Laser-induced damage measurements in CdTe and other
`ll—Vl materials
`
`M. J. Soileau, William E. Williams, Eric W. Van ‘Stryland,;and M. A. Woodall
`
`Results of laser-induced damage measurements in CdTe and other selected II—VI materials are reported.
`These studies were conducted using pulsed 1.06-um radiation from a Nd:YAG laser. The laser pulse width
`was varied from ~40 to 9000 psec (9 nsec). The laser-induced surface breakdown irradiance measured for
`CdTe over this pulse width range scaled as t;’/2 [tp is the laser pulse width (FWHM)]. This indicates that
`' laser-induced damage in this material is due to linear absorption by a thin surface contamination layer.
`
`I.
`
`Introduction
`
`The II—VI_ compounds and various single element
`semiconductors are used as window materials for con-
`tinuous output lasers operating in the 10.6-um region.
`These materials are also used as the high index of re-
`fraction component of multilayer dielectric coatings for
`laser and other infrared optics applications. While
`these materials have proved useful for nonlaser infrared
`applications and for many continuous output laser ap-
`plications, they are generally avoided in pulsed laser,
`systems.
`There are a variety of fundamental reasons to avoid
`the high index materials for pulsed laser applications.
`The high index means very high Fresnel losses at in-
`terfaces, and the field enhancement associated with
`surface scratches and other defects” is more pro-
`nounced. Bettis et al.3 have argued that high index
`materials will have lower damage thresholds based on
`local field considerations. Finally, Wang’s rule4 for
`nonlinear indices of refraction, n2, implies that a ma-
`terial with a large linear index will also have a large P12,
`and thus the problems associated with self-focusing will
`be most pronounced for high index materials.
`Despite these problems there are a variety of pulsed
`laser applications which require the use of the II-VI
`compounds, e.g., a system that has optics which are
`shared by lasers of different frequencies and different
`pulse widths. These materials are also of interest in
`integrated optics and phase conjugation; two applica-
`tions for which a large nonlinear index of refraction is
`advantageous.
`In any such application it is important
`to know the operating limits set by laser~induced
`damage to the materials.
`
`The authors are with North Texas State University, Center for
`Applied Quantum Electronics, Physics Department, Denton, Texas
`76203.
`Received 1'7 July 1982.
`0003-6935/82/224059-04$01.00/0.
`© 1982 Optical Society of America.
`
`In this study we measured the pulsed laser-induced
`surface damage threshold of CdTe, ZnSe, CdS, and
`ZnTe at 1.06 ,um. The study emphasized CdTe because
`of its use as a pulse shaping device5 in laser fusion sys-
`tems and other applications and because it is a good
`model system for use in studying nonlinear absorption.
`The band gap in CdTe is 1.32 eV, which lies between
`1.06 and 0.53 um. The position of the band gap in
`CdTe means that it is transparent to 1.06—um radiation
`but has a relatively large two-photon absorption coef-
`ficient. These experiments show that absorption in
`CdTe is dominated by two-photon absorption and ab-
`sorption by the two-photon generated excess carriers
`at 1.06 ,uIn near the damage threshold. However, the
`surface damage threshold irradiance scales as t;“2 (tp
`is the laser pulse width) which is characteristic of
`damage due to surface contamination.
`
`ll. Experimental
`
`The laser source for the picosecond studies was a
`passively mode-locked microprocessor-controllede
`Nd:YAG laser system operating at 1.06 pm. A single
`pulse of measured Gaussian spatial and temporal pulse
`shapes was switched from the mode—locked train and
`amplified. The temporal pulse width was Variable
`between 30 and 200 psec [full widths at half—maximum
`(FWHM)] by selecting various etalons as the output
`coupler. The width of each pulse was monitored by
`measuring the ratio R of the square of the energy in the
`fundamental (1.06 um) to the energy in the second
`harmonic, produced in a LiIO3 crystal. This ratio is
`directly proportional to the laser pulse width as long as
`the spatial profile remains unchanged.7 The ratio was
`calibrated by measuring the pulse width using type—I
`second~harmonic autocorrelation scans. The observed
`three—to-one signal-to—background ratios indicated
`clean mode locking.8 To ensure that the ratio R is
`proportional to the pulse width and provides a valid
`pulse width monitor, scans were performed for all out-
`put coupler etalons.
`The laser beam was focused onto the sample surface
`with a single element lens of best form design, i.e., de-
`signed for minimum spherical aberrations. The focal
`
`4059
`15 November 1982 / Vol. 21, No. 22 / APPLIED OPTICS
`ASML 1415
`
`

`
`1—GHz bandwidth oscilloscope. The focal radius of the
`Q—switched beam at the sample surface was 26 pm as
`calculated using Gaussian optics and the measured
`unfocused beam parameters. This laser system is de-
`scribed more completely elsewhere.“
`The CdS and ZnTe samples were single crystals. The
`CdTe was large grain-size polycrystalline material
`_ grown by chemical vapor deposition}?
`T
`III. Results and Discussions
`
`.
`
`The results of the picosecond surface damage mea-
`surements are summarized in Table I. The threshold
`values given are the peak irradiance levels, IB, which
`produce damage 50% of the time as determined using
`the procedure described by Porteus et al.13 In all the
`picosecond measurements the pulsewidth of each shot
`was measured and the irradiance calculated. Only
`shots within the pulse width range shown in Table I
`were used in determining the damage thresholds.
`The laser-induced damage observed in this work al-
`ways occurred on the front or entrance surface. The
`imaging lens for the picosecond measurements was used
`at f/116. This means that the depth of focus was very
`large and thus the beam radius was approximately the
`same at the front and exit surfaces. For this situation
`(neglecting absorption) the local electric field of the
`light beam at the rear surface is larger than that at the
`front surface by a factor 2n/(n + 1), where n is the index
`of refraction of the specimen.” Thus, the exit surface
`should fail at irradiance levels 4n2/(n + 1)? lower than
`the front surface. For CdTe this factor is ~2.3 and thus
`the exit surface should fail at irradiance levels equal to
`~0.43 times the front surface damage threshold.
`
`Table I. Surface Damage Threshold Dala
`
`Surface Damage Tnreshold Irradlance (GW/cmz)
`
`Material
`
`‘
`
`Focal Radius’ (l/e2 Point of the Irradlance)
`
`a) 1.24 1 0.04
`
`a) 0.82 i: 0.03
`
`
`tp—‘-1881:1595
`up = 1241: 10 ps
`
`
`b) 1.96 1: 0.10
`b) 1.18 1: 0.06
`
`0.10 :1: 0,01
`
`
`hp = 9000 :t 500 ps
`
`
`
`
`
`Cd’IVe
`
`
`
`cp=l0Bt10ps
`c) 2.06t0.10
`
`tP=42t3p5
`
`tp=46i:$ps
`
`0.43 1: 0.03
`
`
`
`
`
`
`up = 45 2!: 5 ps a) 12.4 i 0.9
`
`ZnSe
`
`tp=41:t3ps
`b) X4.9 1: 1.5
`
`l:p=34:t3ps
`
`
`
`1.0
`
`
`
`RELATIVEFl.UEN(2E‘an
`
`a7
`
` ...|_ '-'-.-.
`n
`-
`.
`,_
`lflll
`l25
`
`5
`125
`no
`75
`so
`25
`a
`2'5
`in
`_
`Posmnu [pm]
`.
`_
`V
`Fig. 1. Pinhole scan of the focused beam from the picosecond laser.
`The dots are the actual data points and the solid line is an ideal
`Gaussian of width 77 pm. After correction for the finite size of the
`. pinhole used to make this scan the focused beam radius was deter-
`mined as 75 :i: 5 pm.
`
`.
`
`position (547 :1: 2 mm from the lens mechanical center)
`was determined by a series of pinhole scans at various
`distances from the focusing lens. The 1/e2 radius of the
`irradiance of the unfocused beam was 2.35 mm resulting
`in an f/116 imaging system. The focal radius at the 1/e2
`point of the irradiance was 75 :l: 3 pm as determined by
`pinhole scans. Figure 1 is a plot of the beam scan data
`and a Gaussian best fit to the data. Some of the data
`were taken with the sample at the focal position and
`some data were taken with the specimen 2.70~cm behind
`the focal Dpint. The beam radius was calculated using
`Gaussian optics to be 139 :l: 4 ,u.m at a position 2.70~cm
`behind focus. The beam was not scanned at this posi»
`tion, however; it was scanned at 2.25—cm behind focus.
`The measured radius at this position was 122 i 7 ,um,
`which is within 5% of the 116—um value predicted by
`Gaussian optics.
`The beam irradiance was controlled by varying the
`voltage in the amplifier stage of the laser. Beam scans
`were conducted at various amplifier settings thereby
`verifying that the beam spatial profile was unchanged
`over the range of amplifier settings used. The pulse
`energy into the specimen and transmitted through the
`specimen was monitored for each laser shot.
`A piezoelectric transducer was mounted on the
`samples using pressure contact. The transducer was
`used to monitor the acoustic signal generated in the
`sample by linear and nonlinear absorption of light by
`the sample. The optoacoustic technique used in this
`experiment is described elsewhere? The optoacoustic
`signal and the transmission of the sample were moni-
`tored as the laser output was increased to a value which
`produced damage.
`In this experiment damage was
`taken to be any perceptible change in the same as
`viewed with a 20X microscope. The results of the
`nonlinear absorption measurements are discussed in
`detail in Ref. 10. The nanosecond data were taken
`using a Q~switched Nd:YAG laser operating at 1.06 pm.
`The pulse width for this device was 9-nsec (FWHM) as
`measured using a high speed photodiode read by a
`
`4060
`
`APPLIED OPTICS / Vol. 21, No. 22 / 15 November 1982
`
`

`
`
`
`'o.o
`
`1.2
`as
`0.4
`IRRAIJIAHBE (aw/cm 2;
`
`1.6
`
`2.0
`
`Fig. 2. Nonlinear absorption in CdTe. Here the transmission of
`~150-psec (FWHM) 1.06—,um pulses is plotted as a function of the
`input irradiance. The solid line is the theoretical curve for two-
`photon absorption. The dashed line is a fit including the absorption
`by photogenerated carriers using a two-photon absorption coefficient
`of 35 cm/GW and a carrier absorption cross section of 2 X 10”!“ cm”.
`The zero irradiance transmittance of this CdTe sample at 1.06 pm is
`0.63 and is determined by the Fresnel reflection losses at the sample‘
`surface.
`
`The fact that front surface damage precedes exit
`surface damage in these experiments can be understood
`in terms of the nonlinear absorption which preceeds
`damage in the material studied. Figure 2 is a plot of the
`transmission as a function of incident irradiance for
`CdTe. The laser pulse width for these data was ~15O
`psec. The points are actual data and the solid line the
`theoretical prediction based on two—photon absorp-
`tion.1°v15*17 The dashed line is a theoretical fit in-
`cluding the effects of photogenerated carriers with a
`two—photon absorption coefficient of 35 cm/GW and an
`overall excess carrier absorption cross section of 2 X
`10"15 cm2.1° The intercept (zero intensity) corresponds
`to the Fresnel losses and small linear absorption losses.
`Note that the transmission at 0.4 GW/cm? is down by
`more than a factor of 5 or ~2.5 times the loss due to
`Fresnel reflections and linear absorption. This means
`that, for intensities of 0.4 GW/crn2 and above, the field
`enhancement at the exit surface will be negated by the
`high nonlinear absorption. Note that the measured
`damage threshold for the front surface was ~1.2
`GW/cm? for this pulse width and thus the nonlinear
`absorption reduced the flux at the rear surface to a value
`below the measured front surface threshold. The
`nonlinear absorption data are discussed in greater detail
`in Ref. 10. The presence of the photogenerated carriers
`can further reduce the intensity at the rear surface by
`plasma def0cusing18”2° although for these thin samples
`such an effect is negligible.
`Nonlinear transmission and nonlinear optoacoustic
`measurements were made for all the materials listed in
`Table 1. Absorption in CdTe and ZnTe was dominated
`by two—photon absorption and subsequent photogen-
`erated carrier absorption, and absorption in CdS and
`ZnSe was dominated by three-photon absorption and
`subsequent photogenerated carrier absorption.”
`While the absorption in these specimens was dominated
`
`.
`
`by multiphoton processes, damage appeared to be due
`to linear absorption caused by surface defects and/or
`surface contamination as
`the following analysis
`shows.
`The data in Table I indicate that the surface damage
`threshold for CdTe with 40-psec pulses is independent
`of the focal radius to within the uncertainty of, the
`measurement. Bettis et al.21 have used a variety of
`surface damage threshold data to derive a scaling law
`‘ which predicts that the surface damage threshold ir-’
`radiance of dielectrics scales as 1/w, where w is the focal
`radius. However,‘their model is based on the spatial
`spreading of a laser-induced plasma andnoplasma was
`observed in these experiments (i.e.., no visible flash was
`observed upon damage in a darkened room). Exami— .
`nation of the damage sites and the observed pulse width
`dependence both indicate that damage in these mate-
`rials was associated with surface defects and contami-
`nation. We attribute the lack of spot size dependence
`to the fact that surfaces have a relatively high density
`of defects which initiate the damage, and thus the
`probability of finding a defect within the beam radius
`is essentially unity for the smallest spot size used (26-um
`radius).
`The CdTe data in Table I are plotted in Fig. 3. The
`solid line in Fig. 3 is a least-squares fit of the data to a
`t,",‘/2 dependence. Note that the fit to the data is very
`good over the entire 40-9000—psec range. The I3 de-
`pendence on the laser pulse width tp can be understood
`in terms of a simple thermal failure model. Suppose
`that there is a thin absorbing layer on the surface and
`damage is initiated by raising the temperature of this
`layer to a value which results in melting or an irrevers-
`ible phase change which changes the surface appear-
`ance.
`If such a layer has a thickness 5 which is less than
`the thermal diffusion depth (i.e., 5 < V tp , where p =
`thermal diffusivity of the material and $19 is the laser
`pulse width), heat is lost from the irradiated area during
`the laser pulse. The net result is that more energy is
`I“
`‘I
`l
`Non
`
`N=
`
`:..an
`
`r"G
`
`
`
` F’U! DAMAGETHRESHOLDIRRAUIANCEn09mm?)
`
`
`
`0.0
`
`0.0
`
`_u___
`0.5
`
`.
`1.0
`
`J
`1.5
`
`2.0
`
`[LASER PULSEWIDTHVV’ (105 sac"/1)
`
`Fig. 3. Front surface damage threshold irradiance for CdTe vs 6;“.
`The irradiance is given in GW/cmz, and Q71” is in units of 105 sec‘1/2.
`The solid line is the least—squares fit of the data to a t;"2 dependence
`(t,, is the laser pulse width, FWHM). All the data were taken with
`Nd:YAG lasers operating at 1.06 am. The box around the data point
`on the lower left is added for emphasis and does not correspond to the
`experimental uncertainty of this point (which is within the size of the
`data point because of the scale used here).
`
`15 November 1982 / Vol. 21, No. 22 / APPLlED OPTICS
`
`4061
`
`

`
`required to raise the surface temperature a given
`amount for relatively long pulses than for relatively
`short pulses. One-dimensional heat transfer calcula-
`tions (diffusion along the radius of these relatively large
`spots can be ignored) indicate that the threshold energy
`density should scale as t,1,’2 and the threshold irradiance
`should scale as t,;“'‘’.
`.V
`_
`Sparks and Duthler” have modeled thermal damage
`‘due to surface" inclusions.‘ Their solution‘ to the heat
`transport equation predicts that I3 cc t,;“"- for the case
`of metallic _inclusions foriwhich the absorption skin
`, depth is much less than the thermal diffusion depth.
`Their model assumes that the skin depth is much less
`than the inclusion radius and treats absorption by
`spherical inclusions the-same as absorption by a plane
`slab with a 6-function source. Thus, this model should
`work equally well for absorption by a thin contamina-
`tion layer (such as an oxide layer 10-100 A thick). The
`fact that the 15,31/zdependence is seen for pulses as short
`as 42 psec indicates that the absorption which leads to
`damage occurs in a layer of thickness 5 S 2000 A thick.
`Such absorption could be due to metallic inclusions or
`a thin surface contamination layer. Further evidence
`for the dominant role of surface contamination was the
`fact that chemical etching of the CdTe raised its surface
`damage threshold from 2 to 5.8 GW/cm? for 40-p~sec
`pulses.
`A final observation regarding these measurements is
`that we observed visible light emission from the CdS
`and ZnSe samples prior to damage. This visible light
`was emitted in the forward direction and was fairly well
`collimated. Spectral analysis of the emission from ZnSe
`revealed that the light was monochromatic and the
`wavelength was within 1 A (the limit of our spectrom-
`eter resolution) of frequency doubled 1.06 pm, i.e., 0.532
`nm. The ZnSe sample was polycrystalline with random
`crystallite orientation and the observed second har-
`monic generation was angle insensitive though relatively
`inefficient. The threshold for visual detection (viewed
`through 1.06-um laser safety goggles) of the second
`harmonic from the ZnSe was ~4 MW/cm2. This is
`substantially below the intensity required to burn Po-
`laroid film with 40-psec pulses. The relatively low
`threshold intensity for the generation of the second
`harmonic and the angle insensitivity make CVD ZnSe
`a relatively inexpensive, easy to use frequency doubling
`material where high efficiency is not needed. ZnSe thus
`makes a practical device for tracking and monitoring the
`presence of 1.06-um beams.
`
`IV. Summary
`
`The laser-induced surface damage threshold was
`measured for the front surface of CdTe and other II—VI
`materials at a variety of pulse widths and focal spot radii
`at 1.06 ,um. The results indicate that the damage
`threshold irradiance is independent of the focal radius
`and scales as 13;”2. The observed surface damage in
`CdTe appears to be caused by linear absorption by
`surface defects and/or surface contaminants. Two-
`photon absorption and excess carrier absorption by the
`two-photon generated carriers were determined to be
`
`4062
`
`APPLIED opncs / Vol. 21, No. 22 / 15 November 1982
`
`the dominant absorption mechanisms for CdTe and
`ZnTe and three—photon processes plus excess carriers
`were dominant for ZnSe and CdS. The depletion of the
`beam due to these nonlinear absorption processes pre-
`vented rear surface damage in all
`the specimens
`studied.
`
`This work was supported by the office of Naval Re-
`search, the National Science Foundation (-under grant
`7 ECS-8105513), The Robert A. Welch Foundation, and
`. a Faculty Research grant from North "Texas State
`University. _The data on CdTe ‘for 9-nsec pulses were
`taken at the Naval -Weapons Center, China Lake, Calif.
`The authors acknowledge the help of J. B. Franck in
`- setting up and characterizing the nanosecond laser used
`- for this test and for assisting in the measurements. We
`also acknowledge the support of Dwight Maxon and H.
`J. Mackey for providing the data acquisition and mi-
`crocomputer control system used in this work and Ar-
`thur L. Smirl
`for help in spectral emission
`measurements.
`
`References
`
`1. N. Bloembergen, Appl. Opt. 12, 661 (1973).
`2. P. A. Temple and M. J. Soileau, Natl. Bur. Stand. (U.S.) Spec.
`Publ. 462 (1976), p. 371.
`3. J. R. Bettis, A. Guenther, and A. Glass, Natl. Burr. Stand. (U.S.)
`Spec. Publ. 414 (1974), p. 214.
`4. C. C. Wang, Phys. Rev. B 2, 2045 (1970).
`5. R. Ozarski, Lawrence Livermore National Laboratory; private
`communication (1981).
`6. Quantel model YG-40, 928 Benecia Avenue, Sunnyvale, Calif.
`94086.
`
`599°
`
`7. W. H. Glenn and J M. J. Brienza, Appl. Phys. Lett. 10, 221
`(1967).
`D. J. Bradley and G. H. C. New, Proc. IEEE 62, 313 (1974).
`E. W. Van Stryland, and M. A. Woodall in Laser-Damage in
`Optical Materials: 1980, Natl. Bur. Stand. (U.S.) Spec. Pub].
`620 (1980), p. 50.
`10. E. W. Van Stryland, M. A. Woodall, M. J. Soileau, and W. E.
`Williams, in Proceedings, 1981 Conference on Laser-Induced
`Damage to Optical Materials, Boulder, Colo. (National Bureau
`of Standards, Washington, D.C., 1982).
`11. M. J. Soileau, Appl. Opt. 20, 1030 (1981).
`12. The ZnSe material was grown by the Raytheon Research Labo-
`ratories, Bedford, Mass. The CdTe was purchased from II—VI,
`Inc. Saxonburgh, Pa., and CdS and ZnTe were purchased from
`Cleveland Crystals, Cleveland, Ohio.
`13. J. O. Porteus, J. L. Jernigan, and W. N. Faith, Natl. Bur. Stand.
`(U.S.) Spec. Publ. 509 (1977), D. 507.
`14. M. D. Crisp, N. L. Boling, and G. Buge, Appl. Phys. Lett. 21, 354
`(1972).
`15. J. H. Bechtel and W. L. Smith, Phys. Rev. B 12, 3515 (1976).
`16. F. Brynkner, V. S. Dneprovskii, and V. S. Khattaton. Sov. J.
`Quantum Electron. 4, 6 (1975).
`17. M. Bass, E. W. Van Stryland, and A. F. Stewart, Appl. Phys. Lett.
`34, 142 (1979).
`18. R. H. Hellwarth, Natl. Bur. Stand. (U.S.) Spec. Publ. 341 (1970),
`p. 67.
`19. M. J. Soileau, M. Bass, and P. H. Klein, Natl. Bur. Stand. (U.S.)
`Spec. Pub]. 568 (1979), p. 497.
`20. A. A. Barshch, M. S. Brodin, and N. N. Krupa. Sov. J. Quantum
`Electron. 7, (9) 113 (1977).
`21. J. R. Bettis, R. A. House II, and A. H. Guenther, Natl. Bur. Stand.
`(U.S.) Spec. Publ. 462 (1976), p. 338.
`22. M. Sparks and C. D. Duthler, J. Appl. Phys. 44, 3038 (1973).