Scientists working for the Defense Advanced Research Projects Agency (DARPA) are making progress toward the development of robust, integrated devices that can slow, store and process light pulses. Their research will have a far-reaching impact on high-speed optical processing, telecommunications networks and quantum information science.
All-optical data processing operations enabled by slow light. All-optical processing relies on the ability to store, switch and time delay light pulses to create and control data buffers, logic switches and tunable signal delays.
Information processing networks, such as those used in telecommunications and high-speed radio-frequency signal processing, use a physical architecture based on high-speed electrical circuits for data routing and optical fibers for transporting data traffic.
In the future, networks are anticipated to handle much larger bandwidths. Although such bandwidths will be easily transported with photons, they will be difficult to manage using electronic integrated circuit technology. The development of all-optical networks—i.e., networks with nodes devoid of optical-to-electrical converters—will allow for future throughput scalability and eliminate the noise and transmitted bit error rates that arise from converting light signals to electronic signals.
All-optical processing relies on the ability to create and control data buffers, logic switches and tunable signal delays. These operations are not simply a matter of imposing a fixed delay using fiber-optic cable; they involve the ability to store, switch and time-delay optical pulses.
Optical elements that can perform these control functions are still in their infancy. To expedite innovation, DARPA initiated Slow Light, a three-phase program, in August 2004. The program aims to produce robust, integrated devices that can slow, store and process light pulses.
The origin of slow light. The group velocity at which a pulse of light propagates through a material depends not only on the refractive index, but the dispersion (dn/dω) as well. Media with frequency-dependent gain or absorption features exhibit large dispersion, leading to a reduction of the group velocity of the propagating pulse (slow light).
In general, the velocity at which a pulse of light propagates through a medium (group velocity) is given by vgroup = c (n + ω dn/dω)-1 where c is the velocity of light in a vacuum, ω is the light’s angular frequency and n is the refractive index of the medium. The group velocity depends not only on the refractive index, but also on the dispersion (i.e., dn/dω). In most situations, the refractive index dispersion is small, so the group velocity is simply given by c/n.
However, early experiments by Lene Hau at Harvard and Steve Harris at Stanford demonstrated efficient propagation of light pulses in regimes where dn/dω was large and positive, establishing a regime where vgroup << c —in other words, light is slowed. As the pulse is slowed in the medium, nonlinear interactions are greatly enhanced due to compression of the local energy density, allowing for nonlinear processes to occur at much lower operating powers than conventionally required.
High bandwidth analog tunable delay lines
Prior to DARPA’s Slow Light program, slow light demonstrations exhibited narrow bandwidths and were not perceived as practical. In early studies, researchers used the large normal dispersion associated with electronic resonances in atomic systems such as ultra cold sodium gas, rubidium vapor, cryogenically cooled solids and room-temperature ruby. Processes such as electromagnetically induced transparency (EIT) and coherent population oscillations (CPO) were used to generate narrow transparency windows within absorption resonances.
It is now evident that slow light is not limited to these systems. Simply using a medium exhibiting high and steep dispersion—e.g., a medium with frequency-dependent absorption or gain features—will lead to a reduction in the group velocity of a propagating pulse. This condition can be fulfilled with atomic resonances as well as laser-induced amplifying resonances and optical resonances in photonic structures.
The key question is whether one can build an optical delay with enough bandwidth to be useful. Slow Light is tackling this question by exploring a wide range of new approaches for designing tunable delay lines, including fiber-based systems, semiconductor optical amplifiers, 2D photonic crystals, ring resonators and atomic vapors.
Optical fiber systems
One approach pursued by the DARPA Slow Light team exploits stimulated scattering processes commonly found in room-temperature telecom optical fibers. Ordinary nonlinear scattering processes, such as stimulated Brillouin scattering (SBS) and stimulated Raman scattering (SRS), can produce a frequency-dependent variation in the refractive index of the fiber. In general, stimulated scattering is an inelastic process in which a strong incident (pump) beam interacts with density variations in the medium, loses energy, and is converted into a scattered (Stokes) beam of lower frequency.
In SBS, the coupling is mediated via acoustic waves in the medium, whereas in SRS it is mediated via vibrational modes. The process acts as a pump-induced amplifying resonance, resulting in the exponential amplification of the Stokes field. This Stokes gain feature can be made to exhibit steep dispersion, giving rise to slow light. An advantage of this process is that the position of the feature is fixed relative to the pump frequency, allowing for control of the slow light resonance position by tuning the wavelength of the pump. Furthermore, the steepness of the dispersion, which sets the group velocity, is controlled by the pump power, making the delay easily tunable.
DARPA Slow Light researchers have successfully demonstrated fiber-based slow light delays using both SBS and SRS schemes. In the SBS-based work, Okawachi and colleagues used optical fiber with an SBS linewidth of 70 MHz to delay 63 ns pulses by 25 ns with no distortion and shorter 15 ns pulses by 20 ns with some pulse broadening (Phys. Rev. Lett. 94, 153902). Stenner et al. demonstrated that pulse delays can be further increased by using two nearby SBS gain lines to flatten the gain profile and perform dispersion management.
More dramatically, researchers recently showed that a technique involving broadening the spectrum of the SBS pump laser greatly increases the SBS linewidth to more than 12GHz, allowing the delay of 75 ps pulses by up to 47 ps (Zhu et al., 2007). This broadband-SBS technique shows promise and has recently been extended to delay 10 Gb/s phase-modulated data streams by as much as 42 ps (Zhang et al., 2007).
DARPA Slow Light tunable delay lines. Plot of delay versus bandwidth for slow-light-based tunable delay lines. Results are shown for fiber, photonic, SOA and atomic approaches pursued by DARPA Slow Light. Overlaid for reference are the performance goals for the three program phases, performance trade spaces for radar and telecom applications and constant bandwidth delay product lines.
In contrast to SBS, appreciably larger bandwidths (THz for typical fibers) can be obtained by using a stimulated Raman scattering Stokes gain feature. High-bandwidth, fiber-based delays have recently been demonstrated using this process. By taking advantage of THz Raman gain bandwidths in silica, Sharping and colleagues were able to delay short, 430 fs pulses by 370 fs (85 percent of a pulse width) with no distortion (Opt. Express 13, 6092).
Furthermore, this technique can be extended to more compact platforms. For example, SRS-based slow light has also been demonstrated in a compact (8 mm) silicon-on-insulator (SOI) planar waveguide. Using the narrower silicon Raman linewidth of 105 GHz, Okawachi and colleagues generated controllable delays as large as 4 ps for pulses as short as 3 ps in a chip-scale device (Opt. Express 14, 2317). SRS’s ability to accommodate large bandwidths makes it an attractive technique for delaying ultra-short pulses.
Parametric wavelength conversion and dispersion offers an alternative fiber-based approach to slowing light. This technique directly uses the spectral variation of the group index to generate very large delays. Conceptually, it relies on controllably shifting the wavelength of an incoming pulse and sending it through a medium with large group velocity dispersion to impose the group delay. The amount of delay is simply a function of the converted wavelength.
The pulse is then reconverted to the original wavelength, preserving both the spectral and phase information of the input signal. Slow Light researchers have demonstrated large tunable delays using several approaches for wavelength conversion. In one scheme, Sharping et al. used a four-wave mixing process to demonstrate tunable delays of 10 ps input pulses for up to 80 pulse widths (Opt. Express 13, 7872). In another, researchers delayed 350-ps output pulses by 12 pulse widths using spectral broadening via self-phase modulation and subsequent filtering (Opt. Express 14, 12022).
A recent approach uses a periodically poled lithium-niobate waveguide as a rapidly tunable converter in conjunction with a dispersion compensator to continuously delay a 10 Gb/s nonreturn-to-zero data stream up to 44 ns, corresponding to 440 bit slots (Wang et al., 2007).
Semiconductor optical amplifiers
A more compact approach uses semiconductor optical amplifiers to create optical tunable delays. Semiconductor-based devices are attractive candidates due to their small size, room temperature operation and compatibility with existing optical communication systems. Several schemes have been used to generate tunable delays. For example, using an exciton absorption resonance in GaAs quantum well structures, program researchers demonstrated 200 percent fractional delays for 8 ps input pulses (Sarkar et al., 2006).
Using four wave mixing, researchers have realized fractional delays exceeding 50 percent at 0.5 Gb/sec (Pesala et al., 2006). Most recently, scientists have achieved a record room temperature fractional delay of 250 percent for 700 fs pulses using a high-bandwidth spectral hole burning resonance (Sedgwick et al., 2007). These results are very promising and form the basis of ongoing work to use cascaded devices and pulse chirping to increase the delay further.
Resonant photonic systems
Large group delays can also be achieved by taking advantage of the dispersion properties of waveguide-based, optical resonant photonic systems. These offer the added promise of developing truly integrated devices. For example, 2D photonic crystal waveguides are modulated by a periodic array of wavelength-scale holes and can be engineered to produce a photonic band gap with large group velocity dispersion near the edge. In 2005, Vlasov and colleagues demonstrated a more-than-300-fold reduction of the group velocity in 2D photonic crystals. Furthermore, the delay was actively tuned with 100 ns response using integrated micro-heaters.
Coupled high-finesse micro-ring resonators are another example of resonant photonic systems. In this system, very low group velocities result from input light pulses spending the bulk of their time circulating within each resonator rather than propagating between resonators. Furthermore, the strong frequency dependence of the light coupling, arising from the geometric resonance, gives rise to large dispersion.
Slow Light investigators have recently demonstrated on-chip delays exceeding 500 ps using silicon-on-insulator photonic waveguides with up to 100 cascaded micro-ring resonators (Xia et al., 2007). System-level measurements show bit-error-free operation up to 5 Gb/s bit rates. The necessity of fabricating nearly identical resonators remains a challenge in these systems.
Dynamically manipulating the refractive index of such resonant structures while the pulses are in the system has been proposed as a way to not only delay but to stop light. In a complete analogue to atomic systems, Xu and colleagues have used dynamic tuning techniques to demonstrate an all-optical analogue to EIT and on-chip stopping and storing of light. Similar experiments are being pursued to observe EIT-like resonances and light storage in 2D photonic crystals (Fan et al., 2007).
Despite their sizeable footprint, atomic vapor experiments continue to serve as an excellent platform for exploring the physics of slow light. Early work involved the use of single transparency resonances and suffered from narrow bandwidths. However, recent studies in the Slow Light program have yielded appreciably larger bandwidths. As an example, Camacho and colleagues are using two closely spaced absorption resonances in hot cesium vapor to realize delays of multiple pulse widths over a wide bandwidth.
Specifically, they demonstrated the tunable delay of a 275 ps pulse by up to 25 pulse widths and the tunable delay of a 740 ps pulse by up to 80 pulse widths, with little pulse distortion (Phys. Rev. Lett. 98, 153601). It remains a significant challenge to implement atomic vapor schemes in a compact footprint suitable for integration. However, the large spatial extent of the slow light medium does have some unique advantages. In recent work, Camacho et al. have shown the tunable delay by many pulse widths of not only light pulses, but two-dimensional transverse images (Phys. Rev. Lett. 98, 043902). This technique enables new approaches for image processing.
Single photon nonlinearities
DARPA Slow Light seeks to demonstrate practical systems exhibiting ultra-low-light level response. This could enable single-photon components for applications in quantum information science (e.g., communications, cryptography and imaging). The linear coherent processes used to slow and store light can be applied from classical fields down to single photons. These processes can be made strongly nonlinear by enhancing the coupling between the material system and the light field, producing measurable, deterministic effects for single photon inputs.
Single-photon nonlinearities can help realize a source of single photons with controllable spatio-temporal characteristics, a bright source of entangled photon pairs for efficient quantum communication and cryptography, or a single photon, quantum non-demolition (QND) detector.
Controlled, reliable production of nonclassical (i.e., sub-Poissonian) paired photons (biphotons) is an enabling technology for quantum information science. Biphotons were originally produced in traditional nonlinear crystals using spontaneous down-conversion processes. These sources suffered from low brightness, broad linewidths and short coherence lengths. A recent alternative approach uses EIT to generate narrow-band paired photons in a normally opaque atomic medium. DARPA Slow Light researchers are using this approach to generate biphotons with long coherence lengths and controllable bandwidths. Such narrow-bandwidth photon pairs ease many potential integration problems that occur when sending non-classical photons over large distances.
In recent work, Balić et al. demonstrated the generation of narrow-band paired photons of controllable length based on slow light in cold atomic clouds. The generated waveforms have a temporal length determined by the tunable optical group velocity, providing rudimentary waveform control. Counter-propagating paired photons were produced into opposing single mode fibers at a rate of 12,000 pairs/s with a controllable wave packet width in the range of 50 ns.
In subsequent work, researchers achieved similar results using a single laser beam and greatly reduced experimental complexity (Kolchin et al., 2006). These experiments have demonstrated rapid production of correlated photons with nominal coherence times that are orders of magnitude longer than those produced by spontaneous down-conversion.
The ability to combine sources of paired photons with techniques for storing photon excitations in a medium serves as the basis of a scalable quantum communications network, where photons act as information carriers and atomic ensembles act as memory nodes. Techniques to facilitate controlled interactions between single photons and atoms are being actively explored by Slow Light.
Specifically, Eisaman et al. recently demonstrated the slowing and storing properties of EIT down to the single-photon regime. In this work, the group velocity of a single photon was slowed down to about c/300, with a fraction of the incident pulse localized in a material system and later retrieved after microsecond-long time intervals.
In order to make single photons truly practical, Slow Light researchers are pursuing efforts to scale atomic vapor cells from the table-top down to chip-scale dimensions. To that end, program researchers recently fabricated the first monolithically integrated rubidium vapor cell using hollow-core antiresonant reflecting optical waveguides (ARROWs) on a 1 cm2 silicon chip (Yang et al., 2007). In addition, these cells were used to demonstrate rubidium spectroscopy on a chip.
A waveguide-based integrated system offers the combination of tight photon confinement with long atom interaction lengths, as well as the prospect for integration with additional optoelectronic structures for new functionality. Using these cells to demonstrate slow light and single-photon nonlinearities is an ongoing effort.
Slow light at the end of the tunnel
By focusing research on approaches that are compatible with current optical communication technology, DARPA’s Slow Light program has begun transforming this cutting-edge science into useful technology. A variety of applications beyond communication have surfaced as viable uses for slow-light-based components, including test and measurement equipment, radar systems and general signal processing. Fueled by rapid progress in slow light research, slow-light-based devices may enter into the mainstream over the next few years.
The authors gratefully acknowledge the work of current and past members of the DARPA Slow Light program and thank them for their countless hours of dedicated work.
[ Enrique Parra is with Booz Allen Hamilton in Arlington, Va. John R. Lowell is with DARPA in Arlington, Va. ]
About DARPA and Slow Light
DARPA is the central research and development agency within the U.S. Department of Defense. Its mission is to prevent technological surprises that could threaten national security. The agency does this by sponsoring revolutionary research that bridges the gap between fundamental discoveries and military utility.
In other words, DARPA acts as a “technological engine” within DoD, supplying innovative options to the entire department. The agency has a legacy of high-payoff programs that have led to breakthrough developments in military capability (e.g., stealth technology, unmanned aerial vehicles). In some cases, DARPA’s developments have been transitioned to widespread civilian use (e.g., global positioning systems and Arpanet, which is an operational packet switching network and precursor to the Internet).
DARPA’s Slow Light program is managed by the Defense Sciences Office (DSO), which is one of six technical offices within the agency. As the most technologically diverse office, DSO places no limits on the technical opportunities it can pursue. DSO programs cover a broad spectrum of science and engineering research.
Currently in the second of three phases, Slow Light takes its inspiration from the seminal atomic physics experiments by Lene Hau and Steve Harris, in which light pulses propagated through a medium with greatly reduced group velocities and little absorption. In these remarkable experiments, light pulses were slowed by several orders of magnitude (slow light) or stopped in the material (frozen light) and subsequently retrieved.
Slow Light has two main thrusts aimed at exploiting the two principal physical properties of slow light materials: (1) ultra-slow group velocity and (2) high transmission with large effective nonlinearity. For the first area, researchers are focused on demonstrating high-bandwidth analog tunable optical delays that are compatible with modern communications and computational systems. An element with this functionality forms the basis for many all-optical signal processing components such as buffers, equalizers, routers, etc.
Access to a large range of optical delays at high bandwidth translates into signal processing power and—because optical delays maintain phase coherence—additional coherent processing options. Initial applications of these materials are expected to include high-speed optical processing, telecommunication networks, high-speed radio frequency signal processing and radar systems. The program is searching for new material systems that have large bandwidths, continuous delay tunability, fast reconfiguration times and excellent integration characteristics.
The second thrust of the program is to develop systems with unprecedented low-light level response. This would enable single-photon devices, such as efficient, on-demand, single-photon sources and detectors. Of particular interest are non-destructive detectors (quantum non-demolition detectors) that measure the presence of a photon without altering the nature of its information.
The ability to produce on-demand single photons is an enabling technology for quantum information science and is particularly necessary for quantum communication and cryptography. A further interest would be to deterministically change the properties of the single photon, potentially improving detection using matched detector techniques.
References and Resources
>> For more information on DARPA, please visit www.darpa.mil.
>> For a review on early slow light work, see R.W. Boyd and D.J. Gauthier, “Slow and fast light,” Prog. Opt. 43, E. Wolf, ed., Elsevier, Amsterdam, 2002.
Optical fiber systems
>> Y. Okawachi et al. Phys. Rev. Lett. 94, 153902 (2005).
>> M.D. Stenner et al. Opt. Express 13, 9995 (2005).
>> Z. Zhu et al. J. Lightwave Technol. 25, 201 (2007).
>> B. Zhang et al. Opt. Express 15, 1878 (2007).
>> J.E. Sharping et al. Opt. Express 13, 6092 (2005).
>> Y. Okawachi et al. Opt. Express 14, 2317 (2006).
>> J.E. Sharping et al. Opt. Express 13, 7872 (2006).
>> Y. Okawachi et al. Opt. Express 14, 12022 (2006).
>> Y. Wang et al. IEEE Photon. Tech. Lett. 19, 861 (2007).
Semiconductor optical amplifiers
>> S. Sarkar et al. Opt. Express 14, 2845 (2006).
>> B. Pesala et al. Opt. Express 14, 12968 (2006).
>> F.G. Sedgwick et al. Opt. Express 15, 747 (2007).
Resonant photonic systems
>> Y.A. Vlasov et al. Nature 438, 65 (2005).
>> F. Xia et al. Nature Photon. 1, 65 (2007).
>> Q. Xu et al. Nature Phys. 3, 406 (2007).
>> S. Fan et al. Opt. Photon. News 18, 41 (2007).
>> R.M. Camacho et al. Phys. Rev. Lett. 98, 153601 (2007).
>> R.M. Camacho et al. Phys. Rev. Lett. 98, 043902 (2007).
Single photon nonlinearities
>> V. Balić et al. Phys. Rev. Lett. 94, 183601 (2005).
>> P. Kolchin et al. Phys. Rev. Lett. 97, 113602 (2006).
>> M.D. Eisaman et al. Nature 438, 837 (2005).
>> W. Yang et al. Nature Photon. 1, 331 (2007).