Microchip OPOs operate in the infrared

Juan Capmany, Verónica Bermúdez, David Callejo, and Ernesto Diéguez

With the advent of optical parametric oscillators (OPOs), there is now very little of the infrared (IR) spectrum that is out of reach for solid-state laser sources. While OPO-based systems provide a wide range of wavelengths, they are large, unwieldy, and not rugged enough for some applications. Microchip lasers, on the other hand, are miniaturized, compact and reliable diode-pumpable solid-state laser sources that offer an IR wavelength range extending beyond the range that can be generated with diode lasers.¹ Their range of output wavelengths does not, however, equal that of OPOs. The challenge remains, then, to provide a miniature diode-pumpable wavelength-selectable device that covers as much of the IR spectrum as conventional solid-state OPOs. One option might be a microchip OPO.

In search of microchip OPOs
Bonding a small nonlinear crystal to a small IR laser crystal produces a composite microchip laser. The inclusion of the nonlinear crystal permits intracavity nonlinear frequency conversion of the IR laser output. Research has shown that intracavity frequency doubling to the visible range in composite microchip lasers essentially provides the same pump-to-visible efficiency as the corresponding pump-to-IR laser in the absence of the nonlinear crystal.

A simplified and more-integrated version of such a system is the self-frequency-doubled laser, in which the same crystal—a nonlinear crystal doped with laser active ions—acts both as an IR laser oscillator and a nonlinear frequency doubler. Full frequency conversion is obtainable using materials of moderate nonlinearity (around 2 pm/V) and 2- to 3-mm-long crystals. The key to balancing high efficiency with a suitable design involves the use of high-quality crystals with low scattering losses.

FIGURE 1 Shown are the differences in the spectral behavior of Nd³⁺ and Yb³⁺-based microchip OPO's. The simple two-multiplet configuration of Yb³⁺ provides a quasi-three-level laser system without ground- or excited state-absorption in the transparency range of the host crystal.

The principles of intracavity nonlinear frequency conversion extend to continuous-wave (CW) OPOs. Unlike intracavity-doubled or self-frequency-doubled lasers, though, OPOs must reach a threshold before net nonlinear gain is achieved and parametric oscillation becomes possible.

Early CW OPOs operated under external pumping; the high pump-power densities necessary to reach threshold demanded multiwatt-level pumping. While doubly resonant OPOs have lower thresholds, stability problems occur at moderate to medium powers. The output stability improves with singly resonant operation of the OPOs, but only at the expense of increased threshold. With the use of intracavity OPOs, though, the external pump power required to reach threshold can drop significantly—by an order of magnitude or more—because the pump wave for the parametric process is provided by the circulating energy stored in the cavity of the fundamental laser.

FIGURE 2. Although the efficiency obtained in this first demonstration of a PPLN microchip OPO was very low--0.5 mW out of an extractable down-converted power of 30 mW--these devices may someday reach as high an efficiency as two-crystal intracavity singly resonant OPOs.

Theoretically, all the power extractable from the laser can be downconverted intracavity under optimal nonlinear coupling conditions, even in the case of singly resonant OPOs. Recent demonstrations of CW intracavity singly resonant OPOs combining neodymium (Nd3+) based laser oscillators and periodically poled lithium niobate (LiNbO₃), potassium titanyl phosphate (KTP), or rubidium titanium arsenate (RTA) have achieved down-conversion efficiencies in excess of 85% with pump powers at the subwatt level, well within the reach of miniature diode lasers.² As in the case of the self-frequency-doubled lasers, intracavity OPOs can be integrated in a single crystal—a self-pumped OPO or OPO laser—which plays the double role of laser oscillator and parametric generator.³

In attempting to shorten the length of self-pumped microchip OPOs, it helps to combine suitable laser-active ions with a crystal that features as high a nonlinearity as possible and transparency in the infrared. Quasi-phase matching in periodically poled ferroelectric crystals permits exploiting of the higher valued components of the nonlinear tensor. The ion used can be chosen from a relatively large set of available laser-active ions and their laser channels—one desirable characteristic being that the active ion is diode pumpable.

Research results
One outcome of recent cooperative research at the Universidad Miguel Hernández (UMH, Elche, Spain) and Universidad Autónoma de Madrid (UAM, Madrid, Spain) was the demonstration of a 1-mm-long crystal of ytterbium (Yb³⁺) doped periodically poled LiNbO₃ (PPLN) as a self-pumped continuous-wave OPO. At least for microchip OPOs, one advantage of lithium niobate over other ferroelectric crystals such as KTP is that trivalent laser-active ions can be easily incorporated in the concentrations required for laser operation (around 1 mol %).

In this research project, periodically poled LiNbO₃—with a transparency range that permits parametric generation beyond 6 µm in the IR—was used. The main limitation with lithium niobate—the possibility of photorefractive damage (although it is less severe in the infrared and even reduced in PPLN crystals)—is easily overcome by codoping with magnesium oxide (MgO).

FIGURE 3. After broadening the quasi-phase-matched band at the expense of efficiency, the efficiency of periodically poled lithium niobate can still be similar to that of periodically poled KTP.

On the other hand, the Yb³⁺ ion is a particularly interesting option for intracavity OPOs in the IR. Its simple two-multiplet configuration provides a quasi-three-level laser system without ground-state absorption in the transparency range of the host crystal (see Fig. 1). The fundamental laser emission is around 1063 nm, and there is a tunability range of a few nanometers, as well as the potential to generate pulses as short as tens of femtoseconds. The closest competitive ion for pumping around 1060 nm (Nd3+) will not work so well for IR intracavity OPO applications, because of strong idler or signal absorption in some IR regions by the excited states of the ion. The Yb³⁺ ion, though, can be diode-pumped with indium-gallium-arsenide laser diodes around 980 nm.

As with most Yb³⁺-based lasers, because of the small energy difference of pump and laser photons, thermal loading by resonant pumping is very low, and the resulting slope efficiency limit is close to 0.95. In fact, stable, efficient CW operation of Yb³⁺:LiNbO₃:MgO was achieved by the laser spectroscopy group (GIEL) at UAM.⁴ It is also advantageous that Yb³⁺ embedded in lithium niobate has a higher gain in p polarization where it oscillates naturally. This is a requisite for exploiting the nonlinear component d₃₃ in quasi-phase-matched parametric generation.

OPOs with large cross section
The OPO crystals developed at the crystal growth laboratory at UAM were fabricated using a modified Czochralski technique, which creates the domain inversions during the crystal growth process.⁵ This technique provides OPO crystals with a larger cross-sectional area (several-square-mm aperture) than that forcrystals with periodical domain reversals created by electric-field poling (1 mm thick, at best). The Madrid research, however, indicates that creating a uniform domain structure in the crystal with the modified Czochralski technique is easier with Yb³⁺ than with Nd3+ ions, and also shows that Yb³⁺ ions incorporate into lithium niobate easily even if codoped with MgO.

The self-pumped OPO experiments used a quasihemispherical cavity. Nevertheless, deposition of dielectric mirrors on the crystal faces readily provides a diode-pumpable OPO microchip (see Fig. 2). The mirrors were highly reflective for the laser wavelength of the Yb³⁺ ions at 1063 nm. The pump wave at 980 nm was launched longitudinally through the flat input mirror. The output coupler and inner input mirror faces were transparent for the idler wave and reflective for signal waves between 1200 and 1500 nm. Output coupling of 1% was provided for the OPO signal wave. The crystal also had a domain period of approximately 25 µm with a 50% duty cycle. Under these conditions, it was possible to obtain parametric generation at 1360 nm.

Although the efficiency obtained in this first demonstration was very low—0.5 mW out of an extractable down-converted power of 30 mW—there is little doubt that these devices will reach essentially full conversion of the extractable power as two-crystal intracavity singly resonant OPOs do. As with other laser chips, the pump can be fiber-coupled to the chip.

Exploiting domain engineering
Engineering the domain structure provides a flexible tool for quasi-phase matching the parametric process in which idler and signal waves amplify at the expense of the pump-wave photons. The working wavelength of the device can be tailored by choosing the domain spatial period length. The modified Czochralski technique permits control of the final domain period in the crystal during the crystal growth process.

As in the case of birefringent phase-matching in single-domain lithium-niobate OPOs, quasi-phase matching is sensitive to temperature changes. In principle, obtaining a particular output wavelength in a stable manner requires crystal temperature stabilization. If, however, the cavity mirrors impose narrowband resonance conditions for idler and signal (doubly resonant OPO) or signal alone (singly resonant OPO), then it becomes possible to engineer an aperiodic domain structure in the crystal whose detuning curve for quasi-phase matching accounts for the mismatches caused by temperature changes, albeit at the expense of some nonlinear gain (see Fig. 3).⁶ One way to do this is to introduce a small random dispersion or chirp in the domain period length during the crystal growth process.

In many cases, if carefully designed, the resulting system will not require temperature stabilization. If there is enough nonlinearity—as in the case of exploiting d₃₃ in LiNbO₃— there may be extended mismatch bandwidth (tolerance) for quasi-phase matching and high-enough nonlinearity to realize efficient devices capable of extracting all the power available. If electrodes are evaporated onto the microchip, it also may be possible to produce rapid wavelength tuning in a small range or wavelength stabilization via the electro-optic effect.⁷ In principle, other ferroelectric crystals combined with laser ions are candidates for microchip OPOs; the race for efficient and versatile microchip OPOs has just begun.

ACKNOWLEDGMENTS
The authors acknowledge cooperation with Prof. L. E. Bausá·and Prof. J. GarcÍa Solé from GIEL (UAM) in spectroscopy and laser operation of Yb³⁺:LiNbO₃.

JUAN CAPMANY is a professor in the Dpto. FÃsica y Arquitectura de Computadores Universidad Miguel Hern×ndez Avda. Ferrocarril s/n E030202 Elche (Alicante), Spain; e-mail: [email protected]. ERNESTO DIÉGUEZ, VERÓNICA BERMÚDEZ and DAVID CALLEJO (Ph.D. student) are with the Crystal Growth Laboratory (LCC) Dpto. Física de Materiales C-IV Universidad Autónoma de Madrid Ciudad Universitaria de Cantoblanco s/n E28049 Madrid, Spain.

REFERENCES

1. J. J. Zhayhowski, Laser Focus World 32(4), 73 (April 1996).
2. M. Ebrahimzadeh et al., J. Opt. Soc. Am. B16, 1499 (1999).
3. A. R. Geiger et al., Opt. Lett. 21, 201 (1996).
4. E. Montoya et al., Appl. Phys. Lett. 74, 3113 (1999).
5. J. Capmany, V. Bermúdez, and E. Diéguez, Appl. Phys. Lett. 74, 1534 (1999).
6. M. M. Fejer et al., IEEE J. Quant. Electron. 28, 2631 (1982).
7. N. O'Brien et al., Opt. Lett. 24, 1750 (1999).