PPLN—Challenges of Commercialization

Periodically-poled lithium niobate suppliers must overcome damage and physical size issues before the PPLN market can reach full potential.

By: Glenn Nosti, Deltronic Crystal Industries, and Douglas Bamford, Gemfire Corp.

The laser industry has always used nonlinear frequency conversion to increase the range of available output wavelengths. Until recently, efficient nonlinear frequency conversion has only been possible for laser sources with high peak powers…and high purchase prices. During the past few years a new nonlinear optical material, periodically-poled lithium niobate (PPLN), has made possible the efficient nonlinear frequency conversion of less powerful, more economical laser sources, most notably sources based on semiconductor diode lasers (see Figure 1).

FIGURE 1: Periodically-poled lithium niobate can produce higher order harmonics at greater powers than bulk materials, and can be specifically designed to perform at a desired wavelength (Deltronic/Gemfire).

Every nonlinear frequency conversion technique must solve the problem of phase-matching: ensuring that the phase velocities of the interacting waves are properly matched, eliminating destructive interference. One solution to this problem, proposed in the early days of nonlinear optics, is quasi-phasematching (QPM), in which the proper phase relationship between the interacting waves is maintained by periodically reversing the orientation of the nonlinear optical material (see Figure 2).

FIGURE 2: In quasi-phasematching (QPM), the proper phase relationship between the interacting waves is maintained by periodically reversing the orientation of the nonlinear optical material.

Periodic poling
Practical implementation of the QPM scheme did not occur until many years later when appropriate techniques for patterning nonlinear optical materials with the required feature size— typically 1 to 15 µm—became available. In the case of ferroelectric materials, a periodically-varying electric field can be applied to a single crystal (which begins with all the ferroelectric domains pointing in the same direction) in such a way that the ferroelectric domains are periodically inverted This process is called periodic poling.

One method for creating a periodically-varying electric field is to apply a layer of photoresist to the top surface of a wafer, then lithographically pattern it into a series of rectangular bars (see Figure 3). The wafer is then connected to a pulsed high-voltage power supply using an aqueous electrolyte solution to provide electrical contact. The patterned photoresist, which is a dielectric, modulates the electric field near the surface in such a manner that the regions unprotected by the photoresist undergo domain reversal, while the protected regions remain unchanged.

Click here to see Figure 3.

The periodic poling is carried out at room temperature using applied electric fields greater than 20 kV/mm. After poling, the photoresist is removed and the wafer is diced up to produce a large number of PPLN chips. The periodic domain inversion is permanent unless the chips are subjected to very high temperatures (near 1000°C) or very high electric fields (comparable to the field used for poling).

Advantages of PPLN
Periodically-poled materials have significant advantages over conventional nonlinear optical materials. The poling period can be designed to produce phase matching for any nonlinear action involving wavelengths within the transparency range of the crystal.

In conventional materials, the crystal orientations that maximize the nonlinear optical coefficient are generally incompatible with phase-matching. In periodically-poled material, arbitrary interactions can be phase-matched, allowing the crystal to be oriented to maximize the value of the nonlinear optical coefficient. In lithium niobate, for example, the nonlinear optical coefficient increases by a factor of 4.5 when the conventionally phase-matched version of the material is replaced by the periodically-poled version.

In the limit of negligible pump beam depletion, the generated power for a crystal with a fixed length is thus increased by a factor of twenty. For input lasers with low peak powers (tens of milliwatts to tens of watts) this factor of twenty improvement can enable the production of useful frequency-converted power for the first time. Semiconductor diode lasers and diode-pumped solidstate lasers are prominent candidates for nonlinear frequency conversion in this low peak power regime.

PPLN fabrication
Researchers at Sony Corp. (Tokyo, Japan) first demonstrated electric field poling, a practical technique for producing PPLN, in 1993. Refinements were carried out at several research laboratories, including Stanford University (Stanford, CA), University of Southampton (Southhampton, England), Naval Research Laboratory (Washington, DC), Wright Laboratory (Dayton, OH), and Gemfire Corporation (formerly Deacon Research, Palo Alto, CA). The technique was scaled up to larger area, greater crystal thickness, better uniformity, and shorter QPM period.

Eventually the process became practical on a "wafer scale", patterning 75-mm to 100-mm-diameter, 0.5-mm thick lithium niobate wafers in a single step. This achievement raised the possibility of inexpensive PPLN chips, mass-produced from lithium niobate wafers in the same way that computer chips are manufactured from silicon wafers (see sidebar, Glimpses Into Commercialization of PPLN).

Challenges to mass production
Manufacturers face a number of issues before PPLN can achieve its full potential in the marketplace:

• Increasing the PPLN robustness to permit high-average power applications
• Increasing wafer thickness to accommodate larger input beam areas
• Decreasing poling period to permit short-wavelength applications
• Reducing cost

Periodically-poled lithium niobate can suffer damage in high-average-power applications at short wavelengths. In early frequency-doubling experiments with continuous-wave neodymium-doped yttrium aluminum garnet (Nd:YAG) lasers (at Stanford University and Aculight Corp. (Bothell, WA ), PPLN crystals cracked at output powers above 2 W. The mechanism of this crystal damage, which is accompanied by photorefractive beam distortion and green-emission-induced infrared absorption is poorly understood.

The limited thickness of the PPLN wafers forces the use of small beam areas, which limits the frequency-doubled output power achievable for pulsed systems. Although the surface damage threshold of PPLN is identical to that of conventional lithium niobate (1 to 3 J/cm² for a 10-ns pulse at 1064 nm), this threshold can be exceeded by some diode-pumped solid state lasers focused into half-millimeter-thick PPLN.

Chips with shorter poling periods must become available. Shorter periods, appropriate for the frequency-doubling of lasers with wavelengths between 800 nm and 1000 nm, cannot be produced by wafer-scale poling at this time.

Chips must become more economical. The low sales price implied by wafer-scale manufacturing has not been realized because most customers to date have been research workers with specialized needs purchasing one or two chips apiece. Until a "killer application" requiring large numbers of identical chips is found, it will not make economic sense for anyone to produce PPLN in large enough quantities to achieve significant economies of scale.

The solutions
Future work will be aimed at resolving these issues. Researchers are working to understand and control damage mechanisms at high average powers. One approach is to use chemically modified versions of lithium niobate, such as magnesium doped material and stoichiometric material, which could be less susceptible to this damage. Techniques for producing crystals with thicknesses of up to 1 mm have been demonstrated by research groups at Wright Laboratory and Lightwave Electronics (Mountain View, CA), and will be incorporated into commercial production.

Production of QPM periods as short as 3.2 µm, which can now be carried out with low manufacturing yields over centimeter-diameter areas, will be extended to higher yields and larger areas. One promising area of research is wafer-scale poling of lithium tantalate, a material that may be superior to lithium niobate for generation of short wavelengths.

PPLN will be sold to a variety of users worldwide, in the hope that its growing reputation as an efficient, engineered nonlinear optical material will eventually lead to the much-desired "killer application".

About the Authors:
Glenn Nosti is product manager of the crystal group at Deltronic Crystal Industries, 60 Harding Ave., Dover, NJ 07801; Tel: 973-361-2222; Fax: 973-361-0722. Douglas Bamford is director of the sensor program at Gemfire Corp., 2471 E. Bayshore Road, Suite 600, Palo Alto, CA 94303; Tel: 650-849-6831; Fax: 650-849-6900.

A Glimpse Into the Commercial Status of PPLN
Several companies are known to be mass-producing PPLN chips appropriate for nonlinear frequency conversion, but they have apparently chosen to sell vertically-integrated products containing the chips instead of selling the chips directly.

The potential applications for OEM products using commercially purchased PPLN are growing. Aculight Corp. has announced an optical parametric oscillator product incorporating commercially purchased PPLN, and several other laser companies are contemplating similar products.

Gemfire Corp. and Deltronic Crystal Industries have recently teamed to offer PPLN as a catalog item. A variety of QPM periods between 6 µm (appropriate for the frequency doubling of Nd:YAG lasers) and 30 µm (appropriate for optical parametric oscillators pumped by Nd:YAG lasers) are available in pieces 0.5-mm thick, and up to 50-mm long. The chip design can be customized for the customer's application. Previously, PPLN had been sold as a specialty item by both Crystal Technology and Gemfire Corp. Back to text