Speeding Up GaN For Biomedical Sciences
If you step into a life-science laboratory, you’ll probably see a large optical bench crowded with expensive, bulky equipment for time-resolved fluorescence measurements. The make up of this kit will vary from lab to lab, but on the bench you’ll possibly find photomultiplier tubes and a scanning confocal microscope. Nestled amongst some of these items will undoubtedly be a Ti:Sapphire laser oscillator, emitting a steady stream of pulses that could last for a few picoseconds, or maybe even a fraction of that time.
This Ti:Sapphire laser is a complex, cumbersome contraption that is assembled from several units. And snuggled up next to it, taking up yet more space, will be a frequency-doubling component, because this laser emits in the infrared and biological samples and dye markers absorb in the blue-green.
At the heart of Ti:Sapphire laser sits a Ti:Al2O3 crystal, a medium that must be pumped by a powerful green laser source. Unfortunately, semiconductor lasers emitting in the green are still in their infancy, so the tried and tested approach in today’s labs is to use either a Nd:YVO4 or Nd:YAG laser, pumped by an infrared diode. But the emission from the neodymium-based laser, while having the virtue of being efficient, is also in the near-infrared – to be precise, it’s at 1064 nm. So it has to be frequency doubled in a second-harmonic generation (SHG) unit to provide the green pump beam for the Ti:Sapphire oscillator.
Putting this into simpler terms, the generation of ultrashort pulses requires three lasers – a semiconductor laser diode, a Nd:YVO4 laser and a Ti:Sapphire laser – plus two SHG frequency conversion units. Working in combination, these lasers and frequency doublers are incredibly inefficient – a 50 W output from the diode laser is converted to just a few milliwatts of blue-green light. Running costs are also high, partly because there are steep bills for maintenance and repair. Given all this, the obvious question to ask this: Is there an easier way to generate blue-green picosecond pulses?
Well, there isn’t a unit that you could buy off the shelf for doing this today – but there soon could be. That’s because it has been recently shown that is possible to generate really short pulses with a GaN-based laser, which has the great attribute that its spectral range is a perfect match for the absorption spectrum of many organic components. This wide bandgap source also has many key advantages over the Ti:Sapphire incumbent, including low cost, small size and maintenance-free operation. These characteristics enable the GaN laser to provide the first portable source of blue-green ultra-short pulses, opening up the opportunity for a portable, time-resolved fluorescence measurement system to be placed at the point of care for biomedical diagnostics.
Within Europe we are trying to create such a laser through a project called FemtoBlue, which is funded through the European Commission. This effort, which kicked off in September 2009 and is backed by € 2 million of funding, is drawing on a diverse set of talents held by researchers at six institutions: The Swiss innovation and research institute CSEM, which is coordinating the project; Fraunhofer IAF; EPFL; the Lebedev Physical Institute of the Russian Academy of Sciences; the University of Cambridge; and the Technical University of Berlin. The mission for this multinational team is to create an ultra-fast semiconductor laser diode that produces sub-picosecond optical pulses in the blue and violet spectral range.
If successful, the benefits could extend beyond activities in the biological sciences. Blue and violet lasers with ultra-fast pulses could unlock the door to a new, threedimensional optical data storage disc technology that replaces the Blu-ray standard. Other possible applications include multiphoton nano-processing and nano-imaging (see “Further Reading" for details).
Our development of ultrafast GaN lasers is not the only work in this field. A Japanese collaboration between Sony Corporation and Tohoku University’s New Industry Creation Hatchery Center has recently reported the output of 2 ps pulses with a peak power of 20 W and a 1 GHz repetition rate from an external-cavity, multisection laser diode.
By feeding this output into a semiconductor optical amplifier, this research team has boosted peak power to 300 W. Their approach is markedly different from ours, using a relatively large laser system rather than a monolithic cavity design, but it provides another example of the capability of GaN lasers for providing pico-second pulses in a spectral range suited to biomedicine.
Forming ultra-short pulses
The good news for anyone trying to develop picosecond GaN lasers is that they don’t have to re-invent the wheel. Instead, they can exploit all that has been learnt in the evolution of arsenide and phosphide lasers that deliver ultra-short optical pulses. With these AlGaAs and InGaAsP lasers, it has been possible to realize a wide variety of ultra-fast dynamic regimes by applying multiple p-contacts to the top of the device (see Figure 1).
In these modified lasers, one section is positively biased to provide optical gain, while another section takes on the role of a saturable absorber, which is driven as a photodiode with negative bias. With this design, lasing characteristics are governed by the driving conditions for each cavity and its geometry. Through a complex interplay of numerous phenomena, lasing is possible in eight different dynamic regimes, three of which produce ultrafast pulses.
By applying an appropriate combination of applied voltages, the laser can operate in a self-starting passive mode-locking regime. Operating in this fashion, researchers at University of Cambridge have produced a InAs-GaAs quantum dot source emitting pulses with a peak power of 2.25 W, a duration of 0.35 ps, and a repetition rate of 16.8 GHz that is set by the cavity roundtrip time. Applying a tapered waveguide and a longer cavity, researchers at the University of Dundee and Alcatel-Thales III-V Lab have demonstrated a peak power of 15 W, and a pulse duration of 0.8-1 ps at a repetition rate of 10 GHz. Self-mode locking does have the downside of a relatively high timing jitter for the generated optical pulses, but this can be addressed by applying a periodic modulation to the absorber at the pulse repetition rate. Do this and the laser operates in the active mode-locking regime.
A vastly different form of laser output is also possible, which is referred to as the Dicke superradiance regime – it involves spontaneous emission of a solitary coherent optical pulse. Prior to the emission of this pulse, the active region must be densely populated with a nonequilibrium of electron-hole pairs. Creating this condition in a GaAs laser enabled researchers at Lebedev Physical Institute to produce pulses with a peak power of a few hundred watts and a width of just 180 fs, operating in a pulse-on-demand mode.
Bridging the gap
It is tempting to think it would be easy task to transfer the technology used to create ultra-short pulses in GaAs-based and InP-based lasers to blue and violet GaN lasers. But that’s not the case. One must realize that although the requirements for making red and infrared femto-second monolithic semiconductor lasers have been known for more than 20 years, practical realisation of such devices is not that advanced – industrial developers of such sources are, in general, still refining these sources in their labs. The reality is that there are very few commercial monolithic modelocked laser products on the market – we know of only one industrial company offering such a product in reasonable quantities.
One major distinction between the nitrides and their more traditional III-V cousins is a significant difference in their set of intrinsic characteristics. For example, nitrides have a higher effective hole and electron mass. This not only limits the available output power in a GaN laser, but also restricts its pulse width to 30-50 ps (assuming that a conventional design is used and the device is operated in the gain switching regime). However, the large effective hole mass might offer advantages for (quasi-) continuously pumped femtosecond modelocked GaN lasers, thanks to a high density of hole states. The reality is that this work is in its infancy, and there is a whole spectrum of unanswered questions concerning ultra-fast carrier dynamics in InGaN alloys.
Another distinguishing feature of the nitrides is that they have a wurtzite crystalline symmetry, which is markedly different from the zinc blende symmetry associated with conventional III-V materials (see Figure 2).
The wurtzite crystal structure is responsible for an internal spontaneous and piezoelectric polarization field, which pulls apart the electrons and holes in nitride quantum wells, reducing carrier overlap and quashing radiative efficiency. If the well is too wide it can eliminate optical gain in this active region and thereby prevent lasing.
To realize mode-locking and superradiance in any class of semiconductor laser, the device must contain a suitable saturable absorber. Fortunately, it is relatively easy to form this in a monolithic-cavity with multiple contacts. In conventional III-V devices, tweaking the reverse bias voltage of the absorber section provides control of saturable absorption and output pulse parameters via the quantum confined Stark effect (or Franz-Keldysh effect in bulk active layers). In this class of laser, increasing the negative bias produces a red-shift in the absorption edge, and in turn a gradual increase of modal absorption.
With a nitride laser it’s a very difference story, due to competition between the quantum confined Stark effect and the built-in polarization field. Increases in bias initially lead to a reduction in absorption, because the external field just offsets the internal built-in field – only when the voltage is cranked up is the necessary level of absorption realised.
When the influences associated with the nitrides are accounted for, we have found that we are able to build pico-second lasers that behave as one would expect. Our preliminary results that have been acquired with a streak camera reveal a train of periodic pulses from our laser when it is driven in a passive mode-locking regime (see Figure 3). We can also drive our laser in a different manner, so that its emission resembles the regime of superradiance. According to measurements with our sampling scope, our laser produces red-shifted, high amplitude pulses with a large jitter, features indicative of superradiance. Measurements of pulse width and efforts to optimise the production of pulses are on going.