Imagine a multimillion-dollar laser system rendered useless by the absence of a nonlinear crystal needed for shifting the output to the required operating wavelength. This was the fear experienced by laser physicists at Sanders (Nashua, NH) about five years ago while developing a multiwatt, mid-infrared (IR) laser based on diode-pumped, solid-state 2-µm technology and a new nonlinear optical (NLO) semiconductor: zinc germanium phosphide (ZnGeP2).
The first ZnGeP2 optical parametric oscillator (OPO) crystal cut for 2-?m pumping showed outstanding performance, achieving 46% overall efficiency and more than 3 W of 3-5-µm (signal plus idler) output.1 But efforts to produce samples demonstrating similar performance were proving unsuccessful, and talk arose of cutting up the "golden crystal" to provide additional samples. By the time the fully packaged pump laser was ready to drive an OPO, however, ZnGeP2 crystal-growth yields and absorption losses had improved dramatically.
Chalcopyrites can be grown along the crystallographic orientation required for phase-matching, verified here by the author using x-ray diffraction prior to final fabrication.
null
The conversion efficiencies achieved with the lower-loss crystals were much higher than projected, so that system specifications were exceeded despite a slightly underpowered pump source. Similar experiences have occurred on more recent projects, as improving ZnGeP2-crystal quality continues to push OPO performance beyond initial design limits.
Chalcopyrites come of age
Success stories like this one are becoming more common as a result of focused research efforts to improve crystal growth and processing techniques for a class of IR materials known as chalcopyrites. These ternary compounds are derived from binary semiconductors by ordered substitution of cations from adjacent columns of the periodic table. These substitutions provide the birefringence required for phase-matching.
The attractive properties of chalcopyrites-high NLO coefficients and deep-IR transparency-have been known since the late 1960s, but two main obstacles limited their application. Reproducible growth of large, crack-free crystals was difficult to achieve due to highly anisotropic thermal expansion. In addition, absorption and/or scattering losses due to native defects severely limited optical transmission in the near- to mid-IR range. These obstacles have now been overcome by advances made at Sanders during the last ten years (see Laser Focus World, July 1995, p. 87).
A clear breakthrough
One of the most important breakthroughs to accelerate the development of chalcopyrites has been the use of transparent furnaces to grow crystals by the horizontal-gradient-freeze (HGF) technique (see Fig. 1).2 Presynthesized, stoichiometric starting material is loaded into a boat fitted with an oriented seed crystal, vacuum-encapsulated in a heavy-walled quartz ampoule and heated in a two-zone, horizontal transparent furnace until partial melting of the seed occurs. A low axial temperature gradient is established (which helps avoid cracking), and then the furnace is slowly cooled to produce a single crystal by directional solidification.
Horizontal-gradient-freeze growth using transparent furnaces greatly advanced the development of chalcopyrite crystals for infrared laser applications.
null
The viewing capability of this system greatly facilitates the seeding process, and, even after growth has been initiated, any secondary grains that appear can be immediately melted back and monocrystalline growth resumed. This capability is a tremendous advantage over traditional blind Bridgman growth, in which one might wait weeks to find that the seed had melted or polycrystallinity occurred early in the run. The technique is also valuable for rapid prototyping of new materials, especially when fundamental properties like the melting point are not known a priori.
ZnGeP2: setting the standard
ZnGeP2 was the first of the chalcopyrites to benefit from the growth approach described above, and the resulting crystals helped to set new standards for solid-state mid-IR laser technology. Since the introduction of the transparent furnace, ZnGeP2 crystal-growth yields climbed from 10% to 90%, and the overall crystal quality improved substantially. Crystals showed low wavefront distortion and high optical uniformity, so that 12-14 oriented OPO samples (6 x 6 x 14 mm3) could be extracted from each boule. More important, the broad near-IR absorption in ZnGeP2, which extends from the band edge at 620 nm to beyond 2 ?m, was reduced by a factor of five after post-growth annealing. The best samples to date have absorption-coefficient values of 1 cm-1 and 0.05 cm-1 at 1 ?m and 2.05 ?m, respectively, with losses below 0.01 cm-1 between 3 and 8 ?m. The corresponding OPO performance also improved dramatically (see Fig. 2); more than 10 W of mid-IR output power (signal plus idler) were generated from 20 W of input from a diode-pumped, Q-switched (10-kHz, 11-ns) Tm,Ho:YLF pump laser using a 6 x 6 x 14-mm ZnGeP2 crystal cut at a phase-matching angle of 54.5? with an a value of 0.09 cm-1 at 2.05 ?m.
The observed slope efficiency was 63%, but slopes as high as 85% have since been measured using a crystal with a 0.05-cm-1 absorption coefficient at 2.05 µm.3 The high TEM00 beam quality of the pump laser is preserved in the OPO output beam. Even at such high average powers, thermal lensing is minimal because of the very high thermal conductivity of ZnGeP2 (0.35 W/cm K).
AgGaSe2: a new direction
An alternative approach to generating high-average-power mid-IR radiation is to frequency-double the output from a CO2 laser. For this approach, AgGaSe2 emerges as the NLO material of choice. Even though AgGaSe2 has a smaller nonlinear coefficient (39 pm/V vs. 75 pm/V for ZnGeP2) and much lower thermal conductivity (0.011 W/cm K), its transmission losses at CO2 laser wavelengths (9.2-10.8 ?m) are substantially lower than those of ZnGeP2, which is fundamentally limited by high multiphonon absorption around 9 ?m. In addition, the lower nonlinearity of AgGaSe2 can be compensated for to some degree by using longer crystal lengths. To achieve these long lengths (o40 mm), HGF growth in a transparent furnace again offers some significant advantages.
Anisotropic thermal expansion is a major difficulty associated with the growth of AgGaSe2: the material contracts along the a axis but expands along the c axis (or optic axis) during cooling. For vertical Bridgman growth (the standard method applied to AgGaSe2), therefore, seeds must be oriented along the c axis so that expansion on cooldown occurs in the unconstrained vertical direction to avoid cracking the crystal and/or the quartz ampoule.4 Because the phase-matching conditions for CO2 SHG require samples to be oriented at 47° from the c axis in the [110] plane, the maximum crystal length is limited by the boule diameter.
Multiwatt OPO and SHG output powers have been produced from ZnGeP2, AgGaSe2, and CdGeAs2. The data here represent the state of the art for frequency-shifted mid-IR generation.
The HGF technique, however, allows for "phase-matched" crystal growth directly along the orientation required for devices, because the c axis can expand freely at any angle relative to the large exposed crystal surface. Phase-matched growth greatly increases the maximum sample length without scaling to larger diameters, simplifies sample fabrication, and optimizes the device yield per boule with minimal waste. Long antireflection-coated samples (5 x 5 x 37 mm) cut from 10-mm-diameter by 140-mm-long phase-matched AgGaSe2 boules have produced up to 3 W of second harmonic when pumped individually and up to 5.3 W when pumped in tandem by a 12-W, 9.27-µm CO2 laser (100-kHz, 10-ns, 200-?m spot). These high efficiencies and output powers from ZnGeP2 and AgGaSe2-based devices are rapidly moving from the laboratory into ruggedized military hardware. At the same time, the search continues for new NLO crystals offering even higher gains in the mid- to far-IR.
CdGeAs2: the ultimate IR crystal - Cadmium germanium arsenide (CdGeAs2) is in many ways the ultimate choice for long-wavelength applications-the "Holy Grail" of infrared NLO materials. It has an enormous nonlinear coefficient (d36 = 236 pm/V) that is more than three times higher than the nearest alternative (ZnGeP2) and translates into a full order-of-magnitude increase in nonlinear figure of merit (d2/n3). CdGeAs2 also exhibits a large birefringence, which allows phase-matching throughout its wide transparency range (2.3-18 µm), along with sufficient thermal conductivity (0.04 W/cm K) for high-average-power operation.
These outstanding properties have attracted many researchers, but only the development of HGF growth in low-gradient, transparent furnaces has made production of large, crack-free single crystals a reality.5 Absorption losses-which are highest in the mid-IR but affect the entire transparency range-are four times lower than the best samples reported previously, and progress continues as schemes to compensate native electronic defect levels are perfected.
The advances made thus far have allowed efficient (up to 28%), high-power (up to 2.3 W) frequency doubling at 77 K. This performance, achieved with a 13-mm-long CdGeAs2 crystal (at 77 K), approached that of two 37-mm-long AgGaSe2 crystals before rolling off at higher powers. Hopes of pushing this performance toward room temperature and of eliminating thermal roll-off are being fueled by the continuing improvements in crystal quality.6
Room-temperature difference-frequency generation has in fact already been demonstrated: efficient (up to 20%), continuously tunable (6-20 µm) radiation was generated from an uncoated, 6-mm-long CdGeAs2 sample by mixing the signal (4-5 µm) and idler (6.5-9.5 µm) output from an erbium-laser-pumped ZnGeP2 optical parametric generator.7
Birefringence engineering
An alternative to developing new materials with large nonlinear coefficients is to engineer the birefringence of existing materials to achieve phase-matching at an angle qm of 90?. This condition, known as noncritical phase-matching (NCPM), eliminates "walk-off," a phenomenon in birefringent materials whereby ordinary and extraordinary rays (in this case, the second-harmonic and fundamental beams, respectively) diverge as they pass through the crystal.
As the second-harmonic and fundamental beams diverge, the overlap of the two beams decreases until frequency conversion ceases altogether. Noncritical phase-matching preserves the overlap of the input and output beams so that frequency conversion occurs over the entire crystal length, allowing the use of longer crystals to extract more gain. Noncritical phase-matching also maximizes the effective nonlinear coefficient, deff, which is equal to d36sinqm. Because the overall conversion efficiency scales as the square of the length and as the square of deff, these effects can lead to dramatic performance enhancements.
In practice, the birefringence can be engineered by forming a solid solution of two end members with high and low values of Dn, respectively. For example, the birefringence of AgGaSe2 (which is more than enough to compensate for dispersion) can be reduced by alloying with AgInSe2 (which has very low birefringence but a slightly larger d-coefficient).
Large single crystals of AgGa1-xInxSe2 have been grown successfully over a range of indium concentrations using the HGF technique. Here the capability of this method for producing long phase-matched crystals is particularly valuable. Noncritical phase-matching was demonstrated for 9.27-µm-pumped SHG in the crystal grown with 42% indium. Work is ongoing using a modified growth technique to improve the overall crystal quality.
In summary, major innovations in chalcopyrite crystal growth have advanced these materials from laboratory curiosities to reliable system components in high-power mid-IR lasers. Such lasers continue to benefit from ongoing research to minimize absorption losses and engineer the optical properties for optimal performance in specific frequency-conversion applications.
ACKNOWLEDGMENTS
The author is grateful to the Air Force Research Laboratory Materials Directorate (WPAFB, OH) for its long-term support of this work.
REFERENCES
- M. G. Knights et al., in Advanced Solid State Lasers, Technical Digest, 1994, Optical Society of America, Washington, DC, 259 (1994).
- P. G. Schunemann and T. M. Pollak, U.S. Patent No. 5,611,856 (March 18, 1997).
- P. A. Budni et al., in OSA Trends in Optics and Photonics 19, W. R. Bosenberg and M. M. Fejer, eds., Optical Society of America, Washington, DC, 226 (1998).
- G. W. Iseler, J. Crystal Growth 41,146 (1977).
- P. G. Schunemann and T. M. Pollak, J. Crystal Growth 174, 272 (1997).
- For latest advances, see www.sanders.com/lasers.
- K. L. Vodopyanov and P. G. Schunemann, Opt. Lett. 23, 1096 (1998).
PETER SCHUNEMANN is principal physicist at Sanders, a Lockheed Martin Company, POB 868, Mer 15-1813, Nashua, NH, 03061; e-mail: [email protected].