942 PROCEEDINGS OF THE IEEE June

to the Townsend discharge equation where a is the carrier multiplication factor. An in- teresting comparison can be made between the observed negative resistance region and those of the gas discharge tube and also of the double-injection diode.’* The observed symmetry of the characteristics rule out any explanation based on electron and hole injecting contacts of the type discussed by Lampert.6 The observed behavior appears to be a true bulk property and can be ex- plained by a carrier multiplication mecha- nism as in the case of the gas discharge tube. The ionization of traps and_ impurities fit in best with the observed edO dependence of current. The small increase of current in the negative resistance region, a property of both gas discharge and double injection, can be explained on the basis of a recombination process.

The author is indebted to L. C. Bobb for assistance in the preparation of film structures and to R. F. Schwarz for helpful discussions.

Xote Added in Proof: b’e have also ob- served a similar type of negative resistance in thin oxide films of T, and Ti.

K. L. CHOPRA Philco Scientific Lab.

Blue Bell, Pa.

a, ne? principle of operation of semiconductor de- 3 1’. I. Stafeev, “Modulation of diffusion length as

v~ces, Sooicl Phys.-Sdid Slate, vol. 1 . pp. 763-768; December, 1959.

doped germanium diodes, Phys. Rm.. vol. 96, pp. 4 W. W. Tyler, ”Injqtion breakdown in iron-

22G227. October 1954. 6 M . A . Lamp& “Double injection in insulators,”

Phys. Ra., voL,125. pp. 126-141; January, 1962.

sistance InSb diodes with large magnetic field effects,” 6 I. Melngall~s and R. H. Rediker. ‘Negative re-

J . A p p l . Phys.. vol. 33, pp. 1892-1893; May, 1%2.

Application of a Ruby Laser to High-speed Photography*

Multiple pulsing of a ruby laser has been achieved .and incorporated into a high- speed camera. The pulsing is accomplished by means of cavity Q spoiling techniques utilizing a Kerr cell. A rotating mirror camera is used with the laser actin$ as a stroboscopic light source to record the events. Framing rates of over a million frames per second with e x m u r e times of less than 30 nsec are easily obtainable. This in conjunction with the highly intense, monochromatic, coherent and collimated nature of the laser light makes the high- speed laser camera desirable in many areas of research.

The Q spoiling technique of obtaining very short duration pulses has been re- ported by McClung and Hellwarth.1 In their work a single ”giant pulse” was ob- tained and studied in detail as a means of achieving a better understanding of the

received April 2. 1963. This research was sponsored * Received March 1 1 , 1963; revised manuscript

by the Office of Naval Research. IF. J. McClung and R. W. Hellwarth. “Giant

optical pulsation from ruby.” J . A p p l . Phys.. vol. 33, pp. 828-829; March, 1962.

(b) Fig. l-Oscilloscope record of laser output: (a)

Operating in normal mode; upper trace is laser output and lower is pumping light. Time scale is 100 rn per division. (b) Multiple Laser pulses with Kerr cell operating at 500 kc per second.

Fig. 2-Resolution test photograph of microscope scale. Small divisions are 1 0 0 microns apart.

phenomena involved. In this letter the major item to be reported is that of achiev- ing multiple pulses and their utilization in high-speed photography. As a consequence several interesting aspects have occurred, such as a region of repetition rates in which a stability of the amplitude of the pulses occurs. The location of this region is a func- tion of the laser cavity length.

The laser used in these experiments was a 3-inch long &inch diameter ruby rod en- closed in a cylindrical cavity of elliptical cross section. The ruby rod was located a t one focus of the ellipse and the pumping light (an EGG FX-42) at the other. The walls of the cavity were highly polished, thereby furnishing a very efficient means of pumping energy input. A TIR 90”-oriented ruby rod with 0.04 per cent chromium ions was used. The Fabry-Perot inter- ferometric cavity was formed between the wedge end of the ruby and an external di- electric mirror which was located 13 inches away for the figures shown here. The re- flectivity of this mirror was 90 per cent a t

Fig. 3-Generation of a bubble in water by an electri- cal discharge (taken at 200,000 frames per sec- ond).

6943 A. The Kerr cell used as a Q spoiler was filled with nitrobenzene and was operated in the on-off-on mode with a dura- tion of 0.2 pec for the off position.

Fig. 1 shows the results of this method of Q spoiling with Fig. l (a) showing the laser operating in the normal mode with the Kerr cell inoperative. In Fig. l(b) the laser output is seen for the Kerr cell operating a t 500 kc per second. Due to poor frequency response of the photodiode used it is im- possible to compare the amplitudes of these pulses with those of the laser operating in the normal mode. For the Same reason the duration of these pulses is known to be shorter than that indicated in Fig. l (b) (ap- proximately 50 nsec). By changing the cavity length repetition rates of over 1 Mc have been achieved.

Fig. 2 illustrates the resolution possible. I t shows several frames taken a t a rate of 200,000 frames per second of a microscope scale on which the smallest division is 100 microns. The image speed over the film plane was approximately 100,000 inches per second. Fig. 3 shows a bubble being gen- erated in water by means of an electrical discharge. These three frames were taken a t a demagnification of 2 and a t a rate of 200,000 frames per second. I t is of particular interest to note the shock wave in the early frames; a t a later time reflected shocks were also observed. (Sote: a portion of each frame was lost due to the use of a wedge- shaped rotating mirror.) I t is also of interest to note that in order to achieve the proper exposure in these pictures a reduction of intensity by a factor of 800 was necessary.

Thus far the laser camera has yielded ex- cellent results in flow visualization both by direct observation and by means of scattered

elasticity and in high-speed Schlieren ob- light from flow tracers, in dynamic photo-

servations. In particular, flow visualization

1963

by means of forward scattered light from spheres 0.285 micron in diameter have been obtained. Several other areas of ap- plication are being investigated for future work. For example, the high intensity, rapid rise time and accuracy of frequency control for a series of pulses immediately suggest that such a device could be utilized in radar.

A . T. ELLIS M. E. FOURNEY

Calif. Inst. of Tech. Pasadena, Calif.

The Use of a Laser Amplifier in a Laser Communication System*

To make an evaluation of the place of a device in a communication system where the information is transmitted as a signal modu- lation, signal-to-noise ratio may be used as a figure of merit. (This is not true of a de- tection system where the information is essentially the presence or absence of the signal.) However, it is necessary to make precise the definition of signal and noise before such a ratio can be expressed. In this communication a receiver system will be con- sidered, consisting of a laser amplifier fol- lowed by a photomultiplier, and compared with the system without the amplifier. The photon model of Shimoda, et a/.' will be used for the evaluation. The noise will be considered to arise from three sources: background at reception, spontaneous emis- sion and photodetector shot noise.

First, it is necessary to define precisely what is wanted as signal and noise. Let s be the mean number of received signal photons, G be the amplifier gain and E be the quantum efficiency of the detector; then we will define the signal S to be

S = G .

The noise will be defined by the uncer- tainty in the output of the detector relative to the signal. Let P be the detector output; then the noise N will be defined by

where up is the standard deviatioi of the detector output.

Sow this output is calculated with the amplifier present. Let s be the number of photons in the signal and n the number of background photons; then the mean 1 and deviation uL of the output of the amplifier are given by

L = (s + ii)G + a p

uL = G2(0e2 - S + ~n' - i i )

+ (S + f i )G(l + 2P) + ad1 + P )

Project DEFENDEA. sponsored by the Advanced * Received April 8 , 1963. This research is a part of

Research Projects Agency. Department of Defense and technically monitored by the 05ce of the Scientific

673. Research. U. S. Air Force. Contract No. AF 49(638)-

1 K. Shimoda, H. Takahasi and C. H. T o m e s

cation to maser amplifiers." J . Phys. Soc. (Japan). 'Fluctuations in amplification of quanta with appd

PP. 686-700; June, 1957.

Correspondence

where p = K ( G - l )

K=np/ (n?-nl ) n? =upper-state population nl= lower-state population a =number of spontaneous emissions per

number of induced emissions per photon

G = amplifier gain u8 = standard deviation of signal u,, =standard deviation of noise.

Poisson statistics have not been assumed.

detector output are given by The mean P and deviation up of the

- P = (5 + RG)E + ape + 6

upf = (EG)2(Ud' - s + u,2 - R )

+ (S + W)EG(l + 2dP)

+ aV(1 + EP) + gdf

where d and Ud are the mean and deviation of the detector noise and e is the quantum efficiency. For simplicity the following is calculated:

RI = iV2/Sf

943

+ jK [s + R ( 1 + 26) ] - a(1 + 26) e 5 G

- {2G[S + d ( l +a)] + a p ( 1 + a ) ) e 2 > 0. There are two questions to be asked:

the amplifier will help?

G be?

1 ) IVhat are the limits of E for which

2) If the amplifier helps, how big must

To answer the first question, let G+ 50 ; then the inequality simplifies to

fk) = - E 2 + 2 2 s + R ( 1 + 22) K + K

- (2[s + R(l + a)] + a ( l + a ) K } e 2 > 0.

Possible e are then given by 9 _ < e < e a where eo is the positive root of f ( ~ ) = 0 (the other root is always negative):

where

Sf = P(&)'

LV* 0.2 - 9 + U"2 - ii (3 + E ) ( l + 2EP) -E.

Sf S2 + Sf&

If the laser amplifier is not used, the above formula applies with G- 1 and p=O. Let R2 be defined by the above ratio in the absence of the amplifier:

Rz = ~ . ~ - j + u , f - i i S + i i

Sf +x

If the amplifier is to be of any value, the following must hold:

RI < Rz. Dropping the first term, which is un-

affected, and multiplying by 12, we have

(3 + E ) ( 1 + 2EP) EG

a d 1 + ep) + ud2 + (%e + ape + a)e ( 4 2

+ <- + S + i i U 2 + ( & + d ) *

c €2

or

f f i ) ( l + 2eP) f CC€Pp(l + C P ) + Ud'

+ ( & + a p + d ) * < &(S + i i ) + G'[udf + (& + 6 ) f ]

The second question is answered by

+ a(1 + a)Ke? > 0.

H. STEISBERG l ech . Research Group Inc.

Syosset, X. Y .

The Interaction of Visible and Infrared Maser Transitions in the Helium-Neon System*

Quenching effects, due to competition among maser transitions which have com- mon upper or lower energy levels, have been observed in the He-Ne system. Some aspects of thiscompetition havebeen investigated us- ing a one meter long, 3-mm I.D. maser tube. Brewster windows and external spherical mirrors were used and the gas was excited by a dc discharge.

The relevant energy levels of helium and neon are shown in Fig. 1. The 2s2+2p1 (Paschen) transition* at 1.1523 p is normally the strongest of the 2s-2p group of lines, although maser action has been reported in

* Received April 12, 1963. 1 A. Javan. W. R. Bennett, Jr.. and D. R. Herriott.

tion in a gas discharge containing an He-h'e mixture, "Population inversion and continuous maser oscilla;

Phys. Rm. Lcfr.. vol. 6 , pp. 106-110: February, 1961.