Slow progress for OPOs in tunable systems market

17 Jun 2002

The optical parametric oscillator - a source with a wide and continuous tuning range - has been slow to enter the market, despite its advantages over the old dye lasers. Rob van den Berg talks to the market players.

From Opto & Laser Europe March 2002

About 10 years ago, the optical parametric oscillator (OPO) promised to solve the problems that many laser laboratories were experiencing with dye lasers. Before the commercialization of OPOs, almost 20 different dyes were needed to cover the wavelength range between 400 and 1000 nm. The versatile OPO can extend that range up to 2000 nm and hopes were high that it would find uses in many applications, from spectroscopy to telecommunications.

The reality, however, was different. Today only a small number of OPOs are made and sold by a scattering of companies worldwide. Is this the start of a slow revolution, or were OPOs never destined to make it big?OPOs have much to offer. They are available in a plethora of designs: from the mid-infrared to the visible (and even into the ultraviolet by means of frequency doubling or mixing) with pulse durations from femtoseconds to nanoseconds and pulse energies ranging from a few nanojoules to tens of millijoules. According to Eli Margalith, president of Opotek, the first US company to offer a broadband-visible OPO, the device "has finally become a commercial reality".

Although the principle behind the OPO had been known of since 1965, development of a product was initially hampered by the lack of stable non-linear crystals with a high enough damage threshold and the unavailability of stable pump lasers.

The first problem was solved at the end of the 1980s with the discovery of materials like beta barium borate (BBO) and lithium borate. It took just a few more years to overcome the other difficulty: parametric down conversion was found to be extremely sensitive to any variation in the pump beam's intensity profile, pointing, mode structure or collimation. Only when the injection-seeded Nd:YAG, and later the Ti:sapphire laser, proved stable enough did the first commercial OPOs hit the market.

"It was an immature introduction," said Margalith. "The big laser companies sold a few systems based on their reputations, but users soon found that these products did not match up to their promise and had been introduced prematurely. This led to a tremendous loss of confidence. You should never forget that non-linear optics is a very difficult technology to master."Margalith thinks one of the reasons that OPOs are selling slowly is that many users had bad experiences with the systems when they were first introduced onto the market.

Phil Smith, tunable products manager of Continuum in the US, admits that there were growing pains. "BBO is difficult to work with. It attracts moisture and dust, so now we only apply crystals in sealed cavities," he said. "The temperature stabilization of the mount the crystal resides on required a lot of work. We also had to improve the pointing stability of the YAG-pump laser from 100 µrad to less than 30 µrad by re-engineering the bench and mounts - you have to hit the crystal dead centre every time."

Arnd Krueger, marketing manager at Spectra-Physics, believes that the OPO has finally become a success: "We have more than 300 systems out in the field," he said.

Spectra-Physics' nanosecond product combines two oscillators. It was developed to overcome a disadvantage of the first "simple" OPOs: their relatively broad bandwidth far exceeded the transform-limited width of the pump pulse.

Krueger explained: "We use a small amount of the pump beam to pump an OPO (the master oscillator) incorporating a grating instead of a totally reflective mirror. The resulting narrow line output is used to seed a second OPO (the power oscillator)."

This system acts as an optical parametric amplifier. In its simplest non-resonant set-up, a seed pulse at the idler or signal wavelength is combined with a strong pump pulse inside the crystal. As it passes the crystal, parametric down-conversion of the pump occurs and the seed is amplified significantly while maintaining the fidelity of the oscillator's narrow bandwidth.

Alternatively, a resonant amplifier takes advantage of multiple passes of the pump and parametric waves through the crystal to further maximize the gain. "Tuning is completely automated from 220 to 1850 nm," said Krueger. "With a bandwidth as narrow as 0.075 wavenumbers, it is a wonderful tool for high-resolution spectroscopy."

For others, even this wavelength range is not wide enough. Konstantin Vodopyanov, director of mid-infrared systems at BlueLeaf Networks in the US, explained why: "The region between 2 and 20 µm is the molecular fingerprint region in which gases exhibit their characteristic features. Light sources in the long-wavelength infrared would allow new ways to monitor atmospheric pollution and detect the presence of drugs and explosives at ultra-low levels."

Unfortunately, most conventional parametric crystals, such as lithium niobate, BBO and potassium titanyl phosphate, have no transmission above 4 µm. Vodopyanov had to find more appropriate materials that also matched commercially available laser lines. He came up with two promising candidates: zinc germanium phosphide and cadmium germanium arsenide.

Vodopyanov said: "We have shown that all of these crystals can be incorporated into compact OPO devices. Using cavity ring-down spectroscopy, we succeeded for the first time in spectroscopically detecting explosives like TNT."While some are extending the OPO's tuning range, others have focused on the time domain. It is, however, no easy matter to extend OPO operation into the femtosecond regime. Parametric down-conversion is inherently an inefficient process in which both the signal and the idler waves have to "grow out of the noise" by passing through the crystal several times. Unlike a lasing medium, this cannot store energy, so build-up only occurs while the pump pulse is present in the crystal. For a pump pulse that lasts a few nanoseconds this is not a problem, but with picosecond and femtosecond pulses there is not enough time for the down-converted light to acquire intensity.

The solution to this problem is exact repetition, or synchronous pumping. By matching the OPO length to the cavity length of the pump laser, many consecutive pulses from the pump laser can sequentially pump one circulating pulse within the cavity of the OPO. Although the Ti:sapphire laser has proved to be an ideal source in this respect, the development of diode-pumped solid-state lasers has also driven progress. With their high repetition rates and excellent pulse-to-pulse stability, femtosecond OPOs are now the source of choice for time-resolved spectroscopy.

Spectra-Physics' Krueger sees a lot of demand from semiconductor and telecommunications research facilities for tunable femtosecond sources of around 1.3 and 1.5 µm. They are used for evaluating critical semiconductor material properties or for developing techniques to obtain higher transmission rates through optical fibres.

In recent years, much effort has been concentrated on developing a continuous-wave OPO. This would need a single pass gain to exceed the round-trip power loss in the cavity at all times, which either requires extreme pump powers - which no crystal can withstand - or some clever thinking about the optical processes that occur in a parametric crystal. Efficient energy conversion of the pump beam requires a fixed-phase relationship between the beams. However, this relationship changes as the beams travel through the crystal, owing to dispersion. In conventional phase matching, the natural birefringence of the crystal is used to compensate for this change in the index of refraction with wavelength.

If the crystal were not birefringent, the amplitude of the generated waves would reach a maximum when a phase difference of 180° was attained. If after this distance (the coherence length) the domain structure of the crystal is reversed, the phase difference decreases again and the build-up of signal and idler power continues.

This procedure - called quasi-phase-matching - is repeated after each coherence length. It can be achieved by periodic poling in ferroelectric crystals - of which lithium niobate is the most popular - and allows efficient operation at any user-specified wavelength within the transparency window of the material. It is no wonder that periodically-poled materials have replaced conventional non-linear materials in many applications.These materials were the breakthrough that was needed for the realization of the first continuous-wave OPO. In November 2000 the first commercial instrument was launched by German firm Linos. The company's product manager for laser systems, Reiner Urschel, explained why their OS4000 is still the only continuous-wave OPO: "The development costs are rather high. We were lucky to get a license from Jürgen Mlynek and Achim Peters' group at the University of Konstanz. With their basic research and a little bit of development from our side we were able to build it."

The OS4000 has a wide tuning range of between 1.45 and 4 µm. The various poling periods in the crystal can be accessed by moving it with a micrometer. Urschel said the main problem was stabilizing the resonator length: "To this end we used an OPO cavity with pump enhancement. This allows us to stabilize the resonator length and lowers the threshold of the OPO."

This can also be done by putting the crystal inside the cavity of the pump laser, where it can access the higher field present there. This was experimentally demonstrated by Majid Ebrahimzadeh's group at St Andrews University in the UK.

However, the market for continuous-wave OPOs is currently small. Urschel said: "We have sold four to research groups at universities. They are ideal light sources for a number of precision applications such as spectroscopy, metrology, and the analysis and detection of molecules." As a further development he sees a green-pumped continuous-wave OPO for the wavelength range between 700 and 2400 nm. To generate these shorter wavelengths, however, grating periods below 10 µm are necessary. These are far from easy to generate.

Ebrahimzadeh says there are many more areas where novel non-linear materials have led to breakthroughs. "If you want to make continuous-wave OPOs more compact, you have to aim for direct diode-pumping. Two years ago we demonstrated such an OPO using an external-cavity diode-laser pump source by taking advantage of periodically-poled lithium niobate. We also generated ultrashort, five-optical cycle pulses in the mid-infrared."

Despite these results and the progress made by OPO manufacturers, most potential OPO users seem hesitant about picking up on these developments. As Continuum's Smith notes: "We are hovering to see what happens. We are focusing on improving the products we already have rather than developing new ones. In one respect the OPO has not lived up to its promise: it has not replaced the dye laser. We are still selling large numbers of dye lasers."

The OPO is a source with a wide and preferably continuous tuning range. It was designed as a replacement for dye lasers, which are often difficult to handle.

An OPO is a frequency divider: it splits a short-wavelength laser photon into two longer wavelength photons. This requires a non-linear optical crystal, which converts a pump beam into a signal and idler beam. By definition, the idler beam has the longest wavelength.

The crystal is placed in an optical resonator which resonates one or both of the generated waves and optionally also the pump beam. This feedback causes gain in the parametric waves in a process similar to build-up in a laser cavity. The sum of the signal and idler photon energies is equal to the original pump photon energy. Nature requires that momentum must be conserved, which implies a fixed phase relationship between the three beams.

Under most circumstances, the variation of index of refraction with crystal angle and wavelength allows this "phase matching" condition to be met only for a single set of wavelengths for a given crystal temperature, crystal angle and pump wavelength. So by rotating the crystal or varying its temperature, different wavelengths of light can be produced.