Helium Is Good for Silicon Lasers

Breck Hitz

Recently, several research groups demonstrated Raman lasers in silicon, perhaps bringing the union of photonics and microelectronics one step closer. To perform these experiments, however, they have had to employ complex techniques to avoid absorption of the laser photons by free carriers (electrons and holes) in the semiconductor. Now scientists at the Chinese University of Hong Kong have shown that simply implanting helium ions in the silicon can lower the free-carrier lifetime dramatically, reducing the absorption of laser photons.

The free carriers are generated by two-photon absorption of the pump radiation in the silicon. One way to avoid the absorption of laser photons by the free carriers is to pulse the laser so quickly that the pulse is over before the free-carrier density becomes large enough to cause problems, and then to wait until the carriers recombine before pulsing the laser again.

Figure 1. Researchers compared the Raman gain in a silicon waveguide with implanted helium to that of a similar waveguide without helium. They measured the gain by tuning the probe to the Stokes wavelength and monitoring the output at that wavelength. Images ©OSA.

That’s fine for pulsed lasers, but a continuous-wave silicon type requires a different approach. The technique that’s been successful is to fabricate a PN junction around the waveguide and to apply a reverse bias across the junction. The reverse bias sweeps free carriers out of the waveguide and reduces the absorption loss to a value of less than the Raman gain.

The drawbacks of this technique are that it requires the PN junction to be fabricated around the waveguide — which often is not a straightforward process — and it requires an external voltage source to bias the junction.

In contrast, implanting helium ions in a silicon waveguide is relatively straightforward and, once there, the ions require no external voltage to serve their function.

The researchers in Hong Kong have not observed lasing in silicon, but they have shown that free-carrier absorption can be reduced significantly by implanting helium in a silicon waveguide. They used a 4-μm-long silicon rib waveguide and, for experimental purposes, fabricated a PN junction across it by implanting boron and phosphorous on either side of the rib. They implanted helium into the rib at a nominal energy of 0.8 MeV so that the maximum of the helium distribution was at the center of the waveguide. They found that an implant ion density of 10¹²/cm² was sufficient to suppress the free-carrier absorption without deleterious side effects.

Figure 2. In the waveguide with no implanted helium (a), both the Stokes probe at 1556 nm and a non-Stokes probe at 1546 nm decreased with increasing pump power because the free carriers that absorbed the probe photons increased with pump power. In the helium-implanted waveguide (b), free-carrier absorption was reduced enough to allow the Raman gain to dominate.

They pumped the waveguide at 1440 nm and probed it at the Stokes wavelength of 1556.5 nm (Figure 1). They chopped the probe and employed phase-sensitive detection to increase the data’s signal-to-noise ratio. In the initial experiment, they applied no bias across the PN junctions and compared the propagation in a helium-implanted waveguide to that in a waveguide without helium (Figure 2). In the helium-implanted waveguide, the free-carrier absorption was reduced enough to allow a net gain at the Stokes wavelength.

As a further demonstration of the effectiveness of the implanted helium, the scientists forward-biased the PN junctions to generate free carriers in the waveguides and again compared the helium-implanted waveguide to one without helium (Figure 3). The optical attenuation increased rapidly with increasing free-carrier current in the waveguide without helium, reaching 39 dB at a current of 60 mA. In the waveguide with helium, however, attenuation at the same current level was only 1.24 dB. They concluded that the helium produced a significant reduction in the free-carrier lifetime.

Figure 3. When the researchers forward-biased the PN junction across the waveguide, they saw a dramatically greater increase in attenuation at the Stokes wavelength in the waveguide without helium than in the helium-implanted waveguide. The horizontal axis is the current through the junction.

To gain a quantitative understanding of the effect of the implanted helium, the researchers solved the differential equations describing the intensities of the pump and probe radiation in the waveguide. They assumed that the free-carrier lifetime in the waveguide without helium was 100 ns and calculated that the lifetime was reduced to 1.9 ns by the implanted helium. A comparison between the measured gain in the waveguide and that predicted by their mathematical model showed hardly any discrepancy, giving the investigators confidence that their model was accurate.

Optics Letters, June 1, 2006, pp. 1714-1716.