We experimentally examine the noise properties of a two-pump optical parametric amplifier when converting frequencies using the Bragg-scattering (BS) and phase-conjugation (PC) processes. Using co-polarized pumps and signal, we show that the noise performance is limited by spontaneous Raman scattering. The noise performance of BS is superior to that of PC, and should improve with larger frequency excursions.
©2006 Optical Society of America
Optical parametric amplifiers (OPAs), based on four-wave mixing (FWM), can be used to perform a wide variety of functions in fiber-optic communication systems [1, 2]. These functions include amplification, frequency conversion (FC), phase conjugation, and switching on a bit-by-bit or packet basis. However, amplification is inevitably accompanied by noise . An OPA driven by two pump waves involves four product waves that are coupled by what may be characterized as three distinct FWM processes: Modulation instability (MI), phase conjugation (PC), and Bragg scattering (BS) . In MI and PC, pump photons are destroyed in pairs and sideband (signal and idler) photons are produced in pairs. This enables signal amplification and idler generation. However, it also enables the vacuum fluctuations at the frequencies of the signal and idler to be amplified, producing excess noise. In BS, for each idler photon that is created, a signal photon is destroyed. Power is transferred from the signal to the idler. Because the total sideband power is constant, the vacuum fluctuations are not amplified and no excess noise is produced. Recently, it was suggested that this feature might make BS suitable for FC in few-photon systems .
FC of many-photon signals utilizing the BS process was demonstrated over a decade ago , but the noise properties were not examined. In this letter we report the demonstration of low-noise frequency conversion using the BS process. Using co-polarized pumps and signal in a highly nonlinear fiber (HNF), we achieve a wavelength translation of 32 nm, comparable to the wavelength shifts observed in a recent experiment . Noise performance is currently limited by spontaneous Raman scattering (SRS) of the pump light [8–10]. However, the optical signal-to-noise ratio (OSNR) of the translated signal is found to be better than that of a PC signal, and should improve with increases in the pump-signal frequency separation, and with the use of other OPA polarization configurations.
The apparatus used to demonstrate frequency translation via BS is shown in Fig. 1. A two-pump OPA configuration was used. The pumps were two tunable lasers that were separately amplified and filtered, and then combined in a wavelength-division-multiplexing (WDM) coupler. Stimulated Brillouin scattering (SBS) suppression was achieved using a phase-modulation technique . The pumps were then separated, individually amplified and filtered, and recombined in a WDM coupler. Finally, the signal, also from a tunable laser, was combined with the pumps in a 10% coupler, and launched into a 1-km HNF. The output of the second port of the coupler was passed through a polarizing beamsplitter (polarizer) for monitoring on an optical spectrum analyzer (OSA) to insure that the pumps and signal were co-polarized.
The nonlinearity coefficient γ of the HNF was about 17/km-W, and the zero-dispersion wavelength, λ0, was about 1577 nm. The loss, including splices and connectors, was 1.3 dB. For efficient BS, the signal and pump 1 are placed symmetrically about λ0. Due to several constraints, such as amplifier response and WDM coupler passbands, a pump wavelength of 1566 nm and a signal wavelength of 1588 nm were chosen. Pump 2 was placed at 1598 nm. Consequently, the idler produced by BS was generated at about 1556 nm, resulting in a 32-nm (4-THz) translation of the input signal.
The signal was first transmitted through the HNF without the pumps. The power launched into the HNF, measured after the 10% coupler, was -12 dBm, and the power at the output of the HNF was -13.3 dBm. The OSNR of the signal was over 60 dB, measured in a 0.1-nm resolution bandwidth. The optical spectrum at the output of the HNF, taken with a resolution bandwidth of 0.5 nm, is shown in Fig. 2 for two pump scenarios. First, the signal and pump 1, symmetrically placed with respect to the dispersion zero, were transmitted. Second, the signal and both pumps were transmitted. The traces were taken with the pump powers that were found to yield maximum extinction of the signal (and therefore, theoretically, provide maximum transference of power to the idler). About 190 mW (+22.8 dBm) was launched into the HNF at each pump wavelength. Adding pump 1 alone at 1566 nm had almost no impact on the signal power. However, as shown in the second trace, when both pumps were present, signal extinction of about 22 dB was obtained as the signal power was transferred to the idler at 1556 nm. A PC signal at about 1576 nm is also present in the second trace, even though the efficiency of the PC process is reduced by poor phase matching. Moving the second pump to a longer wavelength, and thereby increasing the wavelength shift between the signal and the BS idler, can further reduce the PC idler. Accurate OSNR measurements cannot be directly obtained from the traces, due to a small amount of pump-light scattering in the OSA. OSNR measurements of the idler were made by placing a 3-nm-wide optical bandpass filter (centered at 1556 nm) in front of the OSA to reduce greatly the pump light entering the OSA. The idler OSNR at the optimal pump power was 45.0 dB (measured in a 0.2-nm resolution bandwidth, and normalized to 0.1 nm).
Figure 3 shows the extinction of the signal and growth of the idler with increasing pump power. The signal is maximally extinguished at the optimal pump power. As the pump power is increased beyond this point, the signal begins to grow again as power is transferred back from the idler .
The limiting noise source for the BS process in an OPA is SRS of the pump light [5, 8–10]. The amount of Raman noise is dependent on the wavelength offset from the pump, and also dependent on polarization [12–14]. We measured the noise produced by launching pump 1 alone into the HNF (the pump wavelength was in the normal dispersion regime). A tunable 0.6-nm bandpass filter and a polarizer were placed after the HNF, and the output was monitored on the OSA using a 0.2-nm resolution bandwidth. The polarizer was adjusted for both maximal and minimal readings at wavelengths from 1520 nm to 1620 nm (corresponding to polarizations that were parallel or orthogonal to the pump polarization). The results are shown in Fig. 4.
The noise at our idler wavelength of 1556 nm varies by about 2 dB with polarization. Of course, additional Raman noise at the signal wavelength of 1588 nm is generated by both pump 1 and pump 2, and some of this is transferred to the idler along with the signal power.
These noise sources are also polarization dependent. We confirmed the effect of the partially polarized noise by observing the idler OSNR through the polarizer (adjusted for maximal idler power). If the noise were not polarization dependent, the OSNR would be expected to increase by 3 dB when the polarizer was present. Instead, the OSNR increased by only 2 dB, to 47.0 dB, demonstrating the effect of partially polarized noise. The results are summarized in Table 1.
In addition to measuring the noise produced by a pump at 1566 nm, we also measured the noise produced by launching a single pump at other wavelengths. In particular, we measured the noise for a pump at 1578 nm (about 1 nm from λ0) and a pump at 1590 nm (in the anomalous dispersion regime). Figure 5 summarizes the measurements for all three wavelengths. In order to facilitate a fair comparison, the results are plotted in frequency offset from the pump frequency in each case. For reference, a frequency offset of 1 THz corresponds to a wavelength offset of about -8 nm. The combination of fiber nonlinearity and chromatic dispersion causes the total parallel noise to differ dramatically in the three cases, as seen in Fig. 5(a). The orthogonal noise, however, is almost identical within experimental error, as seen in Fig. 5(b). The difference in parallel and orthogonal noise levels can have important consequences in the performance of particular BS implementations, as will be discussed later.
We next investigated the noise performance of a PC signal. In order to optimize the performance for PC, pump 1 was moved to 1556 nm to be symmetric about λ0 with pump 2. The idler was then generated at 1566 nm (a wavelength shift of 22 nm relative to the signal). Idler OSNR measurements were taken at unity gain and at a gain of about 10 dB, both without and with a polarizer. The results are also shown in Table 1. At a gain of 10 dB, the noise is highly polarized, and therefore the polarizer has little effect on the OSNR (≅0.1 dB). The OSNR is about 2 dB (4 dB) worse than for the BS case without (with) the polarizer. At unity gain, the OSNR values for PC are slightly worse, and a small polarization dependence is observed.
Finally, we measured the OSNR performance of a low-noise erbium-doped fiber amplifier (EDFA). The EDFA had a small-signal gain of 20 dB and a noise figure of about 4 dB. Using an input power of -12 dBm, we obtained OSNR measurements of 41.5 dB (44.7 dB) without (with) a polarizer. For completeness, the results are also shown in Table 1. Note that for an ideal EDFA without a polarizer and with a noise figure of 3 dB, the output OSNR is approximately given by: OSNR (dB)=55 dB+input power (dBm) . For our input power of -12 dBm, this would produce an output OSNR of 43 dB (46 dB) without (with) a polarizer. Thus, even in the ideal case, amplification results in an OSNR that is worse than the OSNR we obtained using BS.
The noise floor in the BS process arises from SRS of the pump light, which has a bandwidth of about ±100 nm. The Raman noise orthogonal to a pump is less than the parallel component. There are BS configurations in which the polarizations of the signal and pumps can be judiciously arranged to take advantage of this lower noise . For example, the launch state of the signal can be aligned orthogonal to the two pumps. In this case, the BS idler is also generated orthogonal to the pumps. The parallel pump noise can then be removed with a polarizer. Although such a configuration is less efficient , it may be that it can further improve our OSNR results when using a polarizer. Also, as technology improves to allow translations well beyond 100 nm, SRS should no longer impose a noise floor on the BS process.
We have experimentally investigated the performance of an optical parametric amplifier when used in the Bragg-scattering (BS) mode to translate optical frequencies. We have demonstrated noise performance better than that achieved with phase conjugation, and limited in the present setup by spontaneous Raman scattering (SRS). We expect the BS noise performance to improve as advances in the manufacturing of highly nonlinear fiber allow larger frequency excursions.
References and links
1. J. Hansryd, P. A. Andrekson, M. Westlund, J. Li, and P. O. Hedekvist, “Fiber-based optical parametric amplifiers and their applications,” IEEE J. Sel. Top. Quantum Electron. 8, 506–520 (2002). [CrossRef]
2. S. Radic and C. J. McKinstrie, “Optical amplification and signal processing in highly nonlinear optical fiber,” IEICE Trans. Electron. E88C, 859–869 (2005).
3. W. H. Louisell, Radiation and Noise in Quantum Electronics (McGraw-Hill, 1964).
6. K. Inoue, “Tunable and selective wavelength conversion using fiber four-wave mixing with two pump lights,” IEEE Photon. Technol. Lett. 6, 1451–1453 (1994). [CrossRef]
7. T. Tanemura, C. S. Goh, K. Kikuchi, and S. Y. Set, “Highly efficient arbitrary wavelength conversion within entire C-band based on nondegenerate fiber four-wave mixing,” IEEE Photon. Technol. Lett. 16, 551–553 (2004). [CrossRef]
8. J. Mostowski and M. G. Raymer, “Quantum statistics in nonlinear optics,” in Contemporary Nonlinear Optics, G. P. Agrawal and R. W. Boyd, eds., (Academic Press, 1992).
10. R. Tang, P. L. Voss, J. Lasri, P. Devgan, and P. Kumar, “Noise-figure limit of fiber-optical parametric amplifiers and wavelength convertors: experimental investigation,” Opt. Lett. 29, 2372–2374 (2004). [CrossRef] [PubMed]
11. S. Radic, R. M. Jopson, A. Gnauck, C. J. McKinstrie, J. C. Centanni, and A. R. Chraplyvy, “Stimulated-Brillouin-scattering suppression using a single modulator in two-pump parametric architectures,” in Proc. OFC 2005, Anaheim, CA, paper OWN5 (2005).
12. R. H. Stolen, “Polarization effects in fiber Raman and Brillouin lasers,” IEEE J. Quantum Electron. 15, 1157–1160 (1979). [CrossRef]
14. X. Li, P. L. Voss, J. Chen, K. F. Lee, and P. Kumar, “Measurement of co- and cross-polarized Raman spectra in silica fiber for small detunings,” Opt. Express 13, 2236–2244 (2005). [CrossRef] [PubMed]
15. J. L. Zyskind, J. A. Nagel, and H. D. Kidorf, “Erbium-doped fiber amplifiers for optical communications,” in Optical Telecommunications IIIBI. P. Kaminow and T. L. Koch, eds., (Academic Press, 1997).