/ quantum information
图2是实验装置简图. 我们通过一个二维磁光阱(MOT1)[60]制备了雪茄形的碱金属铷85(85Rb)冷原子系综, 而后利用泵浦1通过自发拉曼散射(SRS)制备了一个波长在795 nm的反斯托克斯光子(信号光子1), 随后将产生的信号光子1传送到第二个磁光阱MOT2中, 借助一束正交偏振控制脉冲激光对信号光子1进行拉曼存储. 通过这种方法, 建立了两个MOT之间的非经典关联. 在读出MOT2中的信号光子1后, 用泵浦激光2将MOT1中原子系综的集体自旋激发态映射为信号光子2, 进而测量了两个信号光子之间的互相关函数${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$, 结果如图3所示. 信号光子1和信号光子2之间的时域相关性可以通过检验Cauchy-Schwarz不等式[40]是否被违反进行证明. 通常经典光满足不等式$R = \displaystyle\frac{{{{[{g_{{\rm{s1, s2}}}}(\tau )]}^2}}}{{{g_{{\rm{s1, s1}}}}(0){g_{{\rm{s2, s2}}}}(0)}} \leqslant 1$. 如果R > 1, 则光是非经典的. 其中gs1,s2, gs1,s1(0)和gs2,s2(0)分别是光子归一化的二阶互相关和自相关系数. 实验表明, 当信号光子1的存储时间为150 ns时, gs1,s2$\left( \tau \right)$具有最大值24. 同时考虑到gs1,s1(0) = gs2,s2(0) ≈ 2 (信号1和信号2的光子具有典型的热光场统计特性), 此时R = 144, 远大于1, 强烈地违反了Cauchy-Schwarz不等式, 从而清晰地证明了储存过程中非经典关联的保持. 图 2 简化的能级图和实验装置图[29] (a)简化的SRS能级图, 态$\left| 1 \right\rangle $和$\left| 2 \right\rangle $分别对应于85Rb原子的两个亚稳态能级5S1/2 (F = 3)和5S1/2(F = 2), $\left| 3 \right\rangle $和$\left| 4 \right\rangle $分别对应激发态能级5P3/2(F' = 3)和5P1/2(F' = 3); 泵浦光1由外腔二极管激光器(DL100, Toptica) 产生, 波长为795 nm, 与原子跃迁5S1/2(F = 3)→5P1/2(F' = 3)蓝失谐值70 MHz; 泵浦光2来自另一个波长为780 nm的外腔二极管激光器(DL100, Toptica), 对应5S1/2(F = 2)→5P3/2(F' = 3)的原子跃迁; 泵浦光1和泵浦光2被调制成脉冲模式, 脉冲宽度分别为50和160 ns, 上升沿为30 ns; 在存储过程中泵浦1和泵浦2脉冲之间的延迟时间被设置为260 ns; 泵浦1和泵浦2的激光功率分别为0.5和4 mW; 控制光来自于与泵浦光1相同的激光器, 也对应于原子跃迁5S1/2(F = 3)到5P1/2(F' = 3), 并蓝失谐值70 MHz, 功率为12 mW; (b)实验装置简化图, MOT 2中信号1的束腰为63 ${\text{μm}}$(MOT, 磁光阱; FC, 光纤耦合器; SLM, 空间光调制器; PBS, 偏振分束器; $\lambda$/2, 半波片) Figure2. Simplified energy level diagram and experimental setup[29]. (a) Simplified energy level diagram of the SRS. The states $\left| 1 \right\rangle $ and $\left| 2 \right\rangle $ correspond to two metastable levels 5S1/2(F = 3) and 5S1/2(F = 2) of 85Rb atom respectively, $\left| 3 \right\rangle $ and $\left| 4 \right\rangle $ are the excited levels of 5P3/2(F' = 3) and 5P1/2(F' = 3) respectively. The pump 1 laser is from an external-cavity diode laser (DL100, Toptica) with the wavelength of 795 nm, and is blue-detuned to the atomic transition of 5S1/2(F = 3)→5P1/2(F' = 3) with a value of 70 MHz. The pump 2 laser is from another external-cavity diode laser (DL100, Toptica) with the wavelength of 780 nm which couples the atomic transition of 5S1/2(F = 2)→5P3/2(F' = 3). The pump 1 and pump 2 are modulated into pulse modes with a width of 50 and 160 ns respectively, and a rising edge of 30 ns. The delayed time between the pump 1 pulse and the pump 2 pulse is programmed to be 260 ns for the process of storage. The powers of pump 1 and pump 2 are 0.5 and 4 mW respectively. The coupling laser is from the same laser with pump 1 and is also blue-detuned to atomic transition of 5S1/2(F = 3)→5P1/2(F' = 3) with a value of 70 MHz, its power is about 12 mW. (b) Simplified diagram depicting the storage of entanglement of OAM state. The waist of signal 1 at MOT 2 was 63 ${\text{μm}}$. MOT, magneto-optical trap; FC, fibre coupler; SLM, spatial light modulator; PBS, polarisation beam splitter; $\lambda$/2, half-wave plate.
图 3 存储过程中的交叉相关函数${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$[58] (a)信号光子1和信号光子2之间的交叉相关函数${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$, 泵浦1和泵浦2之间延迟时间为260 ns; (b), (c)和(d)是信号光子2与读出的信号光子1之间的时间相关函数${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right),$ 信号光子1的存储时间分别为100, 150和200 ns; (e)在没有输入信号1至MOT2的情况下收集的噪音; 所有数据均为原始数据, 无噪声校正 Figure3. Cross-correlated function of ${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$ in the process of storage[58]: (a) Cross-correlated function ${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$ between signal 1 and signal 2 photons with a delayed time of 260 ns between pump 1 and pump 2; (b), (c) and (d) were the time-correlated function ${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$ between signal 2 photon and the retrieval signal 1 photon with storage time of 100, 150 and 200 ns respectively; (e) the collected noise without the input signal. The signal 1 acted as trigger photon, and the signal 2 acted as stop signal. All data were raw, without noise correction.
随后通过测量信号光子1的条件自相关参数gs1;s1/s2(t)来检验存储过程中单光子特性是否保持. 一个严格的单光子gs1;s1/s2(t) = 0, 一个双光子态gs1;s1/s2(t) = 0.5. 而对于经典场, gs1;s1/s2(t) ≥ 1. 其中${g_{{\rm{s1}};{\rm{s1}}/{\rm{s2}}}}\left( t \right) = \displaystyle\frac{{{P_1}{P_{123}}}}{{{P_{12}}{P_{13}}}}$, P1是信号光子2的计数, P12和P13是信号光子2和两个分离信号光子1之间的二重符合计数, P123是三重符合计数. 因此gs1;s1/s2(t) < 0.5显示了单光子特性. 实验得到存储前gs1;s1/s2(t)为0.074 ± 0.012, 在150 ns的长时间存储后为0.29 ± 0.02, 从而证明了存储后信号光子1的单光子特性保持不变. 实验结果清晰地证明了在冷原子系综中可以通过拉曼方案成功地存储真正的单光子, 实验存储效率为26.7%. 在上述实验中, 信号光子1的带宽和蓝失谐分别为20和70 MHz. 随后, 通过任意函数发生器调制泵浦光1, 进一步增加了信号光子1的带宽. 受实验控制系统分辨率的限制, 信号光子1的最小脉冲宽度约为7 ns (约140 MHz). 图4给出了信号光子1的存储结果, 其中图4(a)表示输入信号, 图4(b)是存储后的结果, 图4(c)是噪声(信号光子1被阻挡). 存储效率为10.3%, 读出的信号光子1和信号光子2之间的二阶互相关(${g_{{\rm{s}}1,{\rm{s}}2}}\left( \tau \right)$)约为13.6, 表明存储后它们之间仍然存在非经典相关性. 如果能提高实验控制系统的时间分辨率, 则可以实现一个脉宽更小的脉冲(< 7 ns)的存储. 图 4 (a) 存储前反斯托克斯和斯托克斯光子之间的符合; (b)存储后斯托克斯光子和恢复反斯托克斯光子之间的符合; (c)噪声符合[58] Figure4. (a) Coincidence between the anti-Stokes and Stokes photons without storage; (b) coincidence between the Stokes and retrieved anti-Stokes photons; (c) coincidence from noise [58].
我们还进一步研究了大失谐条件下的存储性能. 通过改变泵浦光1和控制激光的频率, 使它们与原子跃迁$\left| 1 \right\rangle \to \left| 4 \right\rangle $的失谐为200 MHz, 因此产生的信号光子1也与原子跃迁$\left| 2 \right\rangle \to \left| 4 \right\rangle $失谐200 MHz. 为了在单光子水平上进行拉曼存储, 需要尽可能地增加控制激光的功率. 在存储过程中, 泵浦激光1的脉宽为50 ns, 控制激光的功率为110 mW, 束腰为2 mm. 同时, 在滤波系统中插入了自制的F-P腔滤波器, 最终消光比约为109 : 1, 足以消除控制激光的散射噪声. 实验结果如图5所示, 其中图5(a)为存储前信号光子1与信号光子2之间的符合, 图5(b)为存储后的数据, 图5(c)是噪声. 存储后的gAS, S$\left( \tau \right)$为5.6, 如果使用更多的滤波器来降低噪声, 那么信噪比就会得到提高. 图 5 (a), (b)存储前后单光子失谐200 MHz的反斯托克斯光子与斯托克斯光子的符合计数; (c)噪声记录[58] Figure5. (a), (b) Coincidence between the anti-Stokes and the Stokes photons with a single photon detuning of +200 MHz before/after storage; (c) the recorded noise[58].
22.2.二维OAM纠缠存储 -->
2.2.二维OAM纠缠存储
带有OAM的光子其波前是一个螺旋面[61,62]. 由于OAM空间固有的无限维特性[63-65], 如果将光子编码在OAM空间则可以大幅度提高光子的信息携带量[66], 因此, 基于OAM的量子信息处理也成为近年来的研究热点. 基于OAM量子网络的建立涉及到OAM纠缠光子与物质之间的相干相互作用, 因此存储OAM纠缠态对于建立基于OAM量子网络至关重要. 接下来分别介绍近年来本研究组基于拉曼存储方案实现OAM纠缠存储方面的研究进展. 实验装置与图2相同, 首先利用SRS建立了反斯托克斯光子信号1与MOT1中原子集体自旋激发态之间的OAM纠缠, 其量子态可由$\left| \psi \right\rangle = \sum\limits_{m = - \infty }^{m = \infty } {{c_m}{{\left| m \right\rangle }_{{\rm{s}}1}} \otimes {{\left| { - m} \right\rangle }_{{\rm{a}}2}}} $表示[67,68], 其中下标s1和a2分别标记信号光子1和MOT1中原子系综的集体自旋激发态, ${\left| {{c_m}} \right|^2}$是激发概率, $\left| m \right\rangle $是量子数为m的OAM本征模, 系统的初始线性动量和角动量为零. 由于SRS过程中动量守恒, 因此反斯托克斯光子和原子自旋激发态的总角动量为零, 从而建立了它们之间的OAM关联. 通过拉曼方案将信号光子1存储在MOT2原子系综中, 从而建立了两个原子系综之间的OAM纠缠. 随后通过将两个原子自旋激发态分别映射到两个光子(信号1和信号2光子), 并检查它们的纠缠度来证明原子自旋激发态之间的OAM纠缠. 首先, 实验证明了在一个二维子空间中OAM纠缠的存储(子空间由$\left| m \right\rangle $和$\left| { - m} \right\rangle $基构成), 此时光子纠缠态为$ \left| \varPsi \right\rangle ={1 / {\sqrt 2 }}\left[{\left| m \right\rangle \left| m \right\rangle + }\right.$$\left.{\left| {-m} \right\rangle \left| {-m} \right\rangle } \right]$. 测量了信号光子2与读出信号光子1之间的时间相关函数, 并根据测量的符合计数率重构了密度矩阵[69](如图6所示). 通过与理想密度矩阵${\rho}_ {\rm{ideal}}$相比较, 计算出密度矩阵的保真度为84.6% ± 2.6%, 并与存储前重建的密度矩阵${\rho}_ {\rm{input}}$进行比较, 存储前密度矩阵的保真度为90.3% ± 0.8%. 图 6 存储前后重构的密度矩阵[29] (a)用于重构密度矩阵的OAM态; (b)/(d)和(c)/(e)分别是存储前/后重构密度矩阵的实部和虚部, 背景噪声已被减去; 其中(b)和(c)中每组测量时间为500 s, (d)和(e)中每组测量时间为1000 s; L/R分别表示OAM为1/?1. Figure6. Reconstructed density matrices before and after storage[29]: (a) Four OAM states for reconstructing density matrix; (b)/(d) and (c)/(e) are the real and imaginary parts of the reconstructed density matrix of the state before/after storage respectively. The background noise has been subtracted. The background noise was estimated by repeating the experiment without input signal 1 photon to MOT 2. The measurement time for each data was 500 s in (b) and (c) and 1000 s in (d) and (e). L/R represents OAM = 1/?1.
实现不同量子存储器之间的高维纠缠不是二维纠缠存储的简单扩展, 实验难度很大, 特别是在高维纠缠度的证明、纠缠维数的确定等方面存在着诸多挑战. 我们克服了诸多难题, 实现了两个量子存储器之间高维OAM纠缠的建立, 实验装置如图7所示. 首先在MOT 1的冷原子系综中利用SRS过程建立光子和集体自旋激发态之间的高维OAM纠缠, 然后用拉曼协议将这个光子存入到另一个冷原子系综中, 这样就在两个原子系综之间建立了高维OAM纠缠, 表示形式与前文相同, 为$\left| {\psi '} \right\rangle = \sum\limits_{m = - \infty }^{m = \infty } {{o_m}{{\left| { - m} \right\rangle }_{{\rm{a1}}}} \otimes {{\left| { - m} \right\rangle }_{{\rm{a2}}}}} $, 其中${\left| {{o_m}} \right|^2}$是不同模式m的概率, 下标a1/a2分别是指MOT 1/2中的原子集体自旋激发态. 通过将两个系综中的自旋激发态映射到两个光子(信号1和2的光子), 并对它们的纠缠度进行检验, 从而间接检验原子自旋激发态之间的纠缠. 首先验证一个三维纠缠的量子存储, 将信号光子1和2分别投影到9个不同的态$\left| {\psi _{1 - 9}} \right\rangle $ (分别对应态$\left| L \right\rangle $, $\left| G \right\rangle $, $\left| R \right\rangle $, $(\left| G \right\rangle +\left| L \right\rangle)/2^{1/2}$, $\left(| G \right\rangle\!+\!\left| R \right\rangle)/2^{1/2}$, $(\left| G \right\rangle\!+\!{\rm{i}}\left| L \right\rangle)/2^{1/2}$, $(\left| G \right\rangle\!-\!{\rm {i}} \left| R \right\rangle)/2^{1/2} $, $(\left| L \right\rangle + \left| R \right\rangle)/2^{1/2} $, $(\left| L \right\rangle + {\rm {i}}\left| R \right\rangle)/2^{1/2}$, 其中$\left| L \right\rangle $, $\left| G \right\rangle $和$\left| R \right\rangle $分别对应OAM为$ 1\hbar $, 0和$ - 1\hbar $), 进而重构出存储前(图8(a)和图8(b))和存储后(图8(c)和图8(d))的密度矩阵, 并计算出纠缠度为83.9% ± 2.9%. 图 7 实验装置[30], 其中透镜L1和L2用于将信号1的相位结构映射到MOT2的中心; 利用L3, L4和L5将位于MOT1中心的信号2的相位结构映射到SLM2的表面; L6和L7用于将信号2的OAM模式耦合到C2中; 图的右侧部分用于将信号1存储在MOT2, 并在读出后将其耦合到C1; C, 光纤耦合器; M, 反射镜; L, 透镜 Figure7. Experimental setup[30]. Lenses L1 and L2 are used to focus signal 1 on the centre of MOT 2. L3, L4, and L5 are used to focus the phase structure of signal 2 on the center of MOT 1 onto the surface of SLM 2. L6 and L7 are used to couple OAM mode of signal 2 to C2. There is an asymmetric optical path for coupling signal 1 into C1 in right frame of figure. C, fiber coupler; M, mirror; L, lens.
图 8 构造的三维密度矩阵[30] (a)和(b)分别是存储前的实部和虚部; (c)和(d)是存储后的实部和虚部 Figure8. Constructed density matrix of three-dimensional entanglement[30]. Panels (a) and (b) are the real and imaginary parts before storage; panels (c) and (d) that after storage.
如果系统存在d维纠缠, 则应该违反最大边界Md. 对于一个包含m = 2, 1, 0, ?1组成的量子态, 最大边界M4 = 9. 实验测量得到的存储前和存储后的$M' $值分别为9.30 ± 0.06和9.19 ± 0.06, 结果表明这两个远距离的原子系综之间至少存在一个四维纠缠. 我们还通过对每个基的可见度N进行求和来计算W, 进而确定高维纠缠的维数. 实验测量得到的可见度N如图9(a)和图9(b)表示, 分别对应于存储前和存储后的数据. 维度witness的值Wd[80]由以下公式给出: 图 9 (a) 和(b)是存储前后的可见度之和[30] Figure9. (a) and (b) are the sum of visibilities before and after storage[30].
下面介绍利用拉曼方案实现的两个与偏振DOF相关的存储实验: 1)单光子路径与偏振纠缠存储; 2)在两个冷原子系综中的偏振纠缠存储. 实验装置如图10所示, 利用一个原子系综作为非线性介质来制备单光子或双光子纠缠态, 使用另一个原子系综作为存储介质. 图 10 (a) 简化的能级图, 用于产生和存储偏振纠缠和单光子的生成、存储的时序; P1是泵浦光1, P2是泵浦光2; (b)简化的实验装置; L和R是MOT A中的两个SRS过程, H和V分别代水平极化和垂直极化, P1和P2分别为来自两个声光调制器的25 ns ($\Delta t $)和160 ns的调制脉冲; MOT, 磁光阱; FC, 光纤耦合器; PBS, 偏振分束器; $\lambda$/2, 半波片; $\lambda$/4, 四分之一波片; S, Stokes光子; As, 反-Stokes光子; D1, D2和D3分别是单光子探测器1, 2和3 (PerkinElmer SPCM-AQR-15-FC); PD, 自制光电探测器; PZT, 压电陶瓷; U和D分别是输入到MOT B中的上和下光模式; P, 半波板; $\theta$, 插入相位板的相位[58] Figure10. (a) Simplified energy level diagram used to generate and store the polarization entanglement and the time sequence for the generation, storage and retrieval sequence of a single photon. P1 is pump 1 and P2 is pump 2. (b) Ssimplified setup depicting the storage of the polarization entanglement. L and R are the two SRS processes in MOT A. H and V are the horizontal and vertical polarizations, respectively. P1 and P2 are the modulated pulses with 25 ns ($\Delta t $) and 160 ns from two acoustic optic modulators, respectively. MOT, magneto-optical trap; FC, fibre coupler; PBS, polarization beam splitter; $\lambda$/2, half-wave plate; $\lambda$/4, quarter-wave plate; S, Stokes photon; As, anti-Stokes photon; D1, D2 and D3 are single photon detectors 1, 2 and 3, respectively (PerkinElmer SPCM-AQR-15-FC); PD, home-made photoelectric detector; PZT, piezoelectric transducer; U and D are the up- and down-optical modes input into MOT B, respectively; P, half-wave plate; $\theta$, the phase of the inserted phase plate [58].
表1存储前后的${\bar p_{ij}}$以及$\bar C$ Table1.Values of ${\bar p_{ij}}$ and $\bar C$ before and after storage.
图 11 (a) 和(b) 分别为存储前后探测器D3探测到的Stokes光子与探测器D1 (圆形数据)和探测器D2 (三角形数据)分别探测到的反Stokes光子之间的符合计数; 实线是拟合曲线; 所有实验数据为原始数据, 没有进行误差校正; 误差为 ± 1的标准差[58] Figure11. (a) and (b) Coincidence between the Stokes photon detected by detector D3 and the anti-Stokes photon detected by detector D1 (circular data) and detector D2 (triangular data), respectively, with a different phase before/after storage. The solid lines are the fitted lines. All of the experimental data are raw data without error corrections. The error bars are ± 1 standard deviation [58].
图 12 (a) 存储前输入态密度矩阵; (b)存储后输出态的密度矩阵; 所有的实验数据都是原始数据, 没有进行任何误差修正[58] Figure12. (a) Density matrices of the input state before storage; (b) the output state after storage. All of the experimental data here are raw data without any error corrections[58].
接下来介绍两个冷原子系综中的光子偏振纠缠存储. 由马赫-曾德尔干涉仪中两束波长为795 nm的泵浦光驱动的两个对称SRS过程(L和R)相干叠加, 在MOT A中制备了反斯托克斯光子与原子系综集体自旋激发态之间的偏振纠缠, 纠缠态可以描述为$\left|\psi\right\rangle = \displaystyle\frac{1}{{\sqrt 2 }}(\left|L \right\rangle \left|H \right\rangle + {{\rm{e}}^{{\rm{i}}{\phi }}}\left|R\right\rangle \left|V \right\rangle)$, 其中第一项指代系综L中的SRS过程, 第二项表示系综R中的SRS过程, $\left|H \right\rangle (\left|V \right\rangle )$表示反斯托克斯光子的水平(垂直)偏振, $\left|L \right\rangle(\left. R \right\rangle )$表示原子系综L(R)中的集体自旋激发态, $\phi $是两个泵浦光路径差造成的两个反斯托克斯光子间的相位差, 在实验中将其设置为零. 我们主动锁定了干涉仪, 然后将反斯托克斯光子发送到MOT B中并利用拉曼协议实现存储, 从而在MOT A和MOT B中的原子集体自旋激发态之间建立了纠缠$ \left| {{\psi _{{\rm{aa}}}}} \right\rangle =$$\displaystyle\frac{1}{{\sqrt 2 }}({\left| U \right\rangle _{\rm{A}} }{\left| L \right\rangle _{\rm{B}}} + {\left| D \right\rangle _{\rm{A}}}{\left| R \right\rangle _{\rm{B}}})$. 在将MOT B中原子的集体自旋激发态读出为反斯托克斯光子后, 进而读出MOT A中的原子集体自旋激发态为斯托克斯光子, 从而在光子之间建立了偏振纠缠$ \left| {{\psi _2}} \right\rangle =$$\displaystyle\frac{1}{{\sqrt 2 }}(\left| H \right\rangle \left| V \right\rangle + \left| V \right\rangle \left| H \right\rangle )$. 在此过程中, MOT B中的态存储时间应该小于MOT A中的态存储时间, 从而保证纠缠的存储. 通过检测反斯托克斯光子和斯托克斯光子之间的纠缠证明了两个系综中存在偏振纠缠. 对纠缠态的密度矩阵进行重构, 结果如图13所示, 其中图13(a)和图13(b)分别给出了输入态的重构密度矩阵(${\rho}_ {\rm{input}}$)的实部和虚部, 与理想密度矩阵(${\rho}_ {\rm{ideal}}$)比较后, 计算得到重构密度矩阵保真度(F1)为89.3% ± 1.7%. 图13(c)和图13(d)分别给出了经过存储后密度矩阵的实部和虚部, 计算得到输出的保真度(F2)为85.0% ± 3.4%. 图 13 (a)/(c)和(b)/(d)分别为输入/输出的重构密度矩阵的实部和虚部; 所有实验数据为原始数据, 无误差校正[58] Figure13. (a)/(c) and (b)/(d) Reconstructed real and imaginary parts of the input/output density matrix, respectively. The density matrices were reconstructed with losses. All of the experimental data are raw data without error corrections[58].
为了刻画分立原子系综之间杂化纠缠的性质, 将原子自旋波激发态映射到光子态$\left(\left| {\psi {'_{{\text{2-hybrid}}}}} \right\rangle \right)$上. 通过密度矩阵重构, 得到密度矩阵$\left| {\psi {'_{{\text{1-hybrid}}}}} \right\rangle $和$\left| {\psi {'_{{\text{2-hybrid}}}}} \right\rangle $, 分别对应图15(a)和图15(b)(仅为实部). 通过比较$\left| {\psi {'_{{\text{1-hybrid}}}}} \right\rangle $和$\left| {\psi {'_{{\text{2-hybrid}}}}} \right\rangle $的密度矩阵, 计算出纠缠的保真度为93.6% ± 1.4%. 图 15 重构的杂化纠缠的密度矩阵实部[59] (a)存储前; (b)存储后 Figure15. Real parts of the constructed density matrices for hybrid entanglement: (a) Before storage; (b) after storage [59].
同时测量了双光子干涉来进一步表征杂化纠缠的特性. 图16(a)和图16(b)分别表示存储前后的干涉曲线. 存储前平均可见度为93.1%, 存储后平均可见度为84.6%. 这两个可见度值都大于违背Bell不等式的阈值70.7%, 充分证明了两个原子系综之间确实存在杂化纠缠. 图 16 杂化纠缠的双光子关联干涉曲线[59] (a)存储前; (b)存储后; 误差由泊松统计估计, 表示为 ± s.d; 所有数据均为原始数据, 没有进行误差纠正 Figure16. Interference curves of the two-photon correlations for hybrid entanglement [59]: (a) Before storage; (b) after storage. The error bars are estimated from Poisson statistics and represent as ± s.d. All the data are raw and not subjected to noise correction.