March 15, 1992 / Vol. 17, No. 6 / OPTICS LETTERS 399 Sodium Raman laser: direct measurements of the narrow-band Raman gain Matthew Poelker and Prem Kumar Department of Electrical Engineering and Computer Science,Northwestern University, Evanston, Illinois 60208 Received September 23, 1991 Using a sodium Raman laser that is pumped by a single-frequency cw dye laser, we have measured the narrow- band Raman gain that exists in the vicinity of the sodium Di line. Large gains (>10 cm-') are measured at moderate pump intensities and sodium densities. The high gain allows us to detect directly a 1.772-GHz beat on the pump beam after it makes only a single pass through the sodium cell. By scanning the frequency of the sodium Raman laser while keeping the pump intensity and frequency constant, we are able to map out the Raman gain profile. The narrowness of the Raman gain is consistent with the narrow beat signal between the Raman-laser beam and the pump beam that was reported earlier [Opt. Lett. 16, 1853 (1991)]. Recently there has been increased interest in sodium-vapor oscillators because of their use as self- pumped phase conjugators and precision frequency translators.`14 Different investigators have ex- ploited different physical mechanisms that lead to gain in the vicinity of the sodium D lines to obtain phase conjugation. We are particularly interested in the Raman gain process that was exploited by Donoghue et al. 3 to achieve high-efficiency (5%) self- pumped phase conjugation with a low-optical-power density requirement (20 W/cm 2 ). Using the same gain mechanism, in a previous Letter 4 we showed how a precise frequency translation of 1.772 GHz can be achieved by using a sodium Raman laser op- erating with just 20 mW of pump power. These ex- periments 34 indicate that the Raman gain in the vicinity of the sodium D lines is extremely high. In this Letter we describe experiments designed to measure the single-pass Raman gain directly. In addition, the high gain associated with the Raman process makes it possible to observe a 1.772-GHz microwave beat on the pump beam after only a single pass through the sodium cell. With this observa- tion we have conducted extensive spectroscopic studies on the sodium D lines with a relatively simple experimental setup. Preliminary results are presented that manifest the Zeeman and the ac Stark effects. Tam 5 first observed that the neighboring longitu- dinal modes of a predominantly single-mode dye laser oscillating near the sodium D 2 line were ampli- fied on transmission through a heated sodium cell. In an unrelated experiment designed to study three- photon gain in a two-level medium (also sodium in their experiment), Gruneisen et al. 6 made passing reference to the amplification of a red-detuned probe beam when the pump beam was 1.7 GHz above the probe frequency. In this Letter we use a frequency-tunable sodium Raman laser to probe a sodium-vapor cell that is simultaneously pumped by a single-frequency dye laser. Amplification of the Raman-laser probe beam gives a direct measure- ment of the Raman gain. As first observed by Kumar and Shapiro, 7 when sodium vapor is placed within an optical cavity and pumped with monochromatic laser light tuned near the D 1 (D 2 ) line, frequency-shifted oscillation builds up within the cavity. The oscillation frequency is red (blue) shifted from the pump by 1.77 GHz when the pump is on the low- (high-) frequency side of the line. The gain for this Raman-laser action, it was determined, is a combined result of optical pumping and stimulated Raman scattering. When the pump-laser frequency is held constant, the Raman-laser action occurs only for discrete cavity lengths: those for which an integral number of Raman-shifted wavelengths fit into the ring cavity. Similarly, when the Raman cavity length is held constant, laser action occurs only for discrete values of the pump-laser frequency. These two facts indi- cate that the Raman gain is narrow band. A sodium Raman laser can be tuned in two ways. First, to vary the frequency over a broad range (=20 GHz), one merely varies the pump-laser fre- quency. The Raman-laser frequency follows the pump frequency, always =1.77 GHz below or above the pump frequency, depending on which side of the D line the latter is tuned. The exact frequency dif- ference between the pump and the Raman lasers is a function of many variables, including the pump in- tensity and frequency and the sodium density, and always remains within 1% of 1.772 GHz over a wide range of parameters. 4 ' 8 Second, the Raman-laser frequency can be varied over a much narrower range by changing the Raman-laser cavity length. Figure 1 shows a typical plot of the Raman-laser output as a function of the cavity length that is varied by applying a voltage ramp to a piezoelectric transducer on which one of the cavity mirrors is mounted. The separation between the periodic sets of peaks represents one free spectral range (FSR). The peaks within one set correspond to different transverse modes of the resonator. The maximum output occurs for the lowest-order TEMoo mode be- cause of better spatial overlap with the pump beam in sodium. The smooth, negatively sloped portion 0146-9592/92/060399-03$5.00/0 C 1992 Optical Society of America
Sodium Raman laser: direct measurementsof the narrow-band Raman gain
Matthew Poelker and Prem Kumar
Department of Electrical Engineering and Computer Science, Northwestern University, Evanston, Illinois 60208
Received September 23, 1991
Using a sodium Raman laser that is pumped by a single-frequency cw dye laser, we have measured the narrow-band Raman gain that exists in the vicinity of the sodium Di line. Large gains (>10 cm-') are measured atmoderate pump intensities and sodium densities. The high gain allows us to detect directly a 1.772-GHz beaton the pump beam after it makes only a single pass through the sodium cell. By scanning the frequency of thesodium Raman laser while keeping the pump intensity and frequency constant, we are able to map out theRaman gain profile. The narrowness of the Raman gain is consistent with the narrow beat signal betweenthe Raman-laser beam and the pump beam that was reported earlier [Opt. Lett. 16, 1853 (1991)].
Recently there has been increased interest insodium-vapor oscillators because of their use as self-pumped phase conjugators and precision frequencytranslators.`14 Different investigators have ex-ploited different physical mechanisms that lead togain in the vicinity of the sodium D lines to obtainphase conjugation. We are particularly interestedin the Raman gain process that was exploited byDonoghue et al.3 to achieve high-efficiency (5%) self-pumped phase conjugation with a low-optical-powerdensity requirement (20 W/cm2). Using the samegain mechanism, in a previous Letter 4 we showedhow a precise frequency translation of 1.772 GHzcan be achieved by using a sodium Raman laser op-erating with just 20 mW of pump power. These ex-periments3 4 indicate that the Raman gain in thevicinity of the sodium D lines is extremely high. Inthis Letter we describe experiments designed tomeasure the single-pass Raman gain directly. Inaddition, the high gain associated with the Ramanprocess makes it possible to observe a 1.772-GHzmicrowave beat on the pump beam after only a singlepass through the sodium cell. With this observa-tion we have conducted extensive spectroscopicstudies on the sodium D lines with a relativelysimple experimental setup. Preliminary results arepresented that manifest the Zeeman and the acStark effects.
Tam5 first observed that the neighboring longitu-dinal modes of a predominantly single-mode dyelaser oscillating near the sodium D2 line were ampli-fied on transmission through a heated sodium cell.In an unrelated experiment designed to study three-photon gain in a two-level medium (also sodium intheir experiment), Gruneisen et al.6 made passingreference to the amplification of a red-detunedprobe beam when the pump beam was 1.7 GHzabove the probe frequency. In this Letter we use afrequency-tunable sodium Raman laser to probe asodium-vapor cell that is simultaneously pumped bya single-frequency dye laser. Amplification of theRaman-laser probe beam gives a direct measure-ment of the Raman gain.
As first observed by Kumar and Shapiro,7 whensodium vapor is placed within an optical cavity andpumped with monochromatic laser light tuned nearthe D1 (D2) line, frequency-shifted oscillation buildsup within the cavity. The oscillation frequency isred (blue) shifted from the pump by 1.77 GHz whenthe pump is on the low- (high-) frequency side of theline. The gain for this Raman-laser action, itwas determined, is a combined result of opticalpumping and stimulated Raman scattering. Whenthe pump-laser frequency is held constant, theRaman-laser action occurs only for discrete cavitylengths: those for which an integral number ofRaman-shifted wavelengths fit into the ring cavity.Similarly, when the Raman cavity length is heldconstant, laser action occurs only for discrete valuesof the pump-laser frequency. These two facts indi-cate that the Raman gain is narrow band.
A sodium Raman laser can be tuned in two ways.First, to vary the frequency over a broad range(=20 GHz), one merely varies the pump-laser fre-quency. The Raman-laser frequency follows thepump frequency, always =1.77 GHz below or abovethe pump frequency, depending on which side of theD line the latter is tuned. The exact frequency dif-ference between the pump and the Raman lasers is afunction of many variables, including the pump in-tensity and frequency and the sodium density, andalways remains within 1% of 1.772 GHz over a widerange of parameters.4'8 Second, the Raman-laserfrequency can be varied over a much narrower rangeby changing the Raman-laser cavity length.Figure 1 shows a typical plot of the Raman-laseroutput as a function of the cavity length that isvaried by applying a voltage ramp to a piezoelectrictransducer on which one of the cavity mirrors ismounted. The separation between the periodic setsof peaks represents one free spectral range (FSR).The peaks within one set correspond to differenttransverse modes of the resonator. The maximumoutput occurs for the lowest-order TEMoo mode be-cause of better spatial overlap with the pump beamin sodium. The smooth, negatively sloped portion
0146-9592/92/060399-03$5.00/0 C 1992 Optical Society of America
Fig. 1. Relative output of the Raman laser as the cavitylength is varied.
PBS Heat-Pipe POven
Fig. 2. Schematic of the experimental setup.
of the large TEMoo peak represents the Raman-laserfrequency tuning range.
The Raman laser used in the experiments re-ported here is composed of a triangular 1.75-m ringcavity similar in design to that used in Ref. 4. Anunfocused beam from a single-frequency dye laser isused to pump the Raman laser. A spatial filter isinserted into one arm of the cavity to eliminate las-ing in higher-order transverse modes, whose pres-ence made cavity-length locking a difficult task.The sodium cell is a heat-pipe oven with an =12-cmactive column. The external oven temperature istypically 280'C, and 5 Torr of helium is added to thecell as a buffer gas.
To measure the Raman gain, a portion of the dyelaser (the pump beam) is combined with thefrequency-shifted Raman-laser output (the probebeam) by using a polarizing beam splitter (PBS) asshown in Fig. 2. The two orthogonally polarized,copropagating, collinear beams are then focusedinto the sodium cell with a 25-cm focal-length lens.Gain occurs in a polarization direction orthogonalto that of the pump because of quantum selectionrules pertaining to the Raman process.? An identi-cal second lens recollimates the two beams at theoutput of the sodium cell. When the pump beam isblocked, the probe beam experiences absorption.However, when the pump beam is on, the probebeam is amplified. A polarizer (P) at the cell outputseparates the two beams, and the gain is deter-mined by dividing the output probe-beam powerwith that at the input. Any pump power polarizedparallel to the probe-beam polarization at the celloutput, as a result of intensity-induced birefrin-gence in sodium,2 is measured by blocking the in-put probe beam and subtracted from the outputprobe power.
To investigate the Raman gain across the rela-tively broad D1 line, the probe frequency is varied bytuning the dye-laser frequency. As mentionedabove, the Raman-laser frequency leads or followsthe pump frequency by 1.77 GHz. A servo mecha-nism is employed to maintain the Raman laser at
peak output power in the TEMoo mode. The sodiumcell used for the gain measurements is actuallythe same cell that is used in the Raman laser-theRaman-laser region of the cell does not overlap theregion used to investigate the gain; the two are sep-arated by approximately 1.5 cm. Obviously, theheat-pipe conditions for gain measurements are thesame as those described above for the Raman laser.
Before being focused into the cell, the beams are1.5 mm in diameter. Exact beam diameters
inside sodium are not known because of the self-focusing effects; the diffraction formulas predict afocused spot diameter of 63 ,um. The focused pump-beam intensity is 3.2 kW/cm2 . The focused probe-beam intensities varied from 1.6 to 32 W/cm2 owingto variation in the Raman-laser output power as thedye-laser frequency is tuned across the D1 line. Themeasured Raman gain is plotted in Fig. 3 as a func-tion of the dye-laser frequency. At the frequency ofmaximum gain, approximately 0.2 mW of the probebeam enters the cell and nearly 29 mW exits, corre-sponding to a gain of 12 cm-'. As expected, thepeak gain occurs at a frequency that corresponds tothe peak Raman-laser output. On the blue side ofthe line, some gain values are negative because ofself-focusing-induced pump-beam breakup. At
Fig. 3. Peak Raman gain (squares) in the vicinity of thesodium D1 line. The upper solid trace, which is obtainedby using the method of saturation spectroscopy,'0 cali-brates the frequency scale with the hyperfine doublet(1.77-GHz separation) of the D1 line. Raman gain wasmeasured between the markers on the two sides of thedoublet. The single-pass gain of 180 corresponds to again coefficient of =15 cm-'.
Fig. 4. Beat signals on the transmitted pump beam.(a) Unfocused pump intensity 11 W/cm2, (b) central peak of(a) with magnetic field on, (c) focused pump intensity3.2 kW/cm2, (d) focused pump intensity 6.4 kW/cm2.
-20 -10 0 10 20Raman-Laser Frequency Detuning (MHz)
Fig. 5. Measurement of the Raman gain profile on thelow-frequency side of the D, line. The data (squares)could only be taken on one side of the Raman gain profile.The solid curve is mirrored on the right side to estimatethe gain bandwidth of 21 MHz.
these frequencies, the pump and the probe beamsare no longer collinear, and the latter is absorbed.
The high Raman gain allows us to detect directlya 1.772-GHz beat on the pump beam after only asingle pass through the sodium cell." The datashown in Fig. 4 are obtained by blocking the inputprobe beam, rotating the polarizer at the cell outputby 450, detecting the transmitted pump beam witha fast photodiode, and observing the amplifiedphotocurrent with a microwave spectrum analyzer.No beats are observed when the sodium cell is re-moved or when the pump beam is detuned far fromresonance. The beat signal originates from hetero-dyning of the Raman-shifted amplified noise withthe transmitted pump beam. Observation of thisbeat signal is thus a direct manifestation of theRaman gain, and its shape reflects the Raman gainprofile.
The trace in Fig. 4(a) is obtained with 11 W/cm2 ofunfocused pump intensity tuned 7 GHz below theF = 2 transition of the D, line. The main peak iscentered at 1.7716 GHz and is =30 kHz wide. Thenarrowness of this Raman gain profile is consistentwith the narrow beat signal between the Raman-laser beam and the pump beam that was reportedearlier.4 A stray magnetic field in the laboratoryremoves the degeneracy of the S112 sublevels, whichresults in the side peaks that are present inFig. 4(a). These side peaks decrease in amplitudeand move farther away from the central peak as asmall magnetic field (=3 G) is applied in a directionparallel to the pump-beam propagation direc-tion.8 In Fig. 4(b) we only show the central peakwith the magnetic field on for comparison withFig. 4(a). This behavior is a manifestation of theZeeman effect. The trace in Fig. 4(c) is obtainedwith a focused pump intensity of 3.2 kW/cm2 tuned2 GHz below the F = 2 transition of the D, line andwith no applied magnetic field. It shows a singlepeak at a slightly shifted frequency whose width isconsiderably wider than that shown in Fig. 4(b),which is obtained at a much lower pump intensity.The frequency shift and the increased width are amanifestation of the ac Stark effect. The beat am-plitude increases and splits as the pump intensity isfurther increased to 6.4 kW/cm2 [Fig. 4(d)]. Thesplitting follows a square-root dependence on thepump intensity.8
We also measured the Raman gain linewidth bykeeping the pump frequency fixed while varying theprobe frequency. This is accomplished by changingthe Raman-laser cavity length, in a controlled man-ner, using a side-lock servo mechanism that can lockthe Raman-laser cavity length anywhere on thesmooth negatively sloped portion of the outputprofile shown in Fig. 1. The Raman gain for a3.2-kW/cm2 pump beam tuned 2.5 GHz below theF = 2 transition is plotted in Fig. 5 as a function ofthe Raman-laser frequency. Zero corresponds to afrequency difference between the Raman and pumplasers of 1.772 GHz. The probe detuning is deter-mined by monitoring the Raman-laser output; achange in the output power as the cavity length isvaried can be converted into frequency units by us-ing Fig. 1 and the known FSR of the cavity. Asshown in Fig. 5, the Raman gain linedwidth is=21 MHz, consistent with the beats in Figs. 4(c)and 4(d). Similar results are obtained with a pumpbeam that is tuned above the F = 1 transition.When the data in Fig. 5 were taken, the sodium den-sity was lowered to reduce the strength of the high-order Raman-laser transverse modes that madeactive cavity-length control difficult. The lowersodium density accounts for the lower gain values inFig. 5 compared with those in Fig. 3. These high-order modes also made it impossible to tune theRaman-laser frequency on the right side of theRaman gain profile as shown in Fig. 5.
In conclusion, we have reported direct measure-ments of the narrow-band Raman gain that exists inthe vicinity of the sodium D1 line. The high gainallows us to detect directly a 1.772-GHz beat signalon the pump beam after only a single pass throughthe sodium cell.
References
1. C. J. Gaeta and D. M. Pepper, Opt. Lett. 16, 802(1991); C. J. Gaeta, J. E Lam, and R. C. Lind, Opt.Lett. 14, 245 (1989).
2. M. Vallet, M. Pinard, and G. Grynberg, Opt. Lett. 16,1071 (1991).
3. J. Donoghue, M. Cronin-Golomb, J. S. Kane, and P. R.Hemmer, Opt. Lett. 16, 1313 (1991).
4. M. Poelker, P. Kumar, and S.-T. Ho, Opt. Lett. 16,1853 (1991).
5. A. C. Tam, Phys. Rev. A 19, 1971 (1979).6. M. T. Gruneisen, K. R. MacDonald, and R. W Boyd,
J. Opt. Soc. Am. B 5, 123 (1988).7. P. Kumar and J. H. Shapiro, Opt. Lett. 10, 226 (1985).8. M. Poelker, Ph.D. dissertation (Northwestern Univer-
sity, Evanston, Ill., 1991).9. P. R. Hemmer, Ph.D. dissertation (Massachusetts In-
stitute of Technology, Cambridge, Mass., 1984).10. T. W Hinsch, I. S. Shanin, and A. L. Schawlow, Phys.
Rev. Lett. 27, 707 (1971).11. Coherent Raman beats have previously been observed
in molecular systems in which the ground-level degen-eracy is removed by Stark-pulse switching. See R. L.Shoemaker and R. G. Brewer, Phys. Rev. Lett. 28,1430 (1972); R. G. Brewer and E. L. Hahn, Phys. Rev.A 8, 464 (1973).
