Digitally Pressure-Scanned Fabry-Perot Interferometerfor Studying Weak Spectral Lines

V. G. Cooper, B. K. Gupta, and A. D. May

A system is described for digitally scanning a Fabry-Perot interferometer by changing the pressure ofthe gas in the interferometer chamber by small discrete steps. The system, intended for recording weaksignals, i.e., of the order of a few counts per second, is very linear. Repetitive scans are reproducedwithin -6 of the spectral free range of the interferometer. An example of the depolarized Rayleighline scattered by H2 at 2.5 amagat is given.

The Fabry-Perot interferometer coupled with photon-counting electronics has been widely used for high pre-cision spectroscopy. Two types of scanning are oftenemployed. In one the pressure (density) of the gas inthe Fabry-Perot chamber is continuously varied, whilein the other a voltage-swept piezoelectric mount forone of the interferometer plates is used. For veryweak signals, say of the order of 10 counts per second,there exists a conflict between adequate signal-to-noiseratio and high resolution. The long integration timenecessary for high signal-to-noise ratio will, in con-tinuously scanning systems, lead to a distortion of theprofile, that at best is undesirable and may lead to aloss of detail in the line shape. Obviously a digitalsystem where the frequency of transmission of theFabry-Perot interferometer is fixed during the time ofintegration does not suffer this drawback. At themoment, the development of digitally scanned piezo-electric systems is in its infancy. Problems exist con-cerning the linearity of the sweep, the constancy offinesse with sweep, and long-term stability, all ofwhich are essential for high resolution studies of weaklines. Although actively engaged in the developmentof such interferometers, we have for the past severalyears resorted to the other possible solution, viz., theuse of a digital pressure-scanned Fabry-Perot inter-ferometer. It is the purpose of this note to describethis extremely stable and linear instrument. As anexample we limit ourselves to the description of asystem used to measure the profile of the depolarizedRayleigh light scattered by H2 gas (Gupta and May')

The main components of the system are shown

All authors were at the University of Toronto, Physics Depart-ment, Toronto, when this work was done; V. G. Cooper is nowwith the Hebrew University of Jerusalem, and B. K. Gupta iswith the Physics Department of Hans-Raj College, Delhi.

Received 20 December 1971.

schematically in Fig. 1. Light from a laser, scatteredby the gas in the scattering cell, is analyzed with aparallel-plate Fabry-Perot interferometer and con-ventional photon-counting electronics. For a de-scription of the optical system that is peculiar to thespecific problem being studied, the reader is referred toGupta.2 The pulses are counted by a scaler (Ortecmodel 431) and then printed and punched on tape by acombination of a modified Teletype (model 33 ASR)page printer and a printout control unit (Ortec model432). The time base unit provides pulses at regularintervals r for the programming unit. Over a period of16r the output of the programming unit consists offour properly synchronized signals, d, a, b, and c, tocontrol four different operations, pressure stepping,data printing and punching, scaler gating, and theoperation of the shutter for the laser beam. The systemis designed to have alternate, but not necessarilyequal, periods of time in which the signal and darkcount are recorded. If the dark count can be ignoredor taken as a small constant correction to the signal,the logic circuit described below could be considerablysimplified. Our system is designed for signals that arecomparable to the dark count of a cooled ITT FW 130,i.e., of the order of 1 or 2 counts per second.

Figure 2 shows the details of the programming unit.Pulses from the time base unit were fed to a 4-bitbinary counter, B (DM8533N), and then to a decoder,D. Although we constructed our own decoder frominverters (DM8004N), nand gates (DM8000N), andnor gates (DM8002N), a decoder unit (SN74154) isnow commercially available. There are sixteen outputterminals from the decoder, each of which goes fromlogic zero level (0.2 V) to logic 1 level (3.5 V) for aperiod , once during a complete cycle. During thetime , all the other terminals remain at level zero.Two appropriately chosen logic pulses, x and y, fromthe total of sixteen available are used to generatecontrol pulses for channels a, b, c, and d.

October 1972 / Vol. 11, No. 10 / APPLIED OPTICS 2265

Fig. 1. Schematic diagram of the digitallypressure-scanned Fabry-Perot interferometer.

The time 16T for a complete cycle is divided betweenthe time to collect the dark count of the photomultiplierto print this count, to collect the signal count,.and toprint that count. The printout time was kept to aminimum of T for each operation or a total of 2

T fora single cycle. The remaining time, 14T, was dividedNT/N 2 T between the dark count and signal count timewith N1 + N2 equal to 14. As the operation is cyclical,it is possible to start at any point to describe theoperations occurring in a cycle. We shall start at thetime that a signal from terminal b starts the scaler torecord the scattered light intensity, i.e., when terminaly returns to logic zero level. At a later time, say OT,the counter is stopped and the printout initiated bypulses from outputs b and c, respectively. Thisoccurs when terminal x goes to logic 1 level. At thesame time a signal from terminal c activates theshutter to block the laser beam and a pulse fromterminal d commands the pressure control system toadjust and lock on to a new value of pressure in theFabry-Perot chamber. A period of 1 later, whenterminal x returns to logic zero level, the scaler is re-activated by a signal from b, thus permitting the darkcount of the photomultiplier to be accumulated. Aperiod 4

T later the scaler would be turned off and theprintout initiated. At the same time the shutter wouldbe removed to permit the laser light to enter the scat-tering cell. Finally ir after this, the scaler would beturned on and the signal count accumulated for thenext cycle of operation.

Although the operation of the circuit is straight-forward, there are a few points worth mentioning.Terminal b is connected to the gate of the scaler,which means that counting is suppressed wheneverpoint b is grounded, i.e., for the time that terminalx or y is at logic 1 level and the transistor (type 2N3704)conducts. Switch S1 is used to change the shutterposition so that the interval between the time thatterminal y returns to logic zero and the time thatterminal x goes to logic 1 level corresponds to accu-

mulating, say, the signal and not the dark count aloneof the photomultiplier. In the spectra we haveexamined the instrumental profile is considerablynarrower than the spectral line being studied. Thus weset the pressure step to permit an accurate measure ofthe narrow profile. For the broader profile we recordonly every second step by double-pulsing the steppingmotor. This is accomplished simply by bypassing withswitch S2 the inverters and one-half of the flip-flop F.Finally the resetting of the scaler to zero counts isaccomplished, not by the programming unit, but bythe interconnections between the printout control andthe scaler.

Both the pressure step and the laser shutter lag theactual pulses by some random interval, as they are con-trolled by electromechanical relays. This does notaffect the collected counts, as the scaler is gated offfor a time that is much larger than this time lag. Inpractice the minimum value for is dictated by thetime taken by the Teletype to print and punch the

+5v

_+5 __ _'50o~~

a

Fig. 2. A simple programming unit for the interferometer.

2266 APPLIED OPTICS / Vol. 11, No. 10 / October 1972

Vacuum

2.5 VAC

4

Hg

Fig. 3. The stepping and servosystem for the Fabry-Perotinterferometer.

data and to return from the end of a line to the be-ginning of the next line. This is close to 2 sec. Thetotal delays in electronic components were estimatedto be of the order of 100 ,usec and hence negligible.The gating off and on of the scaler thus provides anaccurate control of the counting periods. In addition,at each stepping of the pressure the servosystem takes2-4 sec to establish a constant pressure in the inter-ferometer. Hence it is essential that the pressure bestepped at the beginning of a dark count period whenthe interferometer is inoperative. There are severalreasons why we have used a sixteen-terminal ratherthan, say, a ten-terminal decoder. One is that thescaler is inoperative for only a period of 2 in 167or - of the time. Another is the flexibility in choosingthe relative lengths of time spent in counting signaland dark count.

For weak signals the basic time interval is of theorder of seconds. Although a commercial scaler-timercan be used as the time base, we have constructed astable and inexpensive time base from a pendulum(grandfather clock courtesy of H. Benzing, Don Mills,Ontario) that cuts the light falling on a photodiode.The signal from the photodiode is amplified andshaped. The period r can be varied either by changingthe length of the pendulum or by frequency dividingwith a selection of binary counters.

The real success of the digital pressure-scannedsystem is due to the linearity and stability of thepressure control unit shown in Fig. 3. The pressure ofthe gas in the Fabry-Perot chamber (F.P.) was moni-tored by means of a mercury U-tube manometer withtwo uniform platinum wires fixed along the axes of thetwo arms. The space above the mercury in the largerarm was evacuated while that in the shorter arm wasconnected to the Fabry-Perot chamber and a cylinderfitted with a movable .airtight piston. The platinumwire in the manometer forms one arm of an ac bridgewith mercury acting as the sliding contact. Theother arm of the bridge is the ten-turn 100-g precisionpotentiometer with a variable 400-S resistor connectedin series on each side. The bridge is excited by 60

cycles, 2.5 V. The out-of-balance voltage between themercury and the moving contact of the potentiometeris amplified and used to drive a servomotor S (Honey-well 362479-10) coupled to the movable piston. Thusthe pressure in the Fabry-Perot chamber is maintained ata value determined by the position of the moving contactof the ten-turn potentiometer. As mentioned above,the stepwise scan of the pressure that is accomplishedby stepping this contact through 100 with the steppingmotor R is controlled by the programming unit. Thepurpose of variable resistances (0-400 2) connected inseries with the potentiometer is to choose the positionand the range of the total pressure change correspondingto a total of 360 steps of the ten-turn potentiometer.

The 60-cycle voltage difference, at a millivolt level,was amplified to the 120-V level by an amplifier thatpreserved the phase of the signal. The amplifier wasof conventional tube design and consisted of a numberof triode amplification stages and an output stage offour pentodes in parallel push-pull array. The outputof the amplifier saturated at just over 120 V rms, sothat full driving torque at 120 V could be supplied bythe motor for even small departures of the Fabry-Perotpressure from its null value without the servomotorbeing driven at arbitrarily high voltages for largedepartures of the pressure from null. The volumechange that the piston's full travel could produce waschosen to provide approximately 1 atm pressure changein the Fabry-Perot chamber. For a typical pressuredecrement equilibrium was reached in about 3 sec. Acertain amount of stray 60-cycle signal was inevitablypresent at the input of the amplifier. Much of it was900 out of phase with the signal and was ignored bythe servomotor. The remainder acted only as a minutebut constant pressure bias that was unimportant, sinceonly pressure changes are of concern.

We have found by scanning several orders of theFabry-Perot to observe the spectral profile of a multi-mode He-Ne laser (modes not resolved) that any non-linearity is not detectable. This probably means alinearity of 0.1% or better. Repeated scans of thesame spectral line showed that random errors, probablydue to the variability of the mercury-platinum contactor the sensitivity of the servosystem, do occur. Thesedo not exceed 1 of a step when the spectral freerange of the interferometer is 160 steps. Thus the un-certainty in the frequency corresponding to a particularpressure is a 6 l-o of a spectral free range.

>I. 0

I Density(aImagat)c 05 ,1 42.50

0.5

z~~- I , . ^ II-27 -18 -9 0 9 18 27

Frequency (GHz)

Fig. 4. An example of a spectral profile recorded with the digi-tally scanned interferometer. The peak count rate was about 6

counts per second.

October 1972 / Vol. 11, No. 10 / APPLIED OPTICS 2267

As an example Fig. 4 shows the depolarized Rayleighline of H2 at 2.5 amagat. The points have been cor-rected for the dark count of the photomultiplier and forparasitic light. The spectral free range of the inter-ferometer was 45 GHz, while the peak count recordedwas just less than 6 per second. The basic time r was14 sec, while N, and N2 were 5 and 9, respectively. Thesolid line is a theoretical profile and includes theinstrumental profile of full width at half-height of1.8 GHz. From our experience, the quality of theexperimental profile is comparable to the qualityof the profile of the polarized spectrum measuredat the same pressure and with the same total scan

time but recorded continuously with a time con-stant short enough to avoid distortion of the profile.(The polarized Rayleigh line in H2 is about 400 timesmore intense than the depolarized Rayleigh line.)Thus with this system one is able to obtain precisespectral line shapes even for very weak signals.

This work was supported in part by the NationalResearch Council of Canada.

References1. B. K. Gupta and A. D. May, Can. J. Phys., in press (1972).2. B. K. Gupta, Ph.D. Thesis, University of Toronto (1971).

OPTICAL SOCIETY OF AMERICAMEETINGS SCHEDULE

NATIONAL MEETINGS

17-20 October 1972 San Francisco Abstractdeadline 7 July 1972

Information: J. W. Quinn, Optical Society ofAmerica, 2100 Pennsylvania Avenue, N. W.,Washington, D. C. 20037or circle no. 64 on readerservice card

13-16 March 1973 Denver Abstract deadline 8December 1972

Information: J. W. Quinn, address above orcircle no. 58 on reader service card

9-12 October 1973 Rochester, N. Y. Abstractdeadline 13 July 1973

Information: J. W. Quinn, address above orcircle no. 57 on reader service card

TOPICAL MEETINGS

2-6 October 1972 THIRD INTERNATIONAL CON-FERENCE ON BEAM-FOIL SPECTROSCOPYTucson, Ariz. Abstract deadline 1 August 1972

Information: S. Bashkin, Department of Phys-ics, University of Arizona, Tucson, Ariz. 85721 orcircle no. 59 on reader service card

19-21 March 1973 OPTICAL STORAGE OF DIGITALDATA, Boulder, Colo. Abstract deadline 1 De-cember 1972

Information: J. W. Quinn, address above orcircle No. 65 on reader service card

OTHER MEETINGS OF INTEREST

9-13 October 1972 NINTH CONGRESS OF THEINTERNATIONAL COMMISSION FOR OPTICS,Santa Monica, Calif. Abstract deadline 1 Febru-ary 1972.

Information: D. Nicholson, ICO-9, P.O. Box95213, Los Angeles, Calif. 90045 or circle no. 63 noreader service card.

23-26 November 1972 FIRST INTERNATIONALCONFERENCE ON MODULATION SPECTROS-COPY, Tucson, Ariz. Abstract deadline 20 June

Information: B. 0. Seraphin, Optical SciencesCenter, University of Arizona, Tucson, Arizona85721

2268 APPLIED OPTICS / Vol. 11, No. 10 / October 1972