We report the observation of Zeno switching through an inverse Raman scattering (IRS) process in an optical fiber. In IRS, light at the anti-Stokes frequency is strongly attenuated in the presence of a pump field, allowing it to be used for all-optical switching and modulation. Our observed level of induced absorption via IRS in the optical fiber is > 20dB in a time scale of less than 5 ps. The full Raman response spectrum was extracted experimentally and excellent agreement was found between the experimental data and theoretical modeling of IRS.
© 2011 OSA
As the demand for higher bandwidth grows, optical switching systems with significantly faster operating speeds and reduced latency will become essential for information processing and communication. All-optical switching, wherein light is used to control light, is considered as an attractive alternative to electrical switching provided that absorptive losses can be minimized. Various all-optical switching schemes including the use of nonlinear directional couplers, optical tuning of photonic crystal cavities or coupled microring resonators, and phase based switching in optical fibers have been reported [1–6]. While each of these schemes has particular advantages, achieving low optical loss, low switching energies and high contrast simultaneously continues to be a major challenge. We report on a new type of all-optical switching, here referred to Zeno switching, with exceptional performance. Our Zeno switch is based on the use of inverse Raman scattering (IRS) in a GeO2 doped silica optical fiber, where a signal beam is switched via optical loss induced by a pump beam [7,8]. The switch has the advantages of minimal optical loss (< 0.02 dB), low dissipation energy (~30 fJ/switch event), high speed (<1ps), high contrast (>20 dB) and broad optical bandwidth (100 nm).
Raman scattering is an inelastic process where there is energy exchange between photons and vibrational modes of molecules or solids (phonons), resulting in the generation of new photons at optical frequencies that are red-shifted (Stokes) and blue-shifted (anti-Stokes) from the initial pump excitation frequency . The conversion efficiency between the pump and Stokes photons in spontaneous Raman scattering is extremely weak but can become very large if the process is stimulated. In stimulated Raman scattering (SRS), the energy transfer from the strong pump field to the Stokes field can be very efficient resulting in significant optical gain. SRS has been widely used for amplification of signals in fiber optic networks and has been implemented, along with associated Raman processes, in three-dimensional multiphoton vibrational imaging [10,11]. As opposed to the growth of the Stokes field in Raman amplification, optical loss can be induced at the anti-Stokes frequency when a pump field is present through a process analogous to SRS known as inverse Raman scattering (IRS) . While both processes involve the transfer of energy from the shorter wavelength field to the longer wavelength field [12–14], the distinction between them is that in SRS the energy transferred from the strong field results in gain for the weaker field, while in IRS the energy transferred to the strong field results in loss for the weaker field. Here, we exploit optically-induced absorption based on IRS for all-optical switching.
From a quantum mechanical viewpoint, the Zeno effect refers to holding a system in a prepared quantum mechanical state by the use of repeated measurements; without such measurements the system changes state. The Zeno effect has been proposed for all-optical switching based on electronically-induced two-photon absorption (2PA) in both a rubidium atomic vapor  and in CdSe-based semiconductor quantum dots  as well as for second order wave mixing processes in waveguides , but, to date, no experimental demonstrations of the effect have been reported (the exception being a preliminary report published in  highlighting the work herein). During SRS (and similarly in IRS) coherent excitation of phonons occurs in the medium through mixing of the pump and Stokes fields, which, in turn, leads to generation of more Stokes photons. If the pump field were turned off, this coherent excitation would naturally evolve to a state with no phase coherence following vibrational dephasing. In this analogy, the pump excitation serves as both the method for measurement and the coherent excitation of the medium into the prepared state.
There are a number of unique practical benefits for IRS-based Zeno switching. Since the frequency difference between the pump and signal (i.e. anti-Stokes) fields must simply match the targeted vibrational resonance frequency, there is great flexibility in the choice of this pair of frequencies. Thus, the pump and signal wavelengths can be judiciously chosen such that minimal optical loss is experienced by either field. Furthermore, in IRS the dissipation energy per switching event is proportional to the difference in energy between the pump and signal photons, whereas in 2PA the sum of the photon energies is dissipated. Finally, since energy is transferred from the signal to the pump in IRS, the pump is conveniently amplified during the switching event, which makes the prospect of pump recycling a possibility. Recently, IRS was observed in a silicon waveguide . However, operation in the telecommunications wavelength bands led to 2PA of the pump beam along with subsequent free-carrier absorption. These mechanisms result in both parasitic optical loss and slower temporal switching response. Such issues could easily be avoided for IRS in a glass optical fiber, which to the best of our knowledge, has not been reported previously.
In this paper, we report on the observation of a Zeno switch based on IRS in a 2 meter long glass optical fiber using an optical signal in the presence of a pump beam. Strong, induced attenuation of the signal is observed with minimal energy dissipation per switching event and virtually no optical loss when either the pump or the signal is present separately. We obtain broad signal spectrum coverage giving rise to an optical bandwidth of ~100 nm, which is among the broadest reported for all-optical switches. Furthermore, due to the ultrafast temporal Raman response of the optical fiber (< 10 fs), a switching bandwidth larger than 100 THz is possible. We further use the observed IRS to extract the Raman response spectrum for the optical fiber. Thus, Raman spectroscopic data for optical fibers can also be obtained using this experimental technique.
2. Experimental methods
The experimental setup for the observation of IRS in optical fiber is shown in Fig. 1 . A mode-locked fiber oscillator operating near 1560 nm was employed, similar to the one reported in . The output of the oscillator was then amplified in an Er-doped fiber amplifier (EDFA). 80% of the power after the amplifier was used to generate a broad supercontinuum in a short piece of highly nonlinear fiber (HNLF) , which was used as the anti-Stokes signal beam. The remaining 20% of the power after the amplifier was passed through a narrow (~1.5 nm FWHM) band-pass filter (working around 1560 nm) to reduce the bandwidth of the pulses. These narrow bandwidth picosecond (~3 ps) pulses passed through a variable delay line, were subsequently amplified using an additional EDFA, and were used as the pump beam in the experiment. The pump and the anti-Stokes signal pulses (pulse durations ~3 ps) were combined using a fiber pigtailed WDM thin film coupler. The beams were then coupled into a short segment of an optical fiber under test (FUT). The average power of the anti-Stokes beam was about 5 mW. The power of the pump beam could be increased up to ~50 mW by adjusting the 976 nm pump power to the corresponding EDFA. The repetition rate of both the pump and the signal was 50 MHz. Two polarization controllers were used to independently optimize the polarization states of the pump and signal beams.The narrowband picosecond pump and broadband anti-Stokes signal pulses were then launched into a 2 m long FUT. Two FUTs were studied: a common single mode fiber (SMF 28, 8.3 μm diameter) and a small core, high numerical aperture optical fiber (Nufern, UHNA7, ~3 μm diameter). When the variable time delay was adjusted such that the pump and signal pulses were overlapped in time, significant loss was observed in the anti-Stokes spectral range of the signal.
In order to observe modulation in the time domain data, the pump and signal pulse trains were coupled out of the FUT using a collimator. The collimated beam was sent through a transmission grating to separate the spectral components in space. Two independent fast detectors (<1 ns response time) were used to measure the pump (around 1550 nm) and signal intensities at the same time. A small portion of the signal spectrum (~10 nm) around 1460 nm was captured through the use of an aperture. The signals from the fast detectors were acquired using a fast oscilloscope (6 GHz bandwidth).
3. Results and discussion
Typical measured IRS spectra from the high numerical aperture optical fiber (UHNA7) are shown in Fig. 2 ; the anti-Stokes signal experiences increasing loss as the pump power is increased. At ~50 mW average pump power the maximal loss observed is ~22 dB, exceeding the 20 dB contrast ordinarily required for communications applications. It is worth mentioning that this effect was also observed for several different types of optical fibers (SMF 28, for example). However, the level of IRS induced loss largely depended on the fiber used. The strongest induced loss was observed in the aforementioned UHNA7 small core optical fiber, which possesses a high level of GeO2 doping (30 mole percent) (Fig. 2(a)). For comparison, the largest induced loss in the SMF 28 at 50 mW pump power was only ~3 dB. It is understandable that a nonlinear optical process such as IRS should exhibit a stronger response for fibers with smaller core diameters given the corresponding increase in modal intensity. Furthermore, GeO2 is known to possess a large Raman cross-section [20,21]; given that the nonlinear loss coefficient associated with IRS is proportional to this cross-section, this is consistent with the observation of stronger IRS in these types of fibers.
Time domain data were also acquired to further confirm the observation of IRS switching in the spectral domain. Figure 3 shows oscilloscope traces when the pump and signal pulses were separated in time by 5 ps (set by the delay line, Fig. 3(a)) and when they were perfectly overlapped in time (Fig. 3(b)). The average power of the pump was ~50 mW. It is clear that the signal beam could be switched on and off, with very high extinction ratio, using IRS in the optical fiber. Furthermore, the pump power was modulated by directly modulating the drive current to the corresponding EDFA. This modulation was transferred to the signal pulse train as shown in Fig. 3(c) as expected. Due to the long excited state lifetime of the Er-doped fiber in the EDFA, a modulation frequency of ~100 Hz was used for this demonstration. In the absence of the pump beam, the propagation loss in the 2 m fiber at the signal wavelength (1460 nm) was negligible at the level of +/− 0.1 dB (i.e. loss <0.05 dB/m) with similar results for the pump beam at 1550 nm. The actual losses at these wavelengths are expected to be in the range of 10 dB/km or less, but much longer lengths of fiber would be required to confirm this. The energy dissipation experienced in this IRS process is a product of the energy difference between the pump and signal photons (49 meV) and the number of signal photons absorbed that is estimated via Ns = Is/(fsEs)·δλ/Δλ. Here, Is is the average signal power (5 mW), fs the repetition rate (50 MHz), Es the signal photon energy (0.85 eV), δλ the width of the central signal absorption region (~10 nm), and Δλ the width of the supercontinuum (~1500 nm), resulting into Ns = 4.9x106 absorbed signal photons. This translates to ~30 fJ dissipation per switching event.
The Raman loss spectrum of the FUT was extracted from the above spectral measurements by subtracting the anti-Stokes spectrum at negligible pump power from those at higher pump powers as shown in Fig. 4 . The response function in the time domain can be obtained from the IRS loss spectra of Fig. 4 using the technique described in Ref. 22. Qualitatively, the Raman loss spectra of the Ge-doped fiber (Fig. 4(a)) and SMF28 (Fig. 4(b)) give Raman response functions that are quite similar to published data for Ge-doped and fused silica glasses [23–25]. It is interesting to note that we observe a narrow Raman peak at around 220 cm−1 (distorted at high pump power due to nonlinear spectral broadening of the pump, Fig. 4(a)). This peak is absent in the reported Raman spectra of highly Ge-doped silica fibers [20,21] while it has been observed previously in pure GeO2 crystals . This peak may be attributed to a symmetric bending vibrational mode of trigonal GeO2 groups in the fiber .
Numerical simulations of IRS in optical fiber were performed to provide a better understanding of the observed results (see Numerical Methods below). The theoretical description is based on the coupled nonlinear Schrödinger equations for the anti-Stokes signal and pump pulses propagating through an optical fiber. In the slowly varying envelope approximation, these equations are given by [27,28]. The calculated results shown in Fig. 2(b) are in very good agreement with the measured data presented in Fig. 2(a). In particular, the increased attenuation with increasing pump power can clearly be seen. Furthermore, the shape of the calculated IRS loss spectrum is similar to that of the input Raman loss spectrum used in the calculations. The maximal induced loss of 20 dB, at pump powers corresponding to the experimental values, is accurately reproduced by the theory. This is further verified in Fig. 5(a) which shows the dependence of the attenuation level on the peak pump power. It should be noted that the experimental observation of the decreased IRS-induced loss at high peak pump powers is due to a competing coherent anti-Stokes Raman scattering process [14,17]. We have also confirmed that a small Stokes signal did appear at around 1.66μm at the highest pump power level (~50mW). However, this is not an issue at the low pump powers needed to achieve high contrast switching. It is clear from both the measured data and calculated results that there is no threshold involved. In this way, IRS is different from SRS, the latter occurring only above a particular threshold where the gain due to stimulated emission overcomes the propagation losses. These calculations also indicate (not shown) that the fractional induced absorption from the supercontinuum signal does not depend on the total power of the signal. The properties of IRS in a glass optical fiber are thus similar to those observed in several liquids . The dependence of the maximal IRS-induced loss on the length of the FUT was also investigated both experimentally and theoretically (Fig. 5(b)). The pump power was fixed at ~50 mW in this case. First, the IRS signal was observed to grow rapidly when the pump and signal pulses propagated along the initial couple meters of the FUT. The loss level then is saturated at longer lengths, since pump and signal pulses propagate with different group velocities inside the fiber, leading to reduced temporal overlap between the pulses. Here again, there is excellent agreement between experiment and calculation.
In conclusion, we have demonstrated Zeno switching through IRS in an optical fiber. Our results indicate that IRS can be an effective mechanism for a Zeno-based all-optical switch. Stimulated loss of >20 dB has been achieved in 2 m of highly GeO2-doped optical fiber. Compared to silicon waveguides, IRS in optical fibers does not suffer from additional losses due to 2PA and free-carrier absorption. IRS based Zeno switching in silica-based single-mode optical fibers further benefits from great flexibility with respect to the choice of wavelengths (only the photon energy difference is critical), very low dissipation energy (< 0.1 eV per switched photon) when compared to 2PA Zeno switches (2-3 eV), and practically no off-state insertion loss. Our results suggest that IRS-based Zeno switches can provide an effective path for all-optical switching systems that would be compatible with current optical communication networks.
Appendix - numerical methods
The theoretical description is based on the coupled nonlinear Schrödinger equations for the anti-Stokes signal and pump pulses propagating through an optical fiber. In the slowly varying envelope approximation, these equations are given by [27,28]:
Here, the space-time-dependent anti-Stokes field envelope is denoted by A as(z,t) while the pump field envelope is given by Ap(z,t). In Eqs. (1) and (2), the linear losses are described by α as and αp for the signal and pump, respectively, the dispersion parameters are given by β 2 , as and β 2 ,p, and the complex nonlinear coefficients (which includes the nonlinear loss coefficient associated with IRS) are γ as and γp. The walk-off parameter is defined by d wo= 1/vg, as − 1/vg,p and describes the group-velocity mismatch between the signal and pump pulses. The Raman response enters via the integrals given in Eqs. (1) and (2). We use a standard analytical form for the Raman response function hR(t) which explicitly reads :
with time constants τ 1 and τ 2. τ1 −1 is the frequency of the phonon associated with the Raman response, while τ2 −1, is the bandwidth of the Lorentzian line used to model the frequency response. The parameter fR in Eqs. (1) and (2) denotes the fractional contribution of the Raman response to the generation of nonlinear polarization and ΩR = ω as − ωp represents the difference between the carrier frequencies of the signal and pump.
To determine accurate Raman parameters, values of τ 1 and τ 2 in Eq. (3) were chosen to match the measured Raman response spectrum that was shown in Fig. 4(a) (red curve). The values τ 1= 11.9 fs and τ 2= 65.0 fs produced a very good fit (not shown) to the measured Raman response. In the numerical calculations, these values were used along with fR= 0.18 , the value ordinarily used for silica fibers.
Using the numerical simulations for beam propagation in the fiber, the output signal spectra, following propagation through 2 m of the germanium-doped fiber, were calculated as a function of increasing pump power (Fig. 2(b)). Additionally, the spectral region around the pump is also shown. In Fig. 2(b), the linear losses are assumed to be α as= αp= 0, the dispersions were taken to be β 2 , as = 18 ps2 /km, β 2 ,p = 35 ps2 /km, and the nonlinearities to be γ as= n 2 ω as /(cA eff) = 14 W− 1 /(km), γp= n 2 ωp/(cA eff) = 13 W− 1 /(km) for a fiber diameter of 3.2 μm, an effective area A eff = 8.04 μm2, and nonlinear-index coefficient n 2= 2.6*10− 20 m2 /W. The walk-off parameter is taken to be d wo= 1.8 ps/m . There were no free parameters used in this calculation, with all values taken from Ref. 28, or from the specifications for the GeO2 doped fiber.
This work was supported by the DARPA ZOE program (Grant No. W31P4Q-09-1-0012) and the CIAN ERC (Grant No. EEC-0812072). We would like to acknowledge Alex Gaeta and Michal Lipson of Cornell University and Matt Goodman at DARPA for helpful discussions. We would like to dedicate this article to the memory of Boris P. Stoicheff who originally observed IRS in liquids in 1964.
References and links
1. S. M. Jensen, “The non-linear coherent coupler,” IEEE J. Quantum Electron. 18(10), 1580–1583 (1982). [CrossRef]
3. G. I. Stegeman and E. M. Wright, “All-optical waveguide switching,” Opt. Quantum Electron. 22(2), 95–122 (1990). [CrossRef]
4. K. Nozaki, T. Tanabe, A. Shinya, S. Matsuo, T. Sato, H. Taniyama, and M. Notomi, “Sub-femtojoule all-optical switching using a photonic crystal nanocavity,” Nat. Photonics 4(7), 477–483 (2010). [CrossRef]
5. T. Tanabe, M. Notomi, S. Mitsugi, A. Shinya, and E. Kuramochi, “All-optical switches on a silicon chip realized using photonic crystal nanocavities,” Appl. Phys. Lett. 87(15), 151112 (2005). [CrossRef]
7. B. C. Jacobs and J. D. Franson, “All-optical switching using the quantum Zeno effect and two-photon absorption,” Phys. Rev. A 79(6), 063830 (2009). [CrossRef]
8. L. Schneebeli, T. Feldtmann, M. Kira, S. W. Koch, and N. Peyghambarian, “Zeno-logic applications of semiconductor quantum dots,” Phys. Rev. A 81(5), 053852 (2010). [CrossRef]
9. C. V. Raman and K. S. Krishnan, “A new type of secondary radiation,” Nature 121(3048), 501–502 (1928). [CrossRef]
10. A. Zumbusch, G. R. Holtom, and X. S. Xie, “Three-dimensional vibrational imaging by coherent anti-Stokes Raman scattering,” Phys. Rev. Lett. 82(20), 4142–4145 (1999). [CrossRef]
11. C. W. Freudiger, W. Min, B. G. Saar, S. Lu, G. R. Holtom, C. W. He, J. C. Tsai, J. X. Kang, and X. S. Xie, “Label-free biomedical imaging with high sensitivity by stimulated Raman scattering microscopy,” Science 322(5909), 1857–1861 (2008). [CrossRef] [PubMed]
12. W. J. Jones and B. P. Stoicheff, “Inverse Raman spectra - induced absorption at optical frequencies,” Phys. Rev. Lett. 13(22), 657–659 (1964). [CrossRef]
13. Q. R. An, W. Zinth, and P. Gilch, “In situ determination of fluorescence lifetimes via inverse Raman scattering,” Opt. Commun. 202(1-3), 209–216 (2002). [CrossRef]
14. J. Stone, “Inverse Raman-scattering - continuous generation in optical fibers,” J. Chem. Phys. 69(10), 4349–4356 (1978). [CrossRef]
15. Y.-P. Huang, et al., arXiv: 1008, 2408 [quant.ph] Aug. 14 (2010).
16. K. Kieu, L. Schneebeli, R. A. Norwood, and N. Peyghambarian, “Zeno Switching Through Inverse Raman Scattering in Optical Fiber,” Opt. Photon. News 21(12), 35 (2010). [CrossRef]
17. D. R. Solli, P. Koonath, and B. Jalali, “Inverse Raman scattering in silicon: a free-carrier enhanced effect,” Phys. Rev. A 79(5), 053853 (2009). [CrossRef]
19. K. Kieu, J. Jones, and N. Peyghambarian, “Generation of few-cycle pulses from an amplified carbon nanotube mode-locked fiber laser system,” IEEE Photon. Technol. Lett. 22(20), 1521–1523 (2010). [CrossRef]
20. F. L. Galeener, J. C. Mikkelsen, R. H. Geils Jr, and W. J. Mosby, “The relative Raman cross-sections of vitreous SiO2, GeO2, B2O3 and P2O5.,” Appl. Phys. Lett. 32(1), 34–36 (1978). [CrossRef]
21. J. Bromage, K. Rottwitt, and M. E. Lines, “A method to predict the Raman gain spectra of germanosilicate fibers with arbitrary index profiles,” IEEE Photon. Technol. Lett. 14(1), 24–26 (2002). [CrossRef]
22. R. H. Stolen, J. P. Gordon, W. J. Tomlinson, and H. A. Haus, “Raman response function of silica-core fibers,” J. Opt. Soc. Am. B 6(6), 1159–1166 (1989). [CrossRef]
24. D. Mahgerefteh, D. L. Butler, J. Goldhar, B. Rosenberg, and G. L. Burdge, “Technique for measurement of the Raman gain coefficient in optical fibers,” Opt. Lett. 21(24), 2026–2028 (1996). [CrossRef] [PubMed]
25. S. Smolorz, F. Wise, and N. F. Borrelli, “Measurement of the nonlinear optical response of optical fiber materials by use of spectrally resolved two-beam coupling,” Opt. Lett. 24(16), 1103–1105 (1999). [CrossRef]
26. J. F. Scott, “Raman spectra of GeO2,” Phys. Rev. B 1(8), 3488–3493 (1970). [CrossRef]
27. C. Headley III and G. P. Agrawal, “Unified description of ultrafast stimulated Raman scattering in optical fibers,” J. Opt. Soc. Am. B 13(10), 2170–2177 (1996). [CrossRef]
28. G. P. Agrawal, Nonlinear Fiber Optics (Academic Press, 2007).
29. K. J. Blow and D. Wood, “Theoretical description of transient stimulated Raman-scattering in optical fibers,” IEEE J. Quantum Electron. 25(12), 2665–2673 (1989). [CrossRef]