(Pr,

Figure 7: Simulated packing density in a layer system consisting of a dense layer (Pgr, particle diameter 1.5, mobility value 1.3) and layer with lower packing density particle diameter 1.2, mobility value 1.1). A density of 100 % corresponds to the closest packed structure. P, is the sum of Pr and Pgr.

Figure 8: Simulated packing density of two layers with properties equal to figure 7. The interface between the layers is codeposited. It consists of 6000 particles which are concentrated according to a linear alteration of the deposition rate in the interface region.

Manuscript Received

1-17-89

Measurements of Pulse Damage Thresholds
of AR Coated CdTe at 10.6μm

J. G. Grimm, R. S. Eng, C. Freed, N. W. Harris

Massachusetts Institute of Technology
Lincoln Laboratory

P.O. Box 73

Lexington, MA 02173-0073

R. G. O'Donnell

Ford Aerospace Corp., Lexington, MA 02173

Laser induced damage thresholds (LIDT) measurement results on AR-coated single and polycrystalline CdTe samples using 35μs flat top pulses from a CO2 laser MOPA system are reported. Single-shot LIDT's are in excess of 50 J/cm2. The LIDT's for cumulative pulses in the 50k shots regime and pulse repetition frequency in the 1-5Hz range have been measured. The LIDT has been found to be dependent approximately on the square root of the pulse width. The problem of protecting ARcoatings from aqueous solution has also been investigated.

Key Words: anti-reflection coatings; cadmium telluride; CO2 laser MOPA; laser induced damage threshold; pulse repetition frequency effect.

1. Introduction

The laser induced damage thresholds of a number of semiconductor materials transparent in the IR spectral region have been measured and reported in the open literature. Despite this there are still many unanswered questions on the dependence of laser induced damage threshold (LIDT) of anti-reflection coatings on pulse length, pulse repetition frequency (prf), and cumulative number of shots. We would like to report our method of generating nearly flat-top medium energy CO2 laser pulses of large width and our measurements of the damage thresholds of a number of AR-coated CdTe samples using these wide laser pulses. Single-shot laser induced damage thresholds of AR coated samples exceeding 50J/cm2 have been observed. We have investigated prf and cumulative laser pulse effects on LIDT. Some of the samples have been tested successfully for many thousand laser shots without any observable surface damage.

2. Experimental Setup

There are three major components for the damage test measurement setup, namely, the laser beam from a laser master oscillator-power amplifier (MOPA) system, the damage test sample holder, and the damage monitoring instruments. The CO2 laser beam used in the damage testing station basically comes out of a small laser MOPA system. The output pulse width can be adjusted from 5us to 80us prior to any measurements. For the 35μs wide pulse, in particular, the available output energy is about 230mJ. Mode matching this output pulse to a spot size (1/e2 diameter) of about 0.79mm produces a peak energy density of over 90J/cm2 at the center of the damage test housing for LIDT testing of CdTe crystals (or other infrared optical components). The pulse energy is approximately proportional to the pulse width. Based on our previous work,1 this level of energy should be sufficient for LIDT measurements of AR-coated crystals available at this time.

The main components of the CO2 laser MOPA system are shown in a block diagram in figure 1. A lowpressure hybrid TE CO2 laser, consisting of a cw dc-discharge gain cell in series with a custom made commercial pulsed-discharge gain cell (Pulse Systems, Inc. Model Dual LP-30 laser gain cell) sharing the same stable resonator configuration laser cavity,2 produces a smooth single TEMq00 mode, 80mJ pulsed This work was sponsored by the Department of the Navy for SDIO under contract F19628-85-C-0002. "The views expressed are those of the author and do not reflect the official policy or position of the U.S. Government.'

output with a pulse width of about 70μs (FWHM). Within limits, some degree of pulse width control can be achieved by varying the pressures and mixing ratios of the gas constituents, charging voltages and energy storage capacitors.

The output beam size of the hybrid laser is focussed to a beam waist between two acousto-optics (A-O) cells. They are arranged so that there is no resultant frequency shift of the beam. The output beam of the second modulator is next transformed to a beam waist of about 5mm at the entrance plane of the 5:1 beam expander. The output of the beam expander goes to the pulsed low-pressure CO2 laser amplifier, a modified version of a Model LP-140 from Pulse Systems, Inc. The modified version incorporates electronic triggering circuits for sequentially initiating the dual discharges. This pulsed amplifier has an active gain length of 140cm. The output from the laser amplifier is about 400mJ (full unsliced pulse).

Part of this beam is split off with a 95% beamsplitter and used as an energy monitor. The rest of this beam is focussed by a combination spherical mirror and a lens to a beam spot diameter of 0.79mm at the center of the damage test housing shown in figure 2. The housing is purged constantly with dry filtered N2 during the damage test measurements. The windows are two inch diameter Ge plates with both sides AR coated.

A photomultiplier tube (PMT) is situated looking into the housing at the front surface of the test sample to detect any visible emissions given off as a precursor to damage. A variable attenuator consisting of a rotatable half wave plate and a polarizer is placed before the housing, providing convenient adjustment of the laser pulse energy. The beam expands as it comes out of the damage test housing and ~3% of the beam is split off and is focussed down to a spot less than 1mm in diameter for temporal envelope detection with a highspeed photoelectromagnetic detector. The pulse energy is measured with an energy meter made by Gentec Inc. with a 1 inch diameter sensitive area. A 2 inch x 2 inch Gentec energy meter is used behind the beamsplitter to monitor the output of the LP-140 laser amplifier. Both of these meters have been calibrated with a Scientech power meter operating in the energy mode before the start of the pulse energy measurements.

All the pulsed systems are synchronized through appropriately timed trigger pulses from a pulse generator, Berkeley Nucleonic Model BNC-8010, which drives several pulse delay generators, Ortec Model 416A, for optimization of time delays to the various trigger inputs.

Figure 3a shows a photograph of 50 consecutive energy pulses superimposed measured at the output of the hybrid TE laser. The pulse repetition frequency was 1Hz. Thus we can estimate that the peak-to-peak fluctuation in the output pulse energy is less than ±10%. Figure 3b shows the nearly flat top pulse at the output of the laser amplifier. The pulse width is about 35μs (FWHM).

The spatial profile of the laser output was measured at several locations in the optical path using a 32 x 32 pyroelectric detector array, Spiricon Mode LMP-32 x 32. This was to check the quality of the beam insuring a proper spot at the focus on the test sample. Figure 4 shows two views of the laser beam spatial profile at the output of the pulsed amplifier. Figure 4a is an isometric view of this, and figure 4b shows a slice of data taken through the center which also has a Gaussian curve fitted to it. The solid curve is the data recorded by the array, and dashed curve represents the fitted data. From this it can be seen that we operated with a very nearly Gaussian beam. Based on the above plots of the beam profile, the following formula, which is appropriate for a Gaussian beam, is then used to determine the beam energy density, J, at the center of the beam:

where Et is the total pulse energy and w is the beam spot radius. By measuring the beam size and pulse energy. the peak or on-axis energy density was determined.

The input beam to the LP-140 is about 2.5cm in diameter at the 1/e2 point, therefore, no beam truncation is expected for the 4 x 4 cm2 clear aperture of gain cross section. By examining the input and output beam profiles recorded with the Spiricon 32 x 32 array detector, we see that the beam profile changes very little and remains essentially Gaussian after going through the amplifier for this level of energy extraction. Table I lists the trigger delays used to achieve the highest energy output for a 35μs pulse with minimum temporal variation of pulse energy.

Typically at the start of each measurement, the area of the sample to be targeted was set up and a transmission measurement was made to check the quality of the sample. A series of conditioning runs were made at prf = 1Hz. For each of these conditioning runs the sample was irradiated up to 1000 shots before increasing the fluence level (in general by 5J/cm2 steps). At the start of each level of conditioning visible emissions were often observed (by the photomultiplier) but they gradually grew weaker as time increased. The emissions usually stopped within the first 100 shots. Upon the completion of each conditioning level the sample was removed and inspected under a microscope. No damage was observed in relation to the PMT signals that faded within a hundred shots. This was continued until threshold was reached.

3.2 Results

3.2.1 Single-Shot Damage Thresholds

Table II shows some typical results of LIDT values measured on both polycrystalline and single crystal CdTe test samples. As can be noted there is a large variation in thresholds of the polycrystalline substrate tested. Originally this was thought to be due to the irradiation of crystal boundaries. The mechanism here was that at the boundary of a single crystal cell impurities would collect and/or the boundary layer would be tellurium-rich and absorb more incident radiation. This would lead to a damage center forming perhaps in the interior of the substrate.

*P-1 stands for a polycrystalline witness sample; S-1 stands for a single crystal test sample

We observed two different types of damage, believed to be caused by different damage mechanisms. A typical damage site of the first type is shown in figure 5a, it occurs at the AR-coating only. A typical damage site of the second type is shown in figure 5b, it occurs in the substrate. The typical sizes of the two types are

approximately 1/3mm and 1mm, respectively. The damage sites for both types of damage were identified during the testing by lower transmission associated with large signals from the PMT.

Figures 6a and 6b show the times of damage corresponding to the first and second types of damage, respectively. In both figures, the damages had already occurred during the preceding laser pulses and the transmission through the sample as shown are relatively low. Since the second type of damage was of more catastrophic nature, the drop in transmission as shown in figure 6b is nearly 90% which is more severe than that for the first type of damage. The comparatively late occurrence of the PMT signal may be due to a higher required energy to heat the substrate to the damage threshold. The PMT signals associated with the second type of failure were always larger than those of the first type.

These two types of damage were strongly correlated with a particular surface. The failure of the AR coating in the first type of damage occurred more often on the input surface and there were fewer of these failures altogether. This may imply that this type of failure is due to random imperfections in the AR coating. The second type of damage occurred exclusively on the output surface. Moreover, these damage sites are conically shaped pointing back into the substrate up to a millimeter in depth.

3.2.2 Contamination due to Protective Coating

In the course of routine handling, these crystals were given a protective optical coating of Microstop to keep contaminants and solvents from the AR coating that could possibly damage or remove it. We set out to take a series of measurements to find if the application of this coating had any effect on the LIDT after it was removed. These results can be seen in Table III. On average the sample P-1 damaged at lower fluence levels than the pre-treatment surface. The method of cleaning for the above series was to dissolve off the protective coating using several baths of acetone, changing the acetone each time. Then after the surface appeared to be free of the material to the naked eye there was a final rinse with clean acetone and this was then blotted off the surface before being allowed to dry. We are also trying another rinse of ethyl alcohol which is then blown off with dry N2.

We have measured the laser-induced damage thresholds of several AR coated CdTe samples a function of pulse width in the long pulse regime. Figure 7 shows the plot of the LIDT's of the CdTe samples vs. laser pulse width ranging from Ins to several tens of microseconds. At the long pulse regime are our measurement results using the method just described; they represent the first of its kind in LIDT measurements using truly flat-top CO2 laser pulses, against the method of averaging the power of multiple sub-damage-level picosecond pulses from the free electron lasers recently reported.3.4 At the short and intermediate pulse width regime, LIDT data from several sources are included. 5,6,7,8 We see that the plot tends to confirm that the dependence of LIDT on the pulse width (t) for CdTe is approximately proportional of t1/2 out to the long pulse regime. This probably can be explained by the fact that the thermal diffusivity of CdTe is among the poorest in semiconductors; that is the thermal heating process is diffusion dominated. The present result on the pulse width dependence implies that the most appropriate damage model for AR-coated CdTe samples is that of impurity inclusion in the AR-coated layers. The square root of the pulse width dependence for the LIDT presently obtained allows designers in laser modulator or related fields, for the first time, to scale the LIDT in the long pulse regime with a much higher degree of confidence.