Attenuation of giant laserpulses by absorbing filters

Rory 0. Rice and James D. Macomber

Three types of absorbing filters were tested for suitability as reproducible attenuators for a Q-switchedruby laser beam. None of the filters tested were irreversibly damaged when irradiated with pulses havingpeak powers up to 40 MW (beam cross section 10-4 M2

). However, Tiffen Photar neutral density filtersand Corning filters showed reversible photobleaching (optical saturation) at large pulse energies, whileKodak Wratten neutral density filters did not saturate.

Introduction

We have tested three techniques to reduce quanti-tatively the intensity of optical pulses produced by aQ-switched ruby laser.

In principle, the best of these methods is attenua-tion by diffuse scattering. The laser beam is inci-dent upon a white reflector or transmitter having asurface that is rough on a wavelength scale. Suchsurfaces can be prepared in such a way that the re-flected or transmitted light obeys Lambert's lawquite closely. According to this law, the scattered in-tensity depends inversely upon the square of the dis-tance between the scattering surface and the sensi-tive portion of the photodetector. By fastening thescattering surface on a movable carriage that slidesalong an optical bench calibrated by means of a dis-tance scale, a quantitative and reproducible attenua-tor can be constructed.

There are, however, two practical disadvantages tosuch a device. First, the range of attenuation possi-ble is determined by the square of the ratio of themaximum distances of separation to the minimum.Even if the minimum is as small as 95 mm, a 3-m op-tical bench is required to produce a 3-order-of-mag-nitude range. Second, stray reflections from the un-attenuated beam can easily bypass the diffuser andimpinge directly upon the detector, obscuring the de-sired signal. Even though the entire room was paint-ed flat black to reduce the likelihood of such reflec-tions, and black felt curtains were used to isolate op-

When this work was done both authors were with LouisianaState University, Chemistry Department, Baton Rouge, Louisiana70803. R. 0. Rice is now with Mostek Corporation, Carrollton,Texas 75006.

Received 18 April 1975.

tical paths, reproducible results were never obtainedwith this method.

These problems caused us to investigate attenuat-ing filters that can be placed in a fixed position di-rectly in front of the photodetecting surface. Onetype of filter reflects the undesired portion of the in-cident pulse and transmits a fixed fraction into thedetector. The reflectivity is due to a thin multilayerdielectric coating supported by a glass plate. The re-fractive indices and thicknesses of the successivelayers are very carefully chosen to produce the de-sired transmittance. Unfortunately, filters havingcoatings of sufficiently high quality to be undamagedby 100-MW laser pulses are very expensive. If verymany different levels of attenuation are desired, theymay be achieved without great cost only by using asmall number of filters that can be combined in series(e.g., one, two, or three at a time).

Reflecting filters designed to operate at normal in-cidence are not satisfactory because their opticaldensities are not additive. The optical density D isdefined by

D log(1/T), (1)

where T is the transmittance. The optical density ofa group of filters at normal incidence is not equal tothe sum of the individual densities because of con-structive and destructive interferences between thewaves reflected and transmitted by individual filtersin the set. The amount of interference is very sensi-tively dependent upon the exact placement of the fil-ters. Reflecting filters designed to operate at anangle of incidence of 450 (dielectric beam splitters)could be combined in this way without interferenceeffects, but we did not possess enough of them to usethis attenuation method.

We were therefore finally led to the use of absorb-

September 1975 / Vol. 14, No. 9 / APPLIED OPTICS 2203

A -/ P

F2

G 1 ~ AG1-_7 L_A-- -F I

Fig. 1. First arrangement of apparatus used to test Tiffen Photarneutral density filters. L is the laser, F1 is a pair of Corning fil-ters, Al and A2 are apertures, Gi and G2 are glass flats, P1 andP2 are photodetectors, F2 is a stack of Tiffen filters preceded by

a glass diffuser, and C is the calorimeter.

ing filters. We tried three brands: Tiffen Photar(neutral density), Corning, and Kodak Wratten (neu-tral density). As mentioned above, a satisfactory at-tenuator for a laser must be able to withstand thehigh power light pulse without suffering irreversibledamage, and all three brands passed that test. Alsoa satisfactory attenuator must resist reversible pho-tobleaching processes such as optical saturation.The Tiffen and Corning filters failed this test; onlythe Kodak filters were satisfactory.

Apparatus

The experimental arrangement used to determinethe properties of the Tiffen filters is shown in Fig. 1.The ruby laser was manufactured by Optics Technol-ogy, Inc. (model 130). It was Q-switched by meansof a saturable absorber' (cryptocyanine in dilutemethanolic solution). The pumping energy andcryptocyanine concentration were adjusted to insurethat the laser fired only one pulse each time that thepumping lamps were flashed.

Two Corning 2-58 filters2 were used at the outputaperture in order to block the escape of pumpinglight (with wavelengths shorter than 700 nm) fromthe laser head. (It makes no difference if these fil-ters are saturable.) The beam diameter and direc-tion were controlled by means of a pair of apertures8.33 0.03 mm in diameter and 900 + 3 mm apart.

A glass flat at 45° incidence was used to deflect 8%of the beam into the aperture of photodetector 1 (Op-tics Technology model 620, S-20 surface). The out-put of this detector was fed to a Tektronix 535 oscil-loscope, and the traces were photographed by meansof a Tektronix C-31 camera with a Polaroid roll filmback, using ASA 10,000 speed film (type 410). Thesweep rate ranged between 5 usec/div and 25 gsec/div. This system was used to detect multiple pulseproduction by the laser, in the event that the pumpenergy and cryptocyanine concentration had beenimproperly adjusted. When an oscillogram showed

multiple pulse operation, the corresponding datumwas discarded.

A second glass flat divided the remaining portionof the beam between the calorimeter (Korad modelKJ-2) and the filters under test. The output of thecalorimeter (a thermocouple voltage) was measuredby means of a Keithley model 150A microvoltmeter-milliameter. Thermal stability of this system wasensured by insulating the body of the calorimeter andelectrical stability by inserting a Sola constant volt-age transformer (Catalog 30807) and a Powerstattype 116 voltage regulator between the microvoltmet-er and the 115-V power line. The output of theKeithley was recorded by means of an Esterline-Angus minigraph strip-chart recorder (modelMlA01 B4-000).

Light transmitted through the filters impingedupon the S-1 surface of photodetector 2 (TRG model105B). High voltage necessary for the operation ofthe photodiode was provided by four 510-V batteriesin series. The detector output was fed to the y de-flection plates of a Tektronix model 519 oscilloscope;the traces produced thereby were photographed bymeans of a Tektronix model C-27 camera with anoth-er Polaroid roll film back containing type 410 film.The areas under the oscilloscope traces (voltage vstime) were measured by means of a Lasico Compen-sating Polar Planimeter model 123A.

The experimental arrangement used to determinethe properties of the Kodak filters is shown in Fig. 2.The only essential difference between Figs. 1 and 2 isthat the TRG and OTI photodetectors have been in-terchanged (and relabelled P2 and P1), leaving theoscilloscopes in place. Also, the OTI detector wasmodified to use the 510-V batteries instead of theelectronic power supply provided by the manufactur-er, and the TRG detector was operated by a JohnFluke model 405 power supply. The S-20 surface onthe OTI photodetector is much more sensitive to

F 2

E /a \ B 2

B I </

-- A2

G i/ EF I A I

Fig. 2. Second arrangement of apparatus used to test KodakWratten neutral density filters. L is the laser, Fl is a pair of Cor-ning filters, Al and A2 are apertures, G is a glass plate, BI and B2are dielectric beam splitters, F2 is a stack of Kodak filters, and C

is the calorimeter.

2204 APPLIED OPTICS / Vol. 14, No. 9 / September 1975

6

5+ +

++ +

0"0 0.1 0.2 0.3 0.4

Calorimeter Energy

Fig. 3. Energy dependence of the saturation of Tiffen filters.Energy in joules, area in arbitrary units.

ruby laser photons ( = 694 nm) than is the S-1 sur-face on the TRG photodetector. Consequently, alarger fraction of the laser beam could be deflectedinto the rather insensitive calorimeter (by means of amultilayer dielectric filter oriented at a 45° anglewith respect to the optic axis) and still leave enoughpower to drive the Tektronix 519 oscilloscope.

The light that passed through the first reflector(4% of that striking it) is divided again by a seconddielectric beam splitter. The filters under test, fol-lowed by the OTI photodetector, received about 80%;the balance of 20% continued along the optic axis tobe used for other experiments that need not be de-scribed here.

ExperimentalThe total light energy absorbed by the calorimeter

is the product of the thermocouple voltage (whichvaries from one laser shot to another) with the con-stant sensitivity factor provided by Korad (0.0717 J/gV).

The energy striking photodetector 2 will be pro-portional to the area under the oscilloscope trace,provided that the rise time of the detection system ismuch shorter than the pulse duration. Since the risetimes of the photodetector and of the Tektronix 519are about 0.3 nsec each, the over-all rise time of thesystem should be about 0.5 nsec. The laser pulse du-rations varied somewhat from one shot to the nextbut all exceeded 11 nsec. Since 11 nsec is muchgreater than 0.5 nsec, the trace area was indeed anaccurate measure of the energy. Each oscillogramwas measured ten times with the mean accepted asthe true measure of the area.

If the filters do not saturate, the total energy inthat portion of the pulse striking photodetector 2should be proportional to the energy striking the cal-orimeter. The data presented in Fig. 3 were ob-tained with Tiffen neutral density filters in front ofphotodetector 2, using the experimental arrangementdisplayed in Fig. 1. It can be seen from that figure

that high energy laser pulses registered more heavilyin the photodetector than they did on the calorime-ter, indicating that the Tiffen neutral density filtershad been saturated.

Most mechanisms proposed for the saturation pro-cess suggest that the total energy delivered by thelaser should not be as important as the rate of deliv-ery (e.g., the peak power). The peak power strikingthe calorimeter is proportional to the calorimeter en-ergy divided by the pulse width (duration) at half-height, measured from the Tektronix 519 oscilloscopetrace. The proportionality factor depends upon the

140 - + ++10 - + . +120 - + + ++ 110 - + + *+ +10 0 -- +* + +I

70 _ + + +60 +

i I II

0 ICI 20 30

Peak Power in Megawatts40

Fig. 4. Power dependence of the saturation of Tiffen filters. Theordinate is ratio defined in Eq. (4) of the text; the units are V/MW.The abscissa is photodetector power defined in Eq. (3) of the text;

the units are MW.

90

80

70

60

a

dpdi

0V

S.30

20

10

0

+

+

0 1 2 3 4 5 6 7

Calorimeter Energy8

Fig. 5. Lack of saturation of Kodak filters. The line shown, y =(1.1 0.1) x + (86 2), is best fit in the least-squares sense. The

fit is weighted by separate error estimates for each datum-threevertical bars show typical estimates. Energy in mJ, area in arbi-trary units. The measured energies were reduced (by factors ap-propriate for the various beam splitters) to the amounts actuallystriking the filters. When the equivalent corrections are made forthe data displayed in Fig. 3, the range of pulse energies deliveredto the two different types of filters were found to be quite similar.

September 1975 / Vol. 14, No. 9 / APPLIED OPTICS 2205

a

. . I

shape of the laser pulse; however, since the shapes ofall the pulses were very similar the proportionalityfactor is constant and can be ignored. Also, most ofthe pulses had similar durations (10-40 nsec), al-though the amplitudes varied by as much as a factorof 50. For this reason, the data in Fig. 3 are consis-tent with the idea of power saturation. To provethis, we define

calorimeter power - calorimeter energy pulse width.(in MW) (in M J) (in nsec)

(2)Also,photodetector power(in arbitrary units)

- trace height Tiffen filter transmittance. (3)(in V) (measured at low power)

Finally,ratio - photodetector power calorimeter power. (4)

In Fig. 4, ratio is plotted vs photodetector powerusing the same data used to prepare Fig. 3. If thesaturation depended only upon pulse energy, but notupon pulse power, the points would fall on a horizon-tal line. Conversely, the tilt of the line in Fig. 4clearly indicates that power saturation has occurred.

Corning filters showed much the same kind of sat-uration as did the Tiffen filters. However, no at-tempt was made to collect enough data using them toprepare curves similar to those presented for Tiffenfilters in Figs. 3 and 4.

The data presented in Fig. 5 were obtained withKodak neutral density filters using the experimental

arrangement displayed in Fig. 2. The slope of theline shown was determined using a weighted leastsquares determination forced to the origin using themethods described by Young.3 The excellent fit ofthe points to this line indicates the absence of opticalsaturation in Kodak filters.

Conclusion

Kodak Wratten neutral density filters consist of acolloidal carbon dispersion in gelatin. Such a filtershould not and does not saturate. The disadvantageof these filters is that they are losing favor with pho-tographers to the tougher and more durable TiffenPhotar neutral density filters. Because of this, theKodak filters are no longer available in a mountedform. Considerable care must be taken in mountingthe filters because of the delicate nature of the gela-tin and glass flats, it is best to separate them inten-tionally in order to eliminate wide interference rings.One final advantage to the Kodak filters (besides thefact that they worked) is that they are by far the leastexpensive.

In any experiments using attenuating filters withhigh peak power lasers, reversible optical saturationis a possibility. Failure to ensure that the filters em-ployed are unsaturable may lead to erroneous conclu-sions being drawn from the results of such experi-ments.

References1. B. Soffer, J. Appl. Phys. 35, 2551 (1964).2. E. Garmire, Phys. Dept., MIT (now at EE Dept., Caltech); pri-

vate communication.3. H. D. Young, Statistical Treatment of Experimental Data

(McGraw-Hill, New York, 1962).

Fourth International Conference onRAMAN SPECTROSCOPY

August 25-29, 1974

ABSTRACTS OF TECHNICAL PAPERS

Authors' 200-word abstracts of the papers presented at thismeeting are now available. The 293 papers cover such topicsas: molecular structure, solid-state Raman scattering,instrumentation, band shapes, line profiles, polymer scat-tering, and nonlinear, electronic, and resonance Ramanscattering. The text is offset from the authors' copy: 48pages, 8-1/2 x 11 inches, side stitched.Order from:Raman Abstracts OSA members: $4.00 prepaidOptical Society of America 4.75 invoicedSuite 6202000 L Street, N. W. Nonmembers: $5.00 prepaidWashington, D. C. 20036 5.75 invoiced

2206 APPLIED OPTICS / Vol. 14, No. 9 / September 1975

,I