We report on the generation of quadrature squeezed light at 1.535 µm via single-pass optical parametric amplification in a periodically poled MgO:LiNbO3 waveguide, and detection with a temporally shaped local oscillator. Squeezing of -4.1 dB was directly measured using the shaped local oscillator. Classical parametric gain of the shaped pulse was also investigated; a deamplification gain of -12.1 dB was observed with the amplification gain of only +13.8 dB. We experimentally show that the use of shaped pulse as local oscillator in homodyne detection allows us efficient squeezing detection with near-unit mode-matching efficiency.
©2008 Optical Society of America
Quadrature-squeezed light is an important resource for continuous-variable quantum communication (CVQC) technology, such as quantum teleportation and dense coding . In particular, broadband squeezed light at telecommunication wavelength would be the powerful resource for practical implementations of CVQC, because optical transmission loss in a glass optical fiber is minimum at the telecommunication band, and the broadband source is required for the high-speed and broadband communications [2, 3, 4].
Observation of 10 dB squeezing was recently reported by Vahlbruch et al., showing that strong squeezing as required for practical applications is possible . In their experiment, optical cavities were employed to enhance the optical nonlinearity, and to reduce the phase noise and spatial-mode mismatching with a local oscillator (LO). However, in a cavity-based system the squeezing bandwidth is limited by the cavity bandwidth. Broadband and large squeezing can be obtained by single-pass traveling-wave optical parametric amplification (OPA) pumped by pulsed light . The drawback of pulsed squeezing by single-pass OPA lies in the difficulty of obtaining the LO that is spatiotemporally matched with squeezed pulse. The difficulty about the spatial mode-matching is caused by the phenomenon of gain induced diffraction (GID) in a bulk nonlinear crystal [7, 8]. The phase front of the squeezed light is distorted by the GID in the high parametric gain regime. On the other hand, the temporal mode-mismatching is induced by using the LO pulse whose duration is longer than the pump pulse for the OPA. To obtain the high temporal mode-matching efficiency, the temporal duration of LO pulse should be as the same as or shorter than that of pump pulse [10, 11, 12]. The magnitude of the detectable squeezing is expected to increase with the shortening of LO duration in the case of where the squeezing bandwidth is enough wider than that of LO.
In order to obtain the LO spatiotemporally matched with squeezed pulse, Aytür et al.  utilized OPA process in a KTiOPO4 (KTP) crystal, that is similar to the one used for generating squeezed pulse. Squeezing of 2 dB at wavelength 1064 nm was observed, but observation of higher squeezing was prevented by the uncontrollable phase fluctuations. Kim et al.  generated both squeezed pulse and the spatiotemporally matched LO pulse in a same type II KTP crystal, and observed 5.8 dB of squeezing at 1064 nm. Anderson et al.  used a periodically poled KTP waveguide as nonlinear medium to avoid GID effect. In their experiment, LO with the temporal duration of 2.5 times shorter than that of pump pulse was employed and 0.6 dB (12 %) of squeezing at 830 nm was observed. Recently we have reported 3.2 dB squeezing generated from a single-pass OPA in a periodically poled MgO:LiNbO3 waveguide (PPLN-WG) . We observed -8.9 dB of classical parametric deamplification, showing that GID can be avoided by using a waveguide. However, presence of mode-mismatch between the squeezed pulse and LO pulse was inferred from the experimental data.
In this paper, we report the experimental observation of pulsed squeezed light in which a LO pulse temporally shaped by OPA in a PPLN-WG was employed. The temporal duration of the shaped LO is shorter than that of the pump pulse used for squeezed light generation.
Best squeezing of -4.1 dB was observed with the antisqueezing of +11.5 dB. This represents an improvement over our previous record of -3.2 dB, and it is the highest squeezing obtained using a nonlinearwaveguide, as well as the highest squeezing at telecommunicationwavelength obtained in χ (2) nonlinear media and also the highest squeezing observed in the time domain by pulsed homodyne detection. We investigated the classical parametric gain of OPA in a PPLN-WG using the shaped pulse as probe, and best deamplification of -12.1 dB were observed with the corresponding amplification gain of +13.8 dB. To our knowledge, this is the highest deamplification by OPA.
To investigate the effect of temporal shaping of the LO pulse, we compare the experimental results obtained using shaped LO to that obtained using non-shaped LO which has similar duration to the pump pulse. The results indicate that the spatiotemporal mode-matching efficiency obtained by using the shaped LO pulse is almost unity. Degradation of squeezing can be attributed to uncorrelated noise for squeezed states, which is induced by propagation loss in waveguide, loss of optical components, and imperfect quantum efficiencies of detectors.
2. Experimental setup
Figure 1 shows a schematic diagram of experimental setup for squeezing measurement using the shaped LO pulse. The light source is a passively Q-switched erbium doped glass laser generating pulses at a repetition rate of 5.3 kHz, centered wavelength at 1.535 µm, with pulse duration of 3.7 ns. We classify the setup into 3 parts which are denoted by SHG, Shaping OPA and Squeezing OPA in Fig. 1. Except for the part of Shaping OPA, the setup is similar with that in Ref. . In SHG part, second harmonic pulse (SHP) at wavelength of 767nm is generated in the PPLN-WG (PPLN1) and is employed to pump the following OPAs. In Shaping OPA part, fundamental-wave pulse (FWP) with horizontal polarization is parametrically amplified and shaped in the PPLN-WG (PPLNLO). The shaped FWP is employed as the LO for homodyne detection. The relative phase between FWP and pump pulse is adjusted by angle-tuning of two BK7 plates. In Squeezing OPA part, the shaped FWP, whose polarization is changed into vertical polarization by HWP2 and DPBS2, is injected into the PPLN-WG (PPLN2). Squeezed pulse with horizontal polarization is generated in PPLN2.
3. Experimental results and discussions
3.1. Pulse shaping
To shorten the temporal duration of FWP, which will be used as a LO, we parametrically amplified FWP. Parametric gain in the center portion of FWP is larger than in the tail portion, because parametric gain depends on the intensity of pump pulse. Hence the temporal duration of FWP amplified by OPA process becomes narrower than that of the non-amplified FWP. Figure 2(a) shows the temporal shape of the FWP amplified and shaped by OPA process in PPLNLO (blue envelope) and FWP before PPLNLO (black envelope). In this measurement, the average power of the shaped FWP (output average power is 0.20 µW) is enough smaller than the pump power (0.37 mW), thus pump pulse is not depleted by parametric process in PPLNLO. The amplification gain of the OPA in PPLNLO is +20.2 dB in the average power. The temporal duration of pump pulse is almost the same level as that of input FWP because of high conversion efficiency of SHP in PPLN1 [15, 16]. As was expected, it clearly shows that the temporal duration of the shaped FWP becomes narrower than that of input FWP. The temporal duration ratio of the shaped FWP to the input FWP is 0.45. These temporal durations were calculated from the Gaussian fit. The duration ratio of FWP pulse (the shaped FWP/the input FWP) and the amplification gain of average power versus the output pump average power from PPLNLO are shown in Fig. 2(b). The duration of the shaped FWP is decreased with increasing the pump intensity.
The noise performance of the homodyne detection was measured employing the shaped FWP as LO. If the measured noise is consistent with theoretical shot noise level (SNL), we can accurately confirm the magnitude of the observed squeezing. Figure 3 shows the electron number deviation as a function of the incident electron number per pulse (nLO), where pump average power from PPLNLO was 0.35 mW. The vertical axis was calibrated using the gain factor (voltage/electron) of the homodyne detector. The detector noise corresponds 2.3×102 electrons per pulse when LO intensity is zero. Data in Fig. 3 is the value obtained by subtracting the detector noise. The solid curve, which indicates theoretical SNL curve , is in a very good agreement with the measured values.
3.2. Measurements of classical parametric gain
We investigated the classical parametric gain in PPLN2 using both the shaped pulse and non-shaped pulse as probe. To do this, the shaped probe pulse with horizontal polarization is injected into PPLN2 by removing the HWP2 and the replacement of DPBS2. The replaced DPBS2 is configured to transmit horizontally polarized light at 1.535 µm. Parametric gain of the non-shaped pulse was measured by removing the Shaping OPA part . Figure 4(a) shows the classical parametric gain versus output pump average power from PPLN2. The blue cross marks and red filled circles show the classical parametric gain of average power using the shaped and the non-shaped pulse, respectively. The usable pump power for the shaped pulse is less than that for non-shaped pulse due to the propagation and the coupling losses in PPLNLO (total loss is typically 4 dB). Maximum deamplification of the shaped pulse reached -12.1 dB, which was improved from -8.9 dB when non-shaped pulse was used as probe, at the maximum usable pump power of 0.15 mW. It clearly shows that the shaped probe pulse can be more efficiently deamplified than non-shaped probe pulse. On the other hand, the amplification gain of shaped pulse is slightly larger than that of non-shaped pulse for the same average pump power.
We estimate a mode-matching efficiency between pump pulse and probe pulse, ηcm, using the following equation:
where ηcm means the fraction of overlap between probe pulse and pump pulse, and G ± are the amplification (subscript +) and deamplification gain (-) of the measured average power. G ± is equal to exp(±2r) in the perfect mode-matching (ηcm=1), so exp(±2r) and r are defined as the intrinsic parametric gain and squeezing parameter. ηcm and r are estimated by substituting the observed G + and G - into Eq. (1). The calculated ηcm and r versus the square root of pump average power are shown in Fig. 4(b). The average value of ηcm for the shaped pulse is 0.97±0.01, and are clearly larger than 0.86±0.02 for non-shaped pulse.
3.3. Measurements of squeezing
We measured squeezed state generated from PPLN2 using the shaped LO pulse. Figure 5 shows the measured quantum noise level as the relative phase between pump pulse and LO pulse was discontinuously scanned. In this measurement, output pump average power from PPLN2 was 0.14mW. Pump power from PPLNLO is 0.41mW and amplification gain in PPLNLO is +22.0 dB. The LO intensity is 1.6×107 electrons per pulse. The quantum efficiencies of detectors were measured to be 0.84 and 0.83. The measured squeezing level is -4.1 dB below the SNL and the antisqueezing level is +11.5 dB above the SNL.
The noise levels measured using shaped (blue cross marks) and non-shaped LO (red filled circles) for several pump powers are shown in Fig. 6(a). Squeezing measurements using non-shaped LO are performed by removing Shaping OPA part, and the details of this measurement and analysis are described in Ref. . When non-shaped pulse was used as LO, -3.3 dB squeezing was obtained. In this measurement, the quantum efficiencies of detectors are 0.81 and 0.79. Even if the quantum efficiency is 0.84, the degree of squeezing is increased up to -3.5 dB which is inferior to the degree of squeezing with the shaped LO.
We estimate an overall detection efficiency, η, by substituting the measured antisqueezing, S+ and squeezing level, S- for the following equation:
The calculated η and r are shown in Fig. 6(b). Average values of η for the shaped LO and the non-shaped LO experiment are 0.65±0.01 and 0.52±0.02, respectively. The overall detection efficiency is represented by the degree of mode-matching between squeezed pulse and the LO pulse, ηqm, waveguide transmissivity, ηwaveguide, transmissivity of optical components, ηol, and quantum efficiency of detector, ηPD,
From the analysis in Ref. [15, 16] and ηol=0.96, ηwaveguide is estimated to be 0.79±0.01 (1.2±0.1 dB/cm). Therefore, ηqm=1.02±0.03 and 0.86±0.04 calculated from Eq. (3) for the shaped LO experiment and for the non-shaped LO experiment, respectively. The values of ηqm for the both experiments are almost equal to the values of ηcm. The enhancement from ηqm=0.86±0.04 to 1.02±0.03 is induced by the temporal effect, because it is expected the spatial mode-matching is so high by using waveguide in both experiments. In addition it is shown that, in the shaped LO experiment, the cause of degradation of squeezing can be explained by only the propagation loss in waveguide, loss of optical components, and imperfect detection efficiency since the mode-matching efficiency is near unity. If improvement in the waveguide loss (smaller than 0.3 dB/cm) and detector efficiencies (higher than 0.95) are to be made, a benchmark squeezing factor of 10 dB  would be feasible. One would be able to obtain reliable and dramatically compact source of broadband squeezed light by combination of the pulse-shaping technique and waveguide integration technology . Such source is useful for practical applications of CVQC protocols requiring lots of squeezed light sources, e.g., quantum teleportation network .
In summary, we reported the observation of squeezed light at telecommunication wavelength by single-pass OPA in the PPLN-WG. Both the pulse temporally shaped by the OPA and the non-shaped pulse were employed as LO pulses. We experimentally showed the use of shaped pulse allows us the efficient observation of squeezing. Squeezing of −4.1 dB and classical deamplification of −12.1 dB were observed using the temporally shaped pulse as LO and probe, respectively. We believe that these results will be useful for practical CVQC using waveguides.
One of the authors (Y. E.) is supported by the Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists.
References and links
1. S. L. Braunstein and P. van Loock, “Quantum information with continuous variables,” Rev. Mod. Phys. 77, 513–577 (2005). [CrossRef]
3. Y. Takahashi, J. Söderholm, K. Hirano, N. Namekata, S. Machida, S. Mori, S. Kurimura, S. Komatsu, and S. Inoue, “Effects of dispersion on squeezing and photon statistics of down-converted light,” Phys. Rev. A 77, 043801 (2008). [CrossRef]
4. R. Okubo, M. Hirano, Y. Zhang, and T. Hirano, “Pulse-resolved measurement of quadrature phase amplitudes of squeezed pulse trains at a repetition rate of 76 MHz,” Opt. Lett. 33, 1458–1460, (2008). [CrossRef] [PubMed]
5. H. Vahlbruch, M. Mehmet, S. Chelkowski, B. Hage, A. Franzen, N. Lastzka, S. Goßler, K. Danzmann, and R. Schnabel, “Observation of squeezed light with 10-dB quantum-noise reduction,” Phys. Rev. Lett. 100, 033602 (2008). [CrossRef] [PubMed]
11. M. G. Raymer, P. D. Drummond, and S. J. Carter, “Limits to wideband pulsed squeezing in a traveling-wave parametric amplifier with group-velocity dispersion,” Opt. Lett. 16, 1189–1191 (1991). [CrossRef] [PubMed]
16. Y. Eto, T. Tajima, Y. Zhang, and T. Hirano, “Pulsed homodyne detection of squeezed light at telecommunication wavelength,” Jpn. J. Appl. Phys. , Part 2 45, L821–L823 (2006). [CrossRef]
17. G. S. Kanter, P. Kumar, R. V. Roussev, J. Kurz, K. R. Parameswaran, and M. M. Fejer, “Squeezing in a LiNbO3 integrated optical waveguide circuit,” Opt. Express 10, 177–182 (2002). [PubMed]