QC lasers may provide terahertz bandwidth for communications

by K. Alan Shore

With picosecond-scale electron lifetimes, quantum-cascade lasers may reach terahertz modulation speeds; near-infrared versions would be useful for fiberoptic communications.

The demonstration of the quantum-cascade (QC) laser by Federico Capasso and his group at Lucent Technologies (Murray Hill, NJ) has opened up a challenging area for device physics and engineering.¹ A particularly tantalizing possibility is the development of QC lasers that operate at telecommunications wavelengths. Such a possibility may offer advantages over conventional semiconductor-laser technology; an approach is already being pursued toward this objective.

QC-laser operating principles
In conventional laser diodes, optical processes are mediated by transitions involving electrons and holes in the conduction and valence bands of the semiconductor. Such lasers are thus bipolar—they use both negatively charged electrons and positively charged holes. A particular class of semiconductor optoelectronic devices utilizes so-called quantum-well materials, in which electron-hole transitions occur between quantized energy subbands in the conduction and valence bands—known as intraband transitions. The transition energy and hence the operation wavelength of such devices is largely determined by the semiconductor bandgap. In order to access a desired wavelength, it is thus necessary to identify a material system with appropriate semiconductor bandgap energy.

On the other hand, use can be made of optical transitions between energy subbands within either the conduction or valence band of the quantum well—thus called intersubband transitions. Such transitions involve just one charge carrier: either conduction-band electron transitions or valence-band hole transitions. Therefore, optoelectronic devices exploiting intersubband transitions are unipolar.

Quantum-well materials are obtained by growing thin layers of semiconductor at dimensions comparable to the electron or hole wavelength in the semiconductor material. The quantized conduction- and valence-band energy levels in quantum-well materials are determined by the dimensions of the quantum well. Since those dimensions are determined by the device fabrication process, energy levels are within the control of the device designer; the transition energies no longer depend upon the semiconductor bandgap. This observation points to the opportunity for using the same material system to design laser structures capable of accessing new ranges of operating wavelengths.

Dynamics and nonlinear optics of QC lasers
The dynamic behavior of semiconductor lasers is largely determined by two characteristic lifetimes: the photon lifetime and the charge carrier or electron lifetime. Usually it is the electron lifetime that is of greatest significance. The photon lifetime is governed by optical losses in the laser and hence is particularly significant for determining the conditions for achieving laser action. The photon lifetime in most semiconductor lasers is on the order of 1 ps. In conventional bipolar interband semiconductor lasers, the electron lifetime is generally about 1 ns. The response of any dynamic system is limited by the slowest characteristic time constant. This means that for conventional semiconductor lasers their speed of response, or bandwidth of operation, is normally limited by the electron lifetime. Modulation bandwidths of semiconductor lasers are typically on the order of 10 GHz.

Intersubband semiconductor lasers are characterized by electron lifetimes on the order of 1 ps—comparable to the photon lifetime. This remarkable feature brings several significant effects into play. Most obviously, it is expected that the modulation bandwidth of QC lasers should be significantly greater than those of conventional semiconductor lasers. Theoretical work undertaken by researchers at University of Wales Bangor predicts an opportunity to access terahertz modulation bandwidths using intersubband lasers (see Fig. 1).^{2, 3} While direct experimental verification of such performance has yet to be reported, clear support for the superior dynamic performance of QC lasers has been obtained in experiments on gain-switching and modelocking of QC lasers.^{4, 5, 6}

The availability of picosecond electron lifetimes offers other intriguing opportunities when attention is turned to using the nonlinear optical properties of intersubband transitions. Interest in such work has been motivated by a longstanding observation of strong nonlinearities associated with a large dipole moment obtained in intersubband transitions. Of particular interest is the use of optical nonlinearities to perform frequency translation using four-wave mixing techniques. The electron lifetime determines the frequency range over which efficient four-wave mixing can be performed. The picosecond lifetimes of intersubband transitions offer terahertz-frequency translation ranges.

Both the capability for achieving terahertz modulation bandwidths and the possibility of performing terahertz-bandwidth frequency translation would seem to make QC lasers extremely attractive for use in optical communications systems. There is, however, one major impediment to making full use of the potential of these lasers for such an application. The operating wavelength of available QC lasers is not compatible with optical fiber channels. Fiberoptic communications uses lasers operating at near-infrared wavelengths notably at 1.3 and 1.55 μm.

Given the ready availability of laser diodes at these wavelengths, it is reasonable to wonder why any effort should be made to develop QC lasers operating in the near infrared, and specifically telecommunications wavelengths. The motivation for such work can be challenged even more pointedly when considering that technologies such as wavelength-division multiplexing already offer means for meeting projected demands for terahertz-bandwidth data-carrying capabilities in optical communications systems. In order to deliver a bandwidth of 10 THz using conventional semiconductor lasers, however, it would likely be necessary to multiplex several tens, if not hundreds, of such lasers. The management of an optical communication system of such complexity is quite challenging. On the other hand, if the intrinsic bandwidth of QC lasers can be harnessed at telecommunication wavelengths, then a 10-THz bandwidth can be provided with just a few lasers. The crucial advantage offered by the superior dynamic properties of QC lasers is simplicity.

Near-infrared intersubband emission
The appreciation of the opportunities offered by the dynamic and nonlinear optical properties of intersubband lasers has motivated researchers at Bangor and the University of Manchester Institute of Science and Technology (UMIST; Manchester, England) to take up the challenge to develop a near-infrared intersubband laser (see Fig. 2). Our initial foray into this challenging field was to make basic calculations of the intersubband optical gain for near-infrared operation.^{7, 8}

The main challenge to achieving intersubband lasing at telecommunications wavelengths is the identification of a material platform that will be sufficiently robust to allow appropriate quantum-well structures to be grown reliably. Moving directly to telecommunications wavelengths would be a formidable undertaking, and hence our work is directed first at the demonstration of intersubband emission at a wavelength of 2.0 μm; subsequent efforts will be directed at reducing the operating wavelength to 1.55 μm.

Material platform
In the work being pursued at the University of Wales Bangor and UMIST, the indium gallium arsenide (InGaAs)-and indium aluminum arsenide (InAlAs)-based material system In_xGa_(1-x)As-In_yAl_(1-y)As grown on indium phosphide (InP) has been selected as the semiconductor system for use (in contrast, mid- and far-infrared-emitting QC lasers are based on silicon germanium). The choice is based around a number of considerations, but principally that the processing technology of this material system is well-understood and controlled, thus permitting reproducible and predictable device fabrication. More technically, it is known that appropriate doping of the chosen materials can be achieved in order to facilitate low-resistance ohmic contacts to be applied to the devices.

It is also observed that the material system is compatible with the InGaAs-InGaAsP system used in telecommunications. In a previous project, UMIST grew and successfully fabricated highly strained multi-quantum-wells of In_.7Ga_.3As-In.₅₂Al_.48As that contained up to 100 wells for fast optical switches. It is confidently expected that such structures can be adapted to yield 2-μm-emitting intersubband devices. Access to 1.55-μm intersubband transitions will prove much more challenging, as we expect even more highly strained layers will be needed. Scientists at UMIST have experience growing such highly strained layers in the context of research on pseudomorphic high-electron-mobility transistors, where dislocation-free layers as thick as 10 nm have successfully been grown. The prospects for this material system are rather promising, but serious work remains to be done to achieve the project aims.

Nonlinear measurement of terahertz response
As already emphasized, a major motivation for undertaking the proposed work is the promise of high-speed optical response arising from characteristic picosecond carrier lifetimes in intersubband devices. An important question is how to demonstrate that the expected performance can be achieved. Picosecond carrier lifetimes should provide terahertz direct-current modulation of intersubband lasers. Direct testing of such capabilities poses a number of technical challenges, however, which merit attention in their own right. Our approach to demonstrating this capability is to exploit the second major feature of intersubband processes: the availability of strong nonlinearities. The approach is to evaluate the terahertz response of near-infrared intersubband structures via multiwave mixing. This approach is inherently of interest because it enables verification of additional device functionality.

An open question is whether near-infrared intersubband semiconductor lasers will provide the means for powering terahertz-bandwidth optical communications systems. The effort required to meet these challenges is justified by the tremendous benefits that would be offered by the dynamic and nonlinear optical properties of these devices.

REFERENCES
1. J. Faist et al., Science 264, 553 (1994).

2. W. M. Yee, K. A. Shore, and E. Schoell, Appl. Phys. Lett. 63, 1089 (1993).

3. C. Y. L Cheung and K. A. Shore, J. Modern Optics 45, 1219 (1998).

4. R.Paiella et al., IEEE Photon. Tech. Lett. 12, 780 (2000).

5. R. Paiella et al., Appl. Phys. Lett. 77, 169 (2000).

6. R. Paiella et al., Science 290,1739 (2000).

7. C. Y. L Cheung, P. Rees, and K. A. Shore, IEE Proc. Optoelectronics 146, 9 (1999).

8. C. Y. L. Cheung, P. Reese, and K. A. Shore, J. Mod. Opt. 47,1857 (2000).

K. ALAN SHORE is professor of electronic engineering at the University of Wales Bangor School of Informatics, Dean Street, Bangor, Wales LL57 1UT; e-mail: [email protected].