NONLINEAR OPTICS: PPLN upconversion advances single-photon detection

Long-wavelength pumped, periodically poled lithium niobate waveguides upconvert single photons, allowing efficient, low-noise silicon avalanche photodiodes to detect the signal.

May 1st, 2012

Long-wavelength pumped, periodically poled lithium niobate waveguides upconvert single photons, allowing efficient, low-noise silicon avalanche photodiodes to detect the signal.

Quantum-key distribution (QKD), which allows the production of random cryptographic keys known only to the sender and receiver, is the first successful and widely used application of quantum mechanics to cryptography. The technique requires the detection of single photons, usually transmitted over optical fiber.

Reaching the highest data rates and transmission distances calls for the use of wavelengths in the 1.5 μm region to match the low-loss window of telecommunications fibers. However, such wavelengths are long enough that highly efficient, low-noise silicon (Si) avalanche photodiodes (APDs) cannot be used; instead, QKD systems rely on APDs made of III-V semiconductors such as indium gallium arsenide/indium phosphide, which have lower efficiency and much higher noise.

A group at Stanford University (Palo Alto, CA), along with its collaborators at the National Institute of Standards and Technology (Gaithersburg, MD), the USTC Shanghai Institute for Advanced Studies (Shanghai, China), and the University of Science and Technology of China (Hefei, China) are remedying this problem by developing upconversion-based single-photon detectors that use periodically poled lithium niobate (PPLN) nonlinear crystals to allow signal transmission at 1.5 μm and signal detection in the 0.8 μm region by Si APDs.1 The researchers have even developed a two-stage frequency upconverter that produces a signal in the green region of the visible spectrum.2

Single-stage upconversion

For upconversion to the 0.8 μm region, the researchers use a PPLN waveguide pumped with a monolithic PPLN optical parametric oscillator (OPO) emitting at around 1850 nm (see Fig. 1). A pump wavelength longer, rather than shorter, than the signal wavelength was chosen because it drastically reduces signal noise that could arise from spontaneous parametric downconversion (SPDC) and spontaneous Raman scattering (SRS) of the pump light.

FIGURE 1. A frequency upconverter is based on a PPLN waveguide. (Abbreviations: PC, polarization controller; LPF, longpass filter; BPF, bandpass filter; PH, pinhole; TmDFA, thulium-doped fiber amplifier)

The wavelength of the OPO pump source was tunable so the researchers could characterize the noise-count rate (NCR) of the system as a function of wavelength. The fiber-pigtailed, temperature-controlled PPLN waveguides were 52 mm long with an 18.8 μm poling period and a propagation loss of less than 0.2 dB cm-1. The block of PPLN for the OPO cavity was 52 × 5 × 1 mm, had off-axis spherical facets, and was pumped by an Nd:YAG laser at 1064 nm. The OPO was temperature tuned to produce a wavelength adjustable between 1760 and 1940 nm.

A 1550 nm single-photon-level signal emitted by an attenuated external-cavity laser diode was combined with the pump light from the OPO with a wavelength-division multiplexer (WDM) and sent into the PPLN waveguide. The upconverted light was separated from the pump and signal wavelengths in the waveguide's output with a dichroic mirror and then further filtered using a prism and bandpass filter before being collected by a single-photon-counting module (SPCM).

Conversion efficiency of the waveguide was measured as a function of pump power and signal wavelength, reaching a maximum internal conversion efficiency of 86% for light converted from 1554 to 834 nm at a pump power of 151 mW.

With the signal input (at the WDM) set to about 106 photons/s, the photon-detection efficiency (PDE) and NCR were measured for several pump wavelengths between 1796 and 1859 nm. For an 1810 nm pump, a PDE of about 37% was reached—a figure that the researchers believe they can boost to 45% with changes to the setup.

The NCRs were measured and found to be appreciable (up to about 104 counts/s); but they showed a good match with measurements of the Raman spectrum of the temperature-controlled (heated) PPLN crystal. Future experiments, in which the waveguide would be thermoelectrically cooled to around -50°C, could bring the NCR of the system down to near the intrinsic dark-count rate for a Si APD.

Cascaded upconversion

The second experiment led by the Stanford group is a cascaded upconversion scheme that uses a single PPLN waveguide incorporating two quasi-phase-matched gratings (QPMs). Again, single-photon-level signals at 1550 nm are upconverted for detection by Si APDs; however, here the final wavelength is about 570 nm, suitable for use with a low-timing-jitter APD for detection of high-clock-rate pulsed sources.

The researchers fabricated a 5.2-cm-long two-step upconverting PPLN waveguide containing two QPMs with periods of 18.55 and 8.25 μm. The input signal wavelength was 1548 nm, the pump wavelength was 1801.3 nm, and the temperature of the PPLN crystal to achieve QPM in both gratings was 76°C.

The first-step (ηint) and second-step (η2) conversion efficiencies were measured as a function of pump power; while the second-step efficiency was high, reaching almost 90% for a pump power of 300 mW, the first-step efficiency was much lower, suffering from propagation losses and a less-than-optimal value for the period of one of the gratings (see Fig. 2).

FIGURE 2. Efficiencies for the first (ηint) and second (η2) conversion steps were measured as a function of pump power for a two-stage frequency upconverter.

To demonstrate two-stage upconversion single-photon detection of a high-pulse-rate light source, the researchers chose an actively modelocked 1550 nm laser that emitted 10 ps pulses at a 10 GHz repetition rate, attenuating the signal to about 0.01 photons per pulse. With the pump power set at 100 mW, a Si APD detected the pulses at a DCR of about 850 counts/s.

By measuring the time interval between the clock output of the pulse pattern generator and the output electrical pulses of the APD, the researchers measured a timing jitter of 66.9 ps full-width at half maximum (FWHM). Taking into account the 50 ps manufacturer's specs for the laser and the fact that the time-interval analyzer used in the experiment had a jitter of 25 ps, the researchers concluded that the upconversion did not introduce any additional jitter to the signal pulse timing.


1. J.S. Pelc et al., Opt. Exp., 19, 22, 21445 (Oct. 24, 2011).
2. J.S. Pelc et al., Opt. Lett., 37, 4, 476 (Feb. 15, 2012).

Tell us what you think about this article. Send an e-mail to

More Laser Focus World Current Issue Articles
More Laser Focus World Archives Issue Articles

More in Detectors & Imaging