The speaker was asked how well the surface roughness was characterized and how well the presence of surface impurities was measured in these experiments. He commented that qualitative measures of surface roughness such as measurements of the root mean square deviation of the surface were not used, but that a characterization of rougher or smoother was obtained solely on the basis of visual inspection of electron migrographs. With regard to surface impurities some microprobe analysis, particularly auger analysis was used, but not to a very large degree. Nevertheless, he emphasized the main point that what is crucial in determining the damage resistance of a surface is not simply the surface roughness, no matter how it is characterized, but the history of surface preparation. There was considerable discussion of the difference between polishing methods which bring about a removal of a surface layer and those methods which smooth a surface by means of plastic flow and the creation of a subsurface layer of defects and impurities. It is to be expected that in the former case, higher damage will be observed. Mike Bass of the University of Southern California, offered some clarifying comments with regard to his work which had been referred to in Boling's talk. Bass commented that they took great pains to insure that their studies were carried out on small areas in order to avoid the ambiguity of large area irradiation. They found, he said, that a significant difference was observed in the morphology of surface damage between surfaces polished by conventional methods, which exhibited lower thresholds, and those polished by so-called super polishing techniques, which exhibited higher thresholds. He emphasized that in their work the term roughness or smoothness of the surface referred to small scale, effects observed on small illuminated areas. He indicated that it was crucial in obtaining consistent results to the experiments that the beam size at the damage surface be kept constant, that the focusing geometry be maintained. To this end, short focal length lenses were employed.

Surface damage in LiNbO3 has been studied at 1.06 um as a function of different surface treatments. While the use of ion beam polishing on LiNbO3 has resulted in general degradation of surface finish and decrease in damage resistance, a different type of plasma treatment shows a distinct improvement. This method of surface treatment is one in which the sample is bombarded in a low energy rf-excited plasma of argon and oxygen. Under this type of plasma treatment, the damage resistance is substantially improved (~ 50%) over that of the conventionally polished surface of the same sample. To avoid complications of possible cumulative effects in the damage experiments, each point on the surface is laser irradiated only once at any given power. The results are presented in terms of the fraction of surface which resists damage for a given incident energy.

That this improvement in threshold involves a surface-oxygen effect is supported by another series of preliminary measurements in which damage tests were performed on LiNbO3 at 10 to 15 atm 02 pressure. Under these conditions, the surface was usually seen to completely resist damage at levels of irradiation at least twice that at which damage occurred at ambient conditions or in a high pressure nitrogen environment.

Key words: LiNbO3; 1.06 μm; oxygen deficiency; surface damage; surface treatment.

The relatively low laser-induced surface damage threshold of lithium niobate, LiNbO3, poses stringent limitations on its use as a Q-switch or second harmonic generator. However, its large electrooptic tensor elements and its nonhygroscopic nature render it one of the most useful Q-switch crystals. Thus, a method for increasing its surface damage threshold would certainly be valuable.

1

It has been reported [1,2] that the surface of LiNbO3 is apparently deficient in oxygen although there are various speculations as to its exact role in the damage mechanism. This deficiency is hypothesized to result in greater absorption and thus a lower damage threshold at the surface than in the bulk material. Experimental verification of this hypothesis is presented herein.

It has been found that surface damage in LiNbO3 can be significantly reduced by supplying an excess of oxygen utilizing either of two methods. One method consisted of exposing the LiNbO3 to a low energy rf excited plasma of argon and oxygen. This appeared to supply an excess of oxygen at the near surface, resulting in a residual increase in the surface damage threshold. The other method consisted of surrounding the sample with pressurized oxygen, also resulting in a significantly increased surface damage threshold. However, when the sample was removed from the chamber, no residual effects were observed.

This work was supported in part by the Defense Advanced Research Projects Agency through Air Force Cambridge Research Laboratories.

'Figures in brackets indicate the literature references at the end of this paper.

2. Surface Damage Threshold Measurement Procedure

The technique employed for the determination of the damage thresholds in these experiments was developed to provide a quantitative threshold value and to detect small changes caused by the surface conditioning experiments. It was realized that there are several damage mechanisms in simultaneous operation, some of which are probabilistic in nature. Also, damage in LiNbO3 has been reported to be complicated by multiple shot cumulative effects. [1] For this reason, single pulse damage thresholds were sought. In addition, previous experience had shown that sample crystals of LiNbO3 available had significantly high densities of inclusions which obscured measurements of the intrinsic surface damage threshold.

It was assumed that each point on the surface was characterized by a damage threshold value lying between a power density below which no point on the surface would ever be damaged, and one above which every point on the surface would always be damaged. Thus a surface would have a range of damage thresholds which could be measured by firing the laser one time at a particular site, looking for the occurrence of damage, and moving to another site while keeping the laser output constant. A series of such measurements taken over the range of damage thresholds yields the probability that the single shot damage threshold lies below a given power density.

*

3+

In order to determine the surface damage threshold for a sample, a single shot Nd:YAG laser operating in the TEMO mode at 1.06 μm was focused through a 3.3 cm lens at the entrance surface of the LiNbO3 sample as shown in figure 1. The laser had a pulsewidth of 17.5 nsec (FWHM) and a focused spot size of 64.5 μm (1/e diameter for intensity). The output was varied by rotating crossed polarizers. The damage site was observed with a microscope focused at the LiNbO3 entrance surface utilizing an annular mirror oriented at 45° to the laser axis. The laser output was monitored with a beam splitter and an ITT biplanar photodiode whose output was electrically integrated and displayed on an oscilloscope.

3. Ion Bombardment Treatment

The threshold for laser induced surface damage in optical materials is determined in part by the physical condition of the surface. [3] Previous work by Giuliano [4] has shown that bombardment of sapphire crystal surfaces with energetic Art ion beams resulted in a substantial increase in the surface damage threshold over that of conventionally polished surfaces. Subsequently, attempts were made to similarly condition LiNbO, surfaces.

A series of experiments was performed to determine whether the surface of LiNbO3 could be made more resistant to laser damage. One mechanically polished surface was subjected to various conditioning plans in which time, energy, and current density of an incident Art ion beam were varied. The damage threshold of this surface was then measured and compared with that of another surface of the same crystal which had not been ion bombarded. The results showed that the ion bombarded surfaces were usually slightly more resistant to damage, although the increase in threshold was insignificant.

Considering the surface deficiency of oxygen in LiNbO3, an attempt was made to condition the surface by oxygen ion bombardment and increase the surface oxygen concentration. A surface was sputtered in an MRC rf plasma sputter system. The plasma pressure was 4 mTorr argon with 1 mTorr oxygen. The surface was sputtered for 30 min at 300 V bombardment energy. When the damage threshold of this surface was compared with that of an untreated surface on the crystal, it was found that a significant (~50%) increase resulted. The procedure was repeated on several crystals with similar results as summarized in figure 2. The time between sputtering and damage testing varied from 30 min to several weeks, indicating a retention of oxygen by the surface. From the data shown in figure 2 it can be seen that the treated and untreated surfaces have the same low power damage behavior, but differ significantly for high incident fluxes. Thus for the treated surface, the power density above which damage is observed on every shot is about 50% higher than for the untreated surface. This is interpreted to mean that the damage resistance is improved in the more resistant areas of the surface, (i.e., inclusion-free regions) while the treatment has no effect in improving the less resistant areas of the surface (included regions).

It was found that the power level below which no damage was ever observed was essentially a constant. However, the level above which damage was observed on every shot varied significantly from sample to sample and from conditioned to unconditioned surface on a particular sample.

*

Damage is defined here as being characterized by the appearance of a small crater or pit generally accompanied by a plasma spark.

For the damage tests performed in a pressurized oxygen environment, a windowed chamber was built to house the sample. The surface damage threshold measurement technique is identical with that outlined above, except for the addition of the chamber (see figure 1). With the chamber open to the atmosphere, damage threshold measurements were made. The chamber was then pressurized with oxygen to 250 psig and the same procedure repeated. In the 100 sites on 5 different crystals from one vendor, some of which were uncoated and some ThF4 coated, surface damage was rarely detected with the crystals in 250 psig oxygen. Only internal damage was usually seen at a level of approximately twice that for surface damage. However, crystals grown and polished by others did show more surface damage in the oxygen environment substantiating the need for high quality polishing techniques. [3] In addition, some surfaces coated with materials other than ThF4 showed significant surface damage. This suggests that ThF4 may be porous enough to allow oxygen to diffuse through it in order to be in contact with the LiNbO3

surface.

One important difference was noted between surface damage observed in an oxygen environment and that observed without. Without oxygen, surface damage rapidly avalanched to the catastrophic level. In pressurized oxygen, a surface damage site may begin to form, but it either becomes no worse with additional shots or worsens only slightly. It was never seen to rapidly avalanche into catastrophic damage,

as such a site would always do without oxygen.

In order to verify that the oxygen was responsible for the changed surface damage characteristics of LiNbO3 rather than some other effect due to the presence of pressurized gas, nitrogen was substituted for the oxygen. The pressurized nitrogen had no discernable effect upon the surface damage characteristics. As an additional test, another material, yttrium vanadate, YVO4, was substituted for the LiNbO3 in the pressurized oxygen with no perceptible change in the surface damage characteristics.

To determine the minimum pressure and exposure time required to produce an observable effect upon the damage threshold, these two parameters were individually varied. Minimum pressure appeared to be ~50 psig. Below this level, no observable improvement was detected. Above 100 psig, no additional improvement was seen, but the remaining tests, however, were conducted at 250 psig in order to be assured of maximum benefit. The time during which the sample was exposed to pressurized oxygen prior to being irradiated was varied from a few minutes to 24 hours with no perceptible difference. When the sample was removed from the oxygen, no residual damage resistance was observed.

5. Conclusions

The surface damage characteristics of LiNbO3 are significantly altered by the presence of oxygen. An increased surface damage threshold resulted as well as a marked tendency away from the usual rapid buildup of a damage site with each additional shot.

Two methods for introducing excess oxygen at the LiNbO3 surface were used. One method - bombarding the surface with an argon-oxygen plasma - had the residual effect of increasing the damage threshold after removal from the plasma. The other - immersing the sample in pressurized oxygen - had no residual effect. While in the oxygen, however, the surface damage threshold was increased and the tendency for a damage site to avalanche into catastrophic damage with additional shots was significantly reduced.

Certainly, additional work must be performed in order to more quantitatively describe the effects of oxygen on the surface damage threshold of LiNbO3. However, these results clearly indicate that it is significantly improved by the addition of surface oxygen.

The authors wish to express their appreciation to Dr. Eric Woodbury for discussions which led to the investigation of pressurized oxygen treatment of LiNbO3 and to Dr. H. Garvin for guidance and assistance in the sputtering processes.