Ultrastable CO2 Lasers c. Freed

• This article begins with a briefreview ofgas-laser research and the events that

led to the development ofultrastable CO2 lasers at Lincoln Laboratory. 13 Extremely high spectral purity and short-term stabilities of!1j/j< 1.5 X 10-

have been routinely achieved with these ultrastable CO2 lasers. At MIT/Lincoln Laboratory, we have also invented a novel frequency stabilization method that

enabled the long-term frequency locking ofCO2 (and many other) molecular lasers to the center frequency ofany regular- or hot-band COrisotope-laser

transition. Consequently, line-eenter-stabilized CO2-isotope lasers have become the best secondary frequency standards known to date in the infrared domain. Some ofthe output characteristics, selected applications, and design features of the lasers are also described. The article concludes with a discussion ofsome of the advantages and limitations ofthe present laser design.

arbon-dioxide lasers were invented by mirrors within the vacuum envelope that contained the . GK.N. Patel in 1964 [1], and by mid-I965 he laser gas mixture). Chad demonstrated CW output powers that ex­ The first stable gas laser that we designed and con­ ceeded 10 W [2]. It was the first time that any single structed at Lincoln Laboratory departed from Javan's laser had achieved a CW or average-power level exceed­ design only in minor details; for example, the laser was

ing 1 W Thus the CO2 laser became the first, and 25 madewith demountable metallic (conflat) vacuum seals years later it remains the best, laser for long-range Dop­ insteadofbrazed joints. The features that contributed to pler and imaging optical radars [3-5]. The recent suc­ laser stability, i.e., the rigid invar rod cavity and the

cesses ofthe Firefly tests carried out with the CO2-laser internal mirrors, were essentially retained intact. Two radar facility at the Lincoln Laboratory Firepond site in such He-Ne lasers were built and extensively utilized to Westford, Mass., in March [6] and October 1990 have studythe intensity fluctuations oflaser output just below amply supported the above statement. and above the oscillation threshold. '

The development ofultrastable CO2 lasers was a di­ It should be noted that during the early period of rect outgrowth of previous work and experience with laser history (1960 to 1963) there was a great deal of stable He-Ne lasers at MIT/Lincoln Laboratory. Re­ interest and outright controversy about the nature and search on gas lasers started at Lincoln Laboratory around appropriate analytical description ofthe light emitted by the beginning of1963. At the time, most gas lasers were lasers [10]. GH. Townes suggested to H.A. Haus and based on the He-Ne system that A. Javan had invented me that this topic would be very suitable for us to in 1960 [7]. By early 1963, GH. Townes, Javan, and investigate. Our subsequent research and publications coworkers had clearly ~stablished the inherently good [11-16] on the photocurrent spectrum and photoelec­ spectral purityand short-term frequency stability ofHe­ tron counts produced by a He-Ne gas laser have clearly Ne lasers [8] in a series of experiments to test special shown that the spontaneous emission in the vicinity of - relativity [9]. These experiments, the famous "ether drift" lasing threshold can dominate the amplitude noise ofa measurements carried out in the wine cellar of the laser. However, it did require a great deal ofleal:ning and Round Hill Estate of MIT [8], utilized pairs of very the development ofrefined experimental techniques to stable He-Ne lasers, the design ofwhich was developed operate the lasers with sufficient stability in the vicinity by Javan and his colleagues. Javan's design called for a ofthe oscillating threshold. Consequently; satisfYing the

very rigid optical cavity that consisted of four l-inch­ requirement for stable CO2 lasers at Lincoln Laboratory diameter invar rodspacers and two internal mirrors (i.e., three years later (1966) was straightfOlward: the conver-

VOLUME 3, NUMBER 3, 1990 THE LINCOLN LABORHOHY JOURNAL 479 -FREED

Ultrastable CO2 Lasers sion from He-Ne to CO2 operation was simply a matter At this time I would like to point out that this article of changing two mirrors and an output window, and is only a brief review of more than 25 years of work. providing a water-cooled discharge tube and a CO2 Also, the results reponed here represent the cumulative laser gas mixture. Even our initial measurements with efforts of many individuals in the MIT/Lincoln labo­ the first pair of these lasers convened from He-Ne to ratory community. The acknowledgments at the end list

CO2 operation demonstrated stabilities at least 100 those who most significantly contributed to this times better than previously reponed [17], and the endeavor. demand at Lincoln Laboratory for additional CO2 lasers skyrocketed vinually overnight. Spectral Purity and Short-Term Stability , Thus we undenook a complete redesign ofthe stable The theoreticallinewidth ofa laser above threshold was laser structure at that time, and within about a year (by first derived by Schawlow and Townes and published in early 1968) we had the first batch of the redesigned 1958 [23J, about two years before the actual demonstra­ ultrastable CO2 lasers in operation [18J. (Today, more tion of the first laser, which was made from ruby, by than 20 years later, dozens ofthese ultrastable lasers are T.H. Maiman in 1960 [24]. In its simplest form the still in operation at Lincoln Laboratory.) In 1970 we Schawlow-Townes theory predicted that the output also demonstrated the first completely sealed-off and power Po ofa laser will have a Lorentzian line shape with ultrastable operation of the molecular CO laser system a quantum-noise-limited full width at half maximum by using these redesigned CO2 lasers [19, 20J. (FWHM) !::.f approximated by Within Lincoln Laboratory the ultrastable CO2/CO lasers were, and still are, primarily used in optical radars, !::.f '" anhfo [ fo J2, heterodyne spectroscopy, and optical pumping, and for Po Q.; the study ofvarious processes in laser physics and kinet­ where a, h,lo, Po' and Q.; denote the population-inver­ ics by MIT/Lincoln Laboratory staff and graduate sion parameter, Planck's constant, the center frequency, students. the output power, and the cold-cavity Q of the laser Many additional ultrastable lasers were built else­ (i.e., the Q of the cavity without any active medium), where from the detailed design drawings that Lincoln respectively. Laboratory furnished to qualified researchers at various In a well-designed CO2 laser, Q.; is given by universities and laboratories in the United States and Great Britain. A complete listing of the various users Q.; '" 2 nLfo , (1) and applications is obviously well beyond the scope of c~ this paper; I will briefly mention the two that I person­ where L, c, and ~ denote the cavity length, speed of ally found most gratifYing and rewarding. The first is the light in vacuum, and mirror transmission, respectively. use of ultrastable CO2 lasers for infrared heterodyne (Diffraction losses are not included in Equation 1 be­ spectroscopy in planetary (Mars andVenus) atmospheres cause they are usually negligible compared to the output by C.H. Townes and his collaborators at the University coupling loss). For a small CO2 laser with L =50 cm and 7 ofCalifornia, Berkeley [21, 22J. The second is the high­ T; = 5%, Q.; is on the order of 10 . Thus for a typical altitude balloon-borne experiment that used a sealed-off power output ofl to !OW(which a smallTEMooq-mode

CO laser pump with which C.KN. Patel first demon­ CO2 laser can easily supply) the quantum-phase-noise­ strated the depletion of high-altitude owne by the limited linewidth is less than 10.6 Hz. Note that 10.6 Hz complex interaction ofsunlight and cWorofluorocarbon represents less than 1 pan in 10 19 of the output fre­ 13 compounds. quency lfo - 3 X 10 Hz) ofa CO2 laser. This inherent The main body ofthis article has four sections: spectral purity ofCO2 lasers can be explained as follows: 1. spectral purity and shon-term stability, the linewidth !::.fis inversely proportional to the prod­ 2 2. long-term, line-center stabilization ofCO2 lasers, uct of Po and Q.; , and the combination ofhigh Po and 3. COTisotope lasers as secondary frequency stan­ high ~ can be simultaneously achieved with rela­ dards, and tive ease even in a small COTlaser oscillator. Oscil­ 4. laser design. lators in the radio-frequency (RF) and microwave do-

480 THE LINCOLN LABORATORY JOURNAl VOLUME 3, NUMBER 3, 1990 o FREED

Ultrastable CO2 Lasers main have either high Po or high Qc but not both ment the output beams ofthe two lasers are combined together in a single device. by a beam splitter. The colinearly propagating wavefront Laser stabilities are most frequently determined in of the two lasers is intercepted by a photon detector the laboratory from the results of heterodyne experi­ whose output current i(t) will contain (in addition to dc ments with two lasers. In a typical heterodyne experi- currents) a component i(t)IF at the difference frequency

400 • • 0

• 0

o 0 ~ 300 ~-+--+--+---+ l---t---t---+-~ > o • l 2- --4- ~ Q) o ,---t--"'"f"i--" Cl ~ ." ~ 200

~ ...­o u ...­Q) Q) o 100 •

o (a) --..1 1.- 0.1 MHz

-60

~ E a:l ~ Q) -80 Cfl c o a. Cfl Q) c:: o -100 ...­ u ...-Q) oQ)

-120

(b) --..1 1.- 10 MHz FIGURE 1. Spectrum-analyzer display of beat note between a 240-/1W single-frequency Pbo.saSno.12Te diode laser (well above threshold) and the P(14) transition of the CO2 gas laser. The intermediate-frequency (IF) bandwidth is 10 kHz and the sweep rates and exposure times are (a) 0.2 sec/division and 2 sec, and (b) 0.002 sec/division and 0.5 sec, respectively. The dotted curves correspond to 54-kHz-bandwidth Lorentzian fits to the power spectral density. Note that the photo in b contains -25 consecutively recorded superimposed traces.

VOLUME 3. NUMBER 3, 1990 THE LINCOLN LABORATORV JOURNAL 481 -FREED

Ultrastable CO2 Lasers

(also called the beat frequency, or intermediate frequen­ cy) ofthe two lasers:

... ltl In order to differentiate between the two individual la­ Q) c: ser outputs, the subscripts 10 and s are used to denote ....J the local-oscillator and signal-oscillator parameters, respectively. Laser stabilities can be determined by either frequen­ cy-domain (Fourier spectrum) or time-domain (Allan variance) analysis of the beat-note spectra of the laser pairs. Ifthe spectral purity ofone ofthe lasers (typically ...... 1 1..- the local oscillator) is much better than that ofthe other 500 Hz laser, then the beat-note spectrum may be entirely at­ Beat Frequency tributed to the worse of the two lasers. For example, even for a high-quality lead-salt solitary diode laser the o power output is only on the order ofa milliwatt and ~ is less than 10. Thus the quantum-phase-noise-limited Schawlow-Townes linewidth prediction is ?!ically in the tens ofkilohertz, in contrast with the 10' Hz pre­ lil dicted for a COzlaser. We took advantage ofthese char­ ~ -20 0) acteristics of the COz and lead-salt diode lasers in the 0 ....J historic first experimental demonstration and verifica- -30

Q) -40 L.- ----JL,...; E ,.,i= ...... 1 1..- :!:: c (/) 0 500 Hz c: .­ Q)- o ~ Beat Frequency - Q) ltl (/) ~.D FIGURE 3. Same as Figure 2, but with increased acousti­ ~o cal background noise coupled to the lasers. Note that C.u en Q) the spectral lines have a Gaussian-line-shape envelope, '{l ~ e tion ofthe Lorentzian line shape and FWHM linewidth o 2 3 4 5 ofSchawlow and Townes [25] (Figure 1)...... 1 1..- To establish the spectral purity ofthe COzlaser itself, 500 Hz we can heterodyne two COzlasers ofequal high quality so that the resulting beat-note spectrum can be appor­ Beat Frequency (kHz) tioned equally to each laser. Two problems arise, howev­ FIGURE 2. Real-time power spectrum of the beat signal er, in trying to measure the Schawlow-Townes linewidth between two free-running CO2 lasers. The horizontal scale ofhigh-quality COzlasers. The first ofthese problems is of the figure is 500 Hz/division, which indicates that the instrumental: the stability of the available instrumenta­ optical frequencies of the two lasers producing the beat note were offset by less than 3 kHz. The 10-Hz width of tion is generally insufficient for reliable measure­ 6 the line shown is limited by instrumentation; the linewidth ments of spectral purities of 10. Hz or better. of the beat note falls within this limit. The origin of the second problem is that for well-

482 THE LINCOLN LABORATORY JOURNAL VOLUME 3, NUMBER 3, 1990 • FREED

Ultrastable CO2 Lasers

-20

-45 ~ 60 dB CfJ C Q) 0 ~ -70 .- u Q) a. -3 (f) ~ ~ 9 x10 Hz

~I lM~1 -95 A V11V VV ~) ~W !~Nf n II 1\1 ~nAA W M\~~ ,AI AlJVi III -120 t 0.9 Hz 1.0 Hz 1.1 Hz Frequency

FIGURE 4. Spectral purity of beat note between two CO2 lasers that were phase locked to each other with a frequency offset of 10 MHz. The FWHM spectral width of the beat note was about 9 x 10,6 Hz during the 26.67-min measurement time. designed CO2 lasers the so-called technical noise sources of the spectral width was limited to a 10-Hz resolution dominate over the quantum-phase-noise-limited by the O.l-sec observation time that was set by the Schawlow-Townes linewidth [26]. Examples of techni­ instrumentation, not by the laser stability itself cal noise sources are acoustic and seismic vibrations, and Figure 3 shows another beat note of the same two power-supply ripple and noise. These sources can cause lasers as in Figure 2, bur with increased acoustical back­ frequency instabilities by perturbing the effective cavity ground noise coupled to the lasers [27]. The increased resonance via the sum offractional changes in the refrac­ level ofthe spurious modulation sidebands that resulted tive index n and the optical-cavity length L: from the noise is self-evident in the figure. The loga­ rithmic display of Figure 3 also shows that the spectral ~I = I( :n + ~L )- lines fir within the Gaussian-line-shape envelope given by the expression As an example, a change ofonly 10'3 A(about 1/1000 _l_ _ of the diameter of a hydrogen atom) in a 50-cm-Iong exp [(I - {O)2] a.fi"; 2a CO2-laser cavity will cause a frequency shift ofapproxi­ mately 6 Hz. A 6-Hz variation in the approximately with a = 209 Hz fitted to the beat-note spectrwn. Also

3 X lO13-Hz frequency ofa CO2 laser corresponds to a note that the levels ofthe spurious modulation spectral 13 fractional instability of 2 X 10- . Figure 2, which lines fall at least 35 dB below the beat note's carrier Jhows the real-time power spectrwn of the beat sig­ frequency for frequencies that are spaced more than 1.2 nal between twO free-running lasers that we designed kHz from the carrier frequency. and built at Lincoln Laboratory [18], implies a fre­ The specrra shown in Figures 2 and 3 clearly imply X quency stability at least as good as 2 10,13. The that by frequency- and/or phase-locking two stable CO2 discrete modulation sidebands in the figure were lasers with feedback-loop bandwidths of only a few primarily due to ac-power-line frequency harmonics, kilohertz, we can overcome the masking effects of the cooling-fan noise, and slow frequency drift; however, each spurious modulation sidebands. spectral line was generally within the 10-Hz resolution Figure 4 shows the real-time power spectrwn of the bandwidth ofthe spectrum analyzer. The measurement beat note of two ultrastable CO2 lasers that were phase

'IOLUME 3, NUMBER 3. 1990 THE LINCOLN LABORATORY JOURNAL 483 -FREED

Ultrastable CO2 Lasers

--+-116 msecl'­ 4 msec --+-l .-

u~ Power 1-kW l~ Oscillator 1 Amplifier Du,I,,,, 1 _ Q) Cl ~ f + f 48-i n Telescope III 1 ~ ~ P Doppler o I > 1 e 1 Aberration c ­o Corrector 1 U 1 . f _ 2R I Doppler - A p 1 1

--+-1 f.- 500 Hz

FIGURE 5. Simplified block diagram of a narrowband configuration of the CO2 laser radar at the Lincoln Laboratory Firepond Facility in Westford, Mass. In the figure, wavy and solid lines denote optical and electrical signal paths, respectively. locked with a fixed la-MHz frequency offset between by the round-trip time to and from the target. Hence, the two lasers and with the unity-gain bandwidth ofthe effects due to disturbances with correlation times less servoamplifler set to about 1.2 kHz [28]. Note that the than the round-trip time of the transmitted signal will 2 horiwntal scale in the figure is only 2 X 10- Hz/division be included in the measured beat-note spectrum. and the vertical scale is logarithmic, with 12.5 dB/ Figure 5 shows a greatly simplified block diagram of division. Using the results from Figure 4 and the equa­ a narrowband configuration of the COrlaser radar tion for a Lorentzian line shape, we calculate the FWHM at the Lincoln Laboratory Firepond Facility. The 6 spectral width ofthe beat note to be about 9 X 10- Hz. 0.5-m-Iong local oscillator and the 1.5-m-Iong power It took 26.67 min of measurement time to obtain oscillator were designed and constructed at just a single scan with the frequency resolution ofFigure Lincoln Laboratory. The higher power oscillator was 4. Because tracking even by a very good servosystem phase locked to the local oscillator with a fixed la-MHz would still be limited by quantum phase noise, the frequency offset between the two lasers. Figure 4 has narrow linewidth in Figure 4 is an indirect but clear shown the beat-note spectrum ofthese two lasers. confirmation ofthe high spectral purity ofCO2 lasers, as The frequency stability ofthe COrlaser-radar facili­ predicted by the Schawlow-Townes formula. ty at Firepond was verified from observations on GEOS­ Laser stabilities are most conveniently evaluated from III, a NASA geodetic satellite that is equipped with an the beat-note spectra oftwo lasers, as shown in Figures IRTRAN II solid-cube-corner retroreflector. Doppler 2, 3, and 4. However, the results are not altogether measurements of the radar returns from GEOS-III foolproof because the disturbances that cause frequency helped to determine the radial velocity of the satel­ jitter can be at least partially correlated. In an optical lite, and to set an upper bound to the laser-oscillator radar we can compare a laser with its own output delayed instability [29]. During the measurements the satellite

484 THE LINCOLN LABORATORY JOURNAL VOLUME 3, NUMBER 3, 1990 • FREED

Ultrastable CO2 Lasers range was 1063 lan, which corresponded to a round­ !:!. flo dB = 399 Hz L\ f 30 dB = 1905 Hz trip signal travel time of -7 msec. In this particular a,o dB = 18.5 Hz _ a30 dB = 314 Hz radar experiment a duplexer chopped the constant­ T = 7.09 msec If R = 1063 km frequency amplified COTlaser-output signal so that .... -a=117Hz _t 10 dB a 25%-duty-cycle pulse train consisting of 4-msec­ -t duration pulses spaced 16 msec apart was transmitted 6.088 to the orbiting satellite. Figure 6 shows the spectra that resulted from an RF 6.072 test signal of 4-msec duration processed in the same manner as the optical radar data [28, 29]. In the figure, 6.056 the horiwntal rows simulate logarithmic displays ofthe U 6.040 Q) power spectra ofa consecutive sequence ofradar return U1 Q) 6.024 signals. The [(sin x)/x]2 spectral envelope, which is due E to the 4-msec pulse duration, is quite evident. The f= 6.008 average spectral width 10 dB below the peak is 383 Hz, 5.992 I and the standard deviation ofthe spectral width is 5 Hz. Figure 7 shows the actual COTradar return signal 5.976 from GEOS-III at a 1063-km range [28, 29]. Notice 5.960 the striking similarity to the simulated signal return of Figure 6, which was obtained from a high-quality fre­ 5.944 quency synthesizer that was also part ofthe radar receiv­ er. A comparison of Figures 6 and 7 shows that the 5.928 -10-dB average spectral width of the radar return 5.911 signal broadened from 383 to 399 Hz and the stan­ -20 -18 -1 6 -14 -12 -1 0 -8 -6 dard deviation of this spectral width increased from Frequency (kHz) 5 to 18.5 Hz. According to these figures, the short-term FIGURE 7. A consecutive sequence of actual radar re­ stability of the entire CO2 radar system, including turn signals from GEOS-1I1.

round-trip propagation effects caused by fluctuations in the atmosphere, was better than 1.5 x 10- 13 for the !:!. f 10 dB = 383 Hz !:!.f30dB = 1020 Hz 7-msec round-trip duration. 0;0 dB = 5 Hz a30 dB = 172 Hz _t 10 dB Figure 7 also shows that the standard deviation of -t pulse-to-pulse centroid jitter ofthe spectra was 117 Hz 6.343 I.JUViWVUVVW during the test run. No effort was made to line-center­ U 6.327 11\MJf\,Jvl\rJYMIHm stabilize either ofthe lasers because the long-term stabil­ Q) .!!:­ ity ofthe free-running laser was more than adequate for Q) 6.311 E typical optical radar applications. f= 6.295 Long-Term Line-Center Stabilization 6.279 l'\MII/UVlJvJV\,H WVW' ofCO2 Lasers

6.263 INVV'LNVVVVV\IVW'VlI Lasers can possess exceptionally high spectral purity and short-term frequency stability. Long-term stability, 6.247 1'V\lu~WWWvJ'mvvlIl however, is generally lacking because all lasers are more -6 -4 -2 0 2 4 6 or less tunable over a frequency band that is determined Frequency (kHz) by the detailed physics ofthe gain-profile characteristics FIGURE 6. Simulated radar returns for a 4-msec-duration ofeach particular laser system. RF test signal. In a typical low-pressure (-15 Torr) CO2 laser, the

VOLUME 3. NUMBER 3, 1990 THE LINCOLN LABORATORY JOURNAL 485 • FREED

Ultrastable CO2 Lasers

4 Hot Band 1 CO 01 1 2 '7 E Detector u 00°1 11 1 0 (I nverted '"0 2 ~ ~ 2 Lamb Dip) >, 02 0 ~ C1l ~"m c ~ [10'0,0"01 UJ Fluorescence 0 00°0

Detector (Saturation Resonance)

FIGURE 8. Graphic illustration of the saturation resonance observed in CO2 fluorescence at 4.3 J-lm. Resonant inter­ action occurs for ro = roo (when k . v =0). The figure shows an internal absorption cell within the laser cavity. External cells can also be used. width of the gain profile is about 53 MHz and is narrow and frequency-stable reference (absorption) line dominated by Doppler broadening. In a high-pressure within the operating range of the laser so that the laser (-760 Torr) COzlaser, the width of the gain profile is can be frequency-locked or frequency-offset-locked to dominated by pressure broadening, which is approxi­ the reference line. Initial attempts to use COz itselfas a mately 4 GHz/760 Torr for typical COrlaser gas mix­ reference via either the Lamb-dip or the inverted-Lamb­ tures. Thus even the ultrastable low-pressure (-15 Torr) dip techniques were not very successful. The poor re­ COz lasers are tunable: they may operate anywhere up sults were obtained because the lower-state rotational­ to at least tens ofMHz away from the center frequency vibrational levels of the COz laser transitions do not ofthe oscillating transition and will generally drift back belong to the ground state, and therefore the absorption and forth due to changes in power-supply current and coefficient of low-pressure room-temperature COz at ambient temperature. 10 J.lm is very low. The low absorption coefficient in In many applications such as high-precision spectros­ turn made it difficult to observe and utilize directly copy and multistatic radar it is highly desirable to find a the inverted Lamb-dip resonance in the full-power .... u0 C1l C1l 0- C1l > E . C-j t . ~ ,j i • 1:.. 5:-, I 1'= i"t l' . 1 l' _:. _ I ::J 0 -Q. -::J 0 Frequency

FIGURE 9. Lamb-dip-like appearance of the resonant change in the 4.3-J.lm fluorescence. The magnitude of the dip is 16.4% of the 4.3-J-lm fluorescence signal. The pressure in the reference cell was 34 mTorr and the laser power into the cell was 1.75 W in the I-P(20) transition. A frequency dither rate of 260 Hz was applied to the piezoelectric mirror tuner.

486 THE LINCOLN LABORATORY JOURNAL VOLUME 3. NUMBER 3, 1990 • FREED

Ultrastable CO2 Lasers output of the CO2 laser. These difficulties were over­ the absorption line center, a resonant change in the come at Lincoln Laboratory in 1970, when, at the sug­ 4.3-.um fluorescence signal appears. This standing-wave gestion ofA. lavan, we first demonstrated [30,31] that saturation resonance results from the nonlinearity ofthe excellent long-term frequency stability and reproduc­ interaction ofthe standing-wave field in the laser cavity ibility of CO2 lasers can be readily obtained (and with CO2 molecules (in the absorption cell) having greatly improved upon if necessary) by the frequency velocities resonant with the Doppler-shifred frequency stabilization of the lasers to the standing-wave satura­ of the field as experienced by the molecules. When the tion resonance observed in the upper-state-to-ground­ laser is tuned to the center frequency of a particular ~ state fluorescence of CO2, as graphically illustrated in transition (w = eua), a narrowband (i1f 1 MHz) reso­ Figure 8. nant dip appears in the intensity ofthe 4.3-.um fluores­ In the experimental procedure, low-pressure room­ cence signal. In this case, the traveling-wave compo­ temperature CO2 gas that serves as the saturable absorb­ nents constituting the standing-wave field interact with er is subjected to the standing-wave field of the laser the same group ofCO2 molecules, namely those mole­ cavity, with the laser oscillating in any preselected reg­ cules which have zero velocity in the direction of the 1 ular (00°1) or hot-band (01 l) transition. We can detect laser's optical axis (k. v = 0). This resonant change in the saturation effect by observing the change in the fluorescence signal is analogous to a Lamb dip. "intensity of the entire collisionally coupled 4.3-.um Figure 9 illustrates the Lamb-dip-like appearance of spontaneous fluorescence emission band as the laser the resonant change in the 4.3-.um fluorescence signal as frequency is tuned across the Doppler profile of the the laser is slowly tuned in frequency across the Doppler­ corresponding 10-,um absorption line. In the vicinity of broadened gain profile [32]. By applying a small fre-

- Voltage

~ 0 13 OJ OJ 0- OJ > E Cfl c OJ I : (/) . - I .- cD Cflro 0 ~ ....a.. 0 OJ OJ ~ 0 >

~ -0- ~ 0- + Voltage

fa Frequency

FIGURE 10. Derivative signal at 4.3 .urn in the vicinity of the standing-wave saturation resonance shown in Figure 9. SNR - 1000, M - ±200 kHz, and t = 0.1 sec (single pole).

VOLUME 3. NUMBER 3, 1990 THE LINCOLN LABORATORY JOURNAL 487 • FREED

Ultrastable CO2 Lasers

4.3-pm Fluorescence Stabilization Loop 1------..,

Mirror • Rare-I sotope 'I~ Laser V>. Beam Splitters...... /\ ~~~~~~~~~~~~~~~ 1-m Grating -- - 7 ~ ~ 12 16 Reference L--...... ". ,...... --.." C 02 Monochromator -- --/:/.c - - - - 0 ) Laser ..- "-----.... Y) ( t) t Attenuator 4.3-pm Fluorescence Stabilization Loop .....------1

HgCdTe Varactor Microwave Photodiode .....I------tt----l Local Oscillator Detector Laser Beat Note Frequency I Intermed iate­ Computer Spectrum Counter ~ Frequency I Analyzer Amplifier ~ Frequency - Counter I Printer I FIGURE 11. Block diagram of the two-channel line-center-stabilized CO2-isotope calibration system. In the figure, wavy and solid lines denote optical and electrical paths, respectively.

-10 quency modulation to the laser as it is tuned in

-20 the vicinity ofthe absorption line center, we can obtain the derivative ofthe 4.3-,um signal [32]. Figure 10, which -30 shows the 4.3-.um derivative signal, was obtained byap­ plying a ±200-kHz frequency modulation to the laser at ~ -40 152 dB a 260-Hz rate. A 1.75-W portion of the laser's output 0> 0 -50 ...J was directed into a small external stabilization cell that

-60 was filled with pure CO2 to a pressure of0.034 Torr at room temperature. It is a straightforward procedure to -70 J line-center-stabilize a CO2 laser through the use of the -80 4.3-,um derivative signal as a frequency discriminant, in 24.4 GHz conjunction with a phase-sensitive detector. Any devia­ tion from the center frequency of the lasing transition Frequency yields a positive or negative output voltage from the FIGURE 12. The 24.4104191-GHz beat note of a 16012C180 phase-sensitive detector. This voltage is then utilized as a laser I-P(12) transition and a 12C1602 laser I-P(6) transi­ feedback signal in a servoloop to obtain the long-term tion. The power levels into the photodiode were frequency stabilization ofthe laser output. 0.48 mW for the 16012C180 laser and 0.42 mW for the Figure 11 shows a block diagram of a two-channel 12C1602 laser. The second harmonic of the micro­ wave local oscillator was generated in the varactor pho­ heterodyne calibration system. In the system, two small, todiode. The IF noise bandwidth of the spectrum ana­ low-pressure, room-temperature COrgas reference cells lyzer was set to 10kHz. external to the lasers were used to line-center-stabilize

488 THE LINCOLN LABORATORY JOURNAL VOLUME 3, NUMBER 3, 1990 • FREED Ultrastable COL Lasers

two grating-controlled stable lasers. The two-channel ofthe frequency stability heterodyne system was used extensively for the measure­ ,------,--,---,--,----- 1 M 2 ment and calibration of CO -isocope-laser transitions 2 a y Cr) = 2M I,(Yj+l - Yj) . [33,34]. )=1 Figure 12 shows the spectrum-analyzer display of a Each measurement consisted of M = 50 consecutive typical beat note of the system shown in Figure 11. samples for a sample time duration (observation time) Note that the signal-co-noise ratio (SNR) is greater of 'l' seconds. Figure 13 shows that we have achieved 12 than 50 dB at the 24.4-GHz beat frequency of the a)'l') < 2 x 10- for 'l' - 10 sec. Thus a frequency two laser transitions [33]. measurement precision of about 50 Hz may be Figure 13 illustrates the time-domain frequency sta­ readily achieved within a few minutes. bility that we have routinely achieved with the two­ The triangular symbols in Figure 13 represent the channel heterodyne calibration system by using the frequency stability of a Hewlett-Packard (HP) model 4.3-,um-fluorescence stabilization technique [34]. 5061B cesium atomic frequency standard, as specified The blue solid and hollow circles represent two sepa­ in the 1990 HP catalog. Clearly, the frequency stabilities rate measurement sequences of the Allan variance ofthe COT and the cesium-stabilized systems shown in

10-10 Measurement Sequence 1 of the 0 • 2.6978648-GHz C02 Beat Note ~ 0 Measurement Sequence 2 of the • 2.6978648-GHz C02 Beat Note 0 HP 5061 B Cesium Atomic Standard 0 6. • CO2 Short-Term Stability \:' £. • ~ 0 b"" 10-11 .~ • .0 Q <0 0 6. U) 0 >. • i u 0 6. c • Ql 0 :J • 0- 0 ~ • ~ 12 • 0 0 L.I.. 10- •

• • 0.01 0.1 1.0 10 100

Sample Time, T(sec)

FIGURE 13. Time-domain frequency stability of the 2.6978648-GHz beat note of the 13C1802 laser I-R (24) transition and the 12C1602 reference laser I-P (20) transition in the two-channel heterodyne calibration system (Figure 11) with the 4.3-,um-fluorescence stabilization technique. For the sake of comparison, stabilities of a cesium clock and short-term stabilities of individual CO2 lasers are also shown. Note that the frequency stabilities of the CO2- and the cesium-stabilized systems shown are about the same and that the CO2 radar has achieved short-term stabilities of at least two to three orders of magnitude better than those of microwave systems.

VOLUME 3. NUMBER 3. 1990 THE LINCOLN LABORATORY JOURNAl 489 • FREED

Ultrastable CO2 Lasers

2 IIIII III I

~- 1 - -- .;. .~ ~ . . , .'".'. .0._ c -0.: .:_. '. o :;::; 0° : ....::••::•• ...... :'''',,: OJ o ~ o .' '.'.' .: . >, .' .,:.. " u .:.0. c .~.:.- .....,. OJ ': ::J c:r ' .~.. J: -1 ~------

-2 I IIII IIII NOON 2:00 4:00 6:00 8:00

Elapsed Time (h), Time of Day

FIGURE 14. Slow drifts in the 2.6978648-GHz beat frequency due to small frequency-offsetting zero-voltage variations of the electronics. The frequency deviations were caused by ambient temperature variations. The beat note was derived from the 13C1802 I-R(24) and the 12C1602 I-P(20) laser transitions. An observation time 'l' = 10 sec and a sample size of M = 8 were used for each data point.

Figure 13 are about the same. observable in Figure 14. The two magenta circles in the lower left corner of The frequency drift was most likely caused by small Figure 13 denote the upper bound of the shorr-term voltage-offset errors in the phase-sensitive detector-driv­ frequency stabilities, as measured in the laboratory en servoamplifier outputs that controlled the piezoelec­ (Figure 2) and determined from CO2 radar returns at trically tunable laser mirrors. Because 500 V was required the Lincoln Laboratory Firepond Facility (Figure 7). to tune the laser one longitudinal mode spacing of Note that the CO2 radar has achieved short-term stabili­ 100 MHz, an output voltage error of ±2.5 mV in ties of at least two to three orders of magnitude better each channel was sufficient to cause the peak-ftequen­ than those ofmicrowave systems. cy deviation of ±1 kHz that was observed in Figure Figure 14 shows the frequency reproducibility ofthe 14. By monitoring the piezoelectric drive voltage two-channelline-center-stabilized CO2 heterodyne cal­ with the input to the lock-in amplifier terminated ibration system. The figure contains a so-called drift run with a 50-Q load (instead of connected to the InSb that was taken over a period of 872 hours beginning at 4.3-,um fluorescence detector), we determined that slow 1 PM [34]. The frequency-stability measurement appa­ output-offset voltage drifts of up to ±2.5 mV were ratus was fully automatic; it continued to take, com­ indeed present in the electronics. These ±5-ppm pute, and record the beat-frequency data ofthe two line­ zero-voltage drifts were the most probable cause of center-stabilized CO2 isotope lasers even at night, when the ±1-kHz frequency drifts observed in Figure 14. no one was present in the laboratory. Approximately It is important to note that no special precautions every 100 sec the system printed out a data point that were taken to protect either the lasers or the associat­ represented the deviation from the 2.6978648-GHz ed electronic circuitty from temperature fluctuations in beat frequency, which was averaged over 872 hours. the laboratory. The fluctuations were substantial­ The system used a measurement time of 'l' = 10 sec plus or minus several degrees centigrade. Significant and M = 8 samples for each data point, yielding a improvements are possible with more up-to-date elec­ measurement accuracy much better than the tronics and a temperature-controlled environment; approximately ± I-kHz peak-frequency deviation these upgrades will indeed become necessary if the

490 THE LINCOLN LABORATORY JOURNAl VOLUME 3, NUMBER 3, 1990 • FREED

Ultrastable CO2 Lasers

,

I 01'1_11' 0 Hot Band IIi=RlI II-P IIi=RlI I-P fIi=Rll II-P I Ii=RlI I-P ~I II-P III-RlI I-P

~l II-P Ii=RlI I-P fIi=Rll II-P ~I I-P ~I II-P fI=Rll I-P

fIi=Rll II-P I II=RlI I-P

I I I I I I I IIII I II I I I I I I I I I 1150 1100 1050 1000 950 900 850 800 Wave Number (cm-') I I I I I II I 9.0 9.5 10.0 10.5 11.0 11.5 12.0 12.5

Wavelength (pm)

FIGURE 15. Frequency and wavelength domain of lasing transitions in nine CO2 isotopes. emphasis changes toward the creation ofa primary fre­ COrlsotope Lasers as Secondary quency standard. Frequency Standards Perhaps the greatest advantage ofthe 4.3-,um fluores­ cence stabilization method is that it automatically pro­ In CO2 lasers, transitions occur between two vibrational vides a nearly perfect coincidence between the lasing states ofthe molecule. Because each vibrational state has medium's gain profile and the line center of the satu­ a whole set of rotational levels, a very large number of rable absorber, because they both utilize the same laser lines, each with a different frequency (wavelength), molecule, CO2, Thus every P and R transition of the can be generated. Moreover, isotopic substitutions of (00°1)-1 and -II regular bands and the (01 11) hot band the oxygen and/or carbon atoms make 18 different may be line-center-locked with the same stabilization isotopic combinations possible for the CO2 molecule. cell and gas fill. Furthermore, as illustrated in Fig­ Approximately 80 to 150 regular-band lasing transi­ ure 8, the saturation resonance is detected separately at tions may be generated for each of the CO2 isotopic the 4.3-,um fluorescence band and not as a fractional specIes. change in the much-higher-power laser radiation at Through the use of optical heterodyne techniques, 10 j.Lm. At 4.3 j.Lm, InSb photovoltaic detectors that beat frequencies between laser transitions ofindividually can provide very high background-limited sensitivity line-center-stabilized CO2-isotope lasers in pairs were are available. accurately measured. As a result, the absolute frequen-

VOLUME 3. NUMBER 3, 1990 THE LINCOLN LABORATORY JOURNAL 491 • FREED Ultrastable CO, Lasers

cies, vacuum wave numbers, band cenrers, and rotation- We can also utilize harmonics and the difference .. ul C 12 l6 13 16 al-v1bratlonal molec ar constants lor C 0 2' CO2' frequencies ofCO2 lasing transitions to synthesize pre­ 13C1802' 12 18 , 12 170 , 16 12 18 , 16 13C 180, C 02 C 2 0 C 0 0 cisely known reference frequencies well beyond the 14 16 d 14 18 . ul I cal ul d CO2' an C 0 2 were Slm taneous y c ate 8.9-to-12.3-)lm range of the COrisotope-laser transi­ from well over 900 beat-frequency measurements [35, tion frequencies illustrated in Figure 15. For instance, 36]. The accuracies of these frequency determinations heterodyne comparisons with the second harmonics were, for the majority of the transitions, within about of appropriately selected CO2 laser transitions have 5 kHz relative to the primary Cs frequency standard achieved the first accurate determination ofseveral CO­ [36]. Consequently, in the wavelength region of 8.9 to isotope laser lines in the 5-to-6-)lm range [37]. Table 1, 12.3 )lm, line-center-stabilized COrisotope lasers which demonstrates infrared (IR) synthesis at 16 )lm (625 I can be conveniently used as secondary frequency cm- ) [38,39], is another example. To generate Table 1, standards. we gave a computer the task of finding the CO2-iso­ Figure 15 graphically illustrates the frequency and tope-laser transitions for which the difference frequencies wavelength domain ofthe nine CO2isotopic species that between a frequency-doubled 14C160rlaser transition have been measured to date. 14C 1602 extends the and any other COrisotope-laser transition fell within 1 wavelength range to well beyond 12 )lm; 12C 1802 the 625.0 ±0.1 cm- frequency range. Synthesized fre­ transitions can reach below 9 )lm. It should be noted quency tables very similar to Table 1 have been used for that the frequency tables generated and published by the accurate determination ofabsorption lines in isotopes Lincoln Laboratory have been extensively employed ofuranium hexafluoride (UF6) near 16 )lm. and reprinred in several publications and books in the Figure 16 illustrates a calibration method for locating United States and other countries. and precisely calibrating reference lines that can be easily

Table 1. An Example of IR Synthesis at 16 pm with Regular-Band C02 Transitions

Wave Number = (2 x Transition 1) - (Transition 2)

Wave Number (em -1) Transition 1 Transition 2

14 16 13 16 II-R (30) 624.906042 C 0 2 I-P (40) C 02 14 16 12 16 18 II-P (16) 624.915792 C 0 2 I-P (28) C 0 0 14 16 12 16 18 II-P (56) 624.917820 C 0 2 I-P (48) C 0 0 - 14 16 13 18 II-P (26) 624.937049 C 0 2 I-P (56) C 02 14 16 12 16 18 II-R (16) 624.945145 C 0 2 I-P (14) C 0 0 14 16 12 18 624.954490 C 0 2 I-P (4) C 0 2 II-R (26) 14 16 13 16 624.978786 C 0 2 I-P (48) C 0 2 II-R (6) 14 16 12 16 18 II-R (1) 624.988389 C 0 2 I-P (20) C 0 0 14 16 12 18 625.007851 C 0 2 I-P (16) C 0 2 II-P (4) 14 16 12 16 625.013741 C 0 2 I-P (6) C 0 2 II-R (56) 14 16 12 16 18 _ (52) 625.044477 C 0 2 I-P (2) C 0 11 R 14 16 14 18 II-R (52) 625.046863 C 0 2 I-P (54) C 02 14 16 12 16 18 II-R (45) 625.091263 C 0 2 I-P (4) C 0 0

492 THE LINCOLN LABORATORY JOURNAL VOLUME 3. NUMBER 3. 1990 -FREED

Ultrastable CO2 Lasers

Tunable Diode Microwave Local Laser Oscillator

Intermediate­ -----I L...... ".~...... ~"l~<'·"""'-~~I HgCdTe Varactor ---.. I --- Attenuator.. --r'\)~~-= ..:;. Photodiode ---+- Frequency )~ "-_Detector...... _. Amplifier )

NH.*3 Microwave Absorption Spectrum Cell Analyzer I

1-m Grating Monochromator

FIGURE 16. High-accuracy calibration method for heterodyne spectroscopy with tunable lasers. In the figure, wavy and solid lines denote optical and electrical paths, respectively.

used to determine the absorption spectra ofUF6isotopes beat note of the two lasers. The detected beat fre­ in the vicinity of 12 pm [38, 39]. In the experimental quency, which is generally in the 0-to-18-GHz range, arrangement, a beam splitter combines the output ofa is further heterodyned to some convenient inter­ tunable lead-salt diode laser and that ofa 14C16021aser. mediate frequency (IF) through the use of readily A fast HgCdTe varactor photodiode [40] heterodynes available commercial RF/microwave-frequency syn­ one part of the combined radiation, the beat note of thesizers and wideband double-balanced mixers. which is displayed and measured by a microwave spec­ The IF output is amplified and amplitude limited trum analyzer (or frequency counter). The other part of by means of low-noise wideband amplifiers and the combined laser radiation is used to probe an absorp­ limiters. The limiter output is, in turn, used as input tion cell that, in this particular experiment, is filled with to a wideband delay-line type of frequency discrimi-

NH3 gas at a pressure of 5 Torr. With the CO2 laser stabilized to its line center and the diode laser locked to the absorption line to be measured, heterodyne calibra­ tion provides an accuracy not currently available by any other method. As an example, Figure 17 shows a hetero­ dyne beat frequency of6775 MHz between a 14C 1602 laser and a diode laser tuned to one ofthe NH3 absorption lines near 12.1 pm. In another project at Lincoln Laboratory, we dem­ onstrated the equivalent of a programmable and highly accurate tunable IR synthesizer, as shown in Figure 18 [41, 42]. In the figure, the IR synthe­ sizer output is derived from a lead-salt tunable diode laser (TDL); a small portion of the TDL output

is heterodyned against a line-center-stabilized FIGURE 17. The 6775-MHz beat note of a 14C1602_laser grating-controlled CO2 or CO molecular laser. A 00°1-[10°0, 02°0] I-band P-transition and a diode laser high-speed HgCdTe varactor photodiode detects the tuned to an ammonia absorption line at 12.1 pm.

VOLUME 3. NUMBER 3. 1990 THE LINCOLN LABORATORY JOURNAL 493 -FREED

Ultrastable CO2 Lasers

Servo- Frequency Printer I I amplifier I Discriminator Current Li ne-Center- Supply Stabilized Microwave C0:2 or CO Synthesizer Computer I Reference Laser o to 18 GHz Bea plitters ) ) Tunable Diode Laser --..., Fast HgCdTe f-+- IF I Frequency (TDL) r~ Detector ® rl Amplifier I Counter Mixer

Grating Spectrum Spectrometer Analyzer

Frequency-Locked Diode Laser Output for Precisely Tunable High-Resolution Spectroscopy

FIGURE 18. Block diagram of an accurate, continuously tunable, computer-controlled, kilohertz-resolution infrared­ frequency synthesizer.

nator (200-to-600-MHz typical bandwidth). ble frequency modulation, which may be necessary to A servoamplifierlintegrator funher amplifies the out­ line-center-lock either one or both lasers. Thus to a great put of the frequency discriminator, and the amplified extent the absolute accuracy of the TOL output fre­ output is then used to control the TDL current, which quencyfmL will depend only on the absolute accuracy, determines the TDL output frequency. Closing the ser­ resettability, and long-term stability of the reference voloop in this fashion frequency-offset-locks the TOL molecular gas laser(s). To date, the most accurate results

output to the combination ofCO2 (or CO) laser, RF/ have been achieved with the use ofCO2 reference lasers. microwave synthesizer, and the center frequency of the wideband IF discriminator, which a frequency counter Laser Design accurately monitors. Most of the stable gas-laser oscillators we designed and A computer controls the entire infrared synthesizer constructed at Lincoln Laboratory have several com­ system shown in Figure 18. If, for instance, the micro­ mon features, described as follows. wave synthesizer is frequency swept under computer A nearly semi-confocal optical-cavity configuration is control, the IRoutput frequency oftheTOL would also used, which yields a ratio of relative diffraction loss of be swept in synchronism with the microwave synthesiz­ about 10 to 1 between the low-loss off-axis TEM10q er because the frequency-offset-locking servoloop forces mode and the desired fundamental TEMooq mode. In the TOL output to maintain the following frequency general, only fundamental TEMOOq-mode operation can relationship: overcome the combined losses, which are due to output coupling and diffraction. The lasers are dc-excited inter­

fTDL = fco2 1co ± fsynrnesizer ± fiF counter' (2) nal-mirror tubes in which four invar alloy rods rigidly space the mirror holders to achieve maximum open­ Either the operator or the computer program prede­ loop stability.

termines the frequency ofthe RF/microwave synthesizer Figure 19 is a photograph ofthe first stable CO2 laser in Equation 2. The IF is very accurately measured, and built at Lincoln Laboratory. This laser also bears the averaged ifso desired, even in the presence ofapprecia- closest resemblance to the Javan design. As mentioned

494 THE LINCOLN LABORATORY JOURNAL VOLUME 3. NUMBER 3. 1990 • FREED

Ultrastable CO2 Lasers in the introduction, this particular laser is a modified ning of 1968 [18]. In the new design, careful choice of version of the He-Ne lasers used in previous experi­ materials and techniques are employed for further en­ ments. The modifications involved changing two mir­ hancing the open-loop stability of the optical cavity. rors and the output window, and substituting a water­ However, in spite ofthe rigid structure, the laser design cooled discharge tube with a cold (not thermionic) is entirely modular and can be rapidly disassembled and cathode. In addition, a differential-screw tuning reassembled; mirrors can be interchanged, and mirror mechanism was affixed to one ofthe mirror holders to holders can be replaced by piezoelectric and grating­ permit a 3.5-pm-per-turn axial motion of one of the controlled tuners. The stainless steel endplates and the laser mirrors [26]. eight differential-alignment screws (Figure 19) have

The next set of lasers, built for the CO2 radar at been replaced by much more stable black diabase end­ Firepond, were in most respects similar to the one shown plates and a novel internal mirror-alignment mecha­ in Figure 19, except that superinvar rods with very low nism that is not accessible from the outside. The new coefficients ofexpansion were used for the optical-cavity lasers are not only more stable, but also much easier to spacers [26]. To the best ofthe author's knowledge, this align and less costly to manufacture. was the first use ofsuperinvar for the optical resonator of All the results described in the previous three sections a laser. Furthermore, acoustic damping, magnetic of this article were obtained with the third-generation shielding, and thermal insulation of the optical cavity CO2 lasers described above. The remainder ofthis sec­ was achieved by a variety ofmaterials surrounding each tion will show photographs of several variants of these superinvar rod in a concentrically layered arrangement. lasers and some of the experimental setups previously Viscous damping compounds, insulating foam, lead, given in block-diagram forms. Mu metal and Co-netic shields, and aluminum foil Figure 20 illustrates the simplest configuration: the provided this isolation ofthe rods. The shielded superin­ laser has two mirrors, one of which is piezoelectrically var cavity lasers yielded an additional factor-of-1 00 im­ tunable. Two-mirror lasers come in various lengths, de­ provement in short-term stability [26] compared to the pending on the output-power requirement. first two CO2 lasers shown in Figure 19. Four two-mirror lasers (lengths of 50, 50, 150, and The above lasers were soon replaced by a family of 236 cm) are contained in Figure 21. The figure shows completely redesigned third-generation CO2 lasers, which the vault that contains the various isotopic CO2 local and have been in use at Lincoln Laboratory since the begin- power oscillators for the optical radar at the Lincoln

FIGURE 19. The first stable CO2 laser built at Lincoln Laboratory.

VOLUME 3. NUMBER 3. 1990 THE liNCOLN lABORATORY JOURNAL 495 • FREED

Ultrastable CO2 Lasers

FIGURE 20. Basic two-mirror ultrastable laser.

Laboratory Firepond Facility. A fifth, grating-controlled first-order reflection ofthe grating was coupled through laser can also be seen in the figure. a partially reflecting output mirror. For heterodyne Figure 22 is a close-up photograph of a grating­ spectroscopy, zero-order output coupling from the grat­ controlled stable TEMooq-mode laser. Many variants of ing is preferable because many more laser transitions can this basic design exist both at Lincoln Laboratory and be obtained with such lasers. elsewhere. This particular unit was built for a relatively Three grating-controlled lasers with zero-order out­ high-power application such as optical pumping and put coupling are contained in Figure 23, a photograph frequency shifring. In the laser shown in Figure 22 the ofthe two-channel heterodyne measurement system of

FIGURE 21. CO2-isotope lasers in the oscillator vault of the Firepond radar.

496 THE LINCOLN LABORATORY JOURNAL VOLUME 3. NUMBER 3. 1990 -PREED

UltrastabLe CO2 Lasers

FIGURE 22. Basic grating-controlled stable TEMOOq-mode CO2 laser.

Figure 11. The two external frequency-stabilization cells, lasers with the 4.3-,um fluorescence technique has been used for the individual line-center locking of lasers in used in all countries (including the United States) that pairs, are also clearly visible in Figure 23. have set up laser chains to tie the frequency of visible Some of the lasers have shorr intracavity absorption lasers to the microwave cesium frequency standard. cells that can be used either for frequency stabilization or The third-generation CO2/CO lasers described above for very stable high-repetition-rate passive Qswitching. and shown in the last five figures are members of the Figure 24 shows a 50-em two-mirror laser with a shorr same family because most ofthe parts used in them are (3 em) internal absorption cell. This laser was the one completely interchangeable. Thus a new laser can be with which the 4.3-,um standing-wave saturation reso­ built or completely reconfigured with spare parts in a nance and the subsequent line-center stabilization of a very shorr period oftime, usually within a few hours.

CO2 laser were first demonstrated through the use of For more than two decades the dual requirements of the 4.3-,um fluorescence signal at Lincoln Laboratory in modularity of laser design and interchangeability of

1970 [30,31]. The line-center locking ofCO2-isotope parts have provided a vast amount of convenience and

FIGURE 23. The optical portion of the two-channel CO2 calibration system.

VOLUME 3. NUMBER 3. 1990 THE LINCOLN LABORATORY JOURNAL 497 -FREED

Ultrastable CO2 Lasers

FIGURE 24. Two-mirror stable laser with short intracavity cell. This laser was used for the first demonstration of the standing-wave saturation resonance observed via the 4.3-,um fluorescence signal.

savings both in time and cost. But such requirements standard. These results have enabled usage of the CO 2 have perforce introduced certain limitations in design system as the most precise secondary frequency standard and performance. Thus, if the need to do so arises, we currendy available in the infrared spectral region. could improve both short-term stability and long-term It should be noted that the results described in this stabilization of the lasers with relative ease by purring article were achieved with laser designs and components less emphasis on modularity and interchangeabilitY. In­ that were developed more than 20 years ago. Extensive deed, several years ago we prepared certain changes in experience gained by working with these lasers clearly design and even procured some of the components for indicates that updated designs could easily improve the this purpose. However, the instrumentation currendy short-term and long-term stabilities by at least one to available is not sufficient to measure definitively even two orders ofmagnitude. the stabilities ofour present lasers. Acknowledgments Conclusions The work reported in this article is the result of the

CO2 lasers have demonstrated greater spectral purity active encouragement and cumulative efforts of many and berter short-term stability than any other currendy individuals in the MIT/Lincoln Laboratory commu­ available oscillator at any frequency. This statement has nity. In particular, the author thanks the contributions been confirmed by laboratory measurements and by the of the following individuals (listed in alphabetical analysis of COTlaser radar returns from orbiting satel­ order): ].W Bielinski, L.c. Bradley, ]A Daley, T.R. lites. Furthermore, we also invented and developed at Gurski, HA Haus, A. lavan, R.H. Kingston, D.G. MIT/Lincoln Laboratory a highly effective long-term Kocher, R.G. O'Donnell, K.L. SooHoo, D.L. stabilization technique that can use low-pressure room­ Spears, L.]. Sullivan,].E. Thomas, and C.H. Townes. temperature CO2 gas as a reference; the technique has achieved long-term COTlaser stabilities at least compa­ rable to commercial-grade atomic clocks. A line-center­ REFERENCES stabilized two-channel COTlaser heterodyne calibration 1. C.K.N. Patel, "Continuous-Wave Laser Action on Vibration­ system was used to determine the absolute frequencies al-Rotational Transitions of CO2,'' Phys, Rev. 136, AilS? of lasing transitions and the rotational-vibrational mo­ (1964). 2. C.K.N. Patel, "Vibration Energy Transfer-An Efficient lecular constants ofnine CO2 isotopic species to within Means of Selective Excitation in Molecules," in Physics of about 5 kHz relative to the primary cesium frequency Quanturll Electronics Conference Proceedings, eds. P.L. Kelley,

498 THE LINCOLN LABORATORY JOURNAL VOLUME 3, NUMBER 3, 1990 - • FREED

Ultrastable CO2 Lasers

B. Lax, and P.E. Tannenwald (McGraw-Hill, 1966), pp. 23. AL. Schawlow and e.H. Townes, "Infrared and Optical 643-654. Masers," Phys. Rev. 112, 1940 (1958). 3. RH. Kingston and L.J. Sullivan, "Coherenr Laser Radar," in 24. TH. Maiman, "Stimulated Optical Radiation in Ruby," Proc. SP/E: 19th Annual Inti. Tech. Symp. on Optical Design Nature 187, 493 (Aug. 1960). Problems in Laser Systems 69, 10 (1975). 25. ED. Hinkley and e. Freed, "Direct Observation of the 4. L.J. Sullivan, "Infrared Coherenr Radar," Proc. SPIE: CO2 Lorenrzian Line Shape as Limited by Quanrum Phase Laser Devices andApplications 227, 148 (1980). Noise in a Laser above Threshold," Phys. Rev. Lett. 23, 277 5. L.J. Sullivan, B.E. Edwards, and W.E. Keicher, "The Fire­ (1969). pond High Power Carbon Dioxide Laser Radar System," 26. e. Freed, "Design and Short-Term Stability of Single-Fre­ Proc. Inti. Con! on Radar, Paris, 24-28Apr. 1989, p. 353. quency CO2 Lasers," IEEE] Quant Electron. QE-4, 404 6. "'Firefly' Laser Experimenr Successful in Measuring Inflat­ (1968). able Decoy Motion," Aviat. Week Space Techno!., p. 75 (23 27. e. Freed, "Progress in CO2 Laser Stabilization," in Proc. 31st Apr. 1990). AnnualSymp. on Frequency Control Atlantic City, Nj, 1-3 7. A lavan, W.B. Bennett, ]r., and D.R Herriott, "Population June 1977. Inversion and Conrinuous Optical Maser Oscillation in a Gas 28. e. Freed, "Ultrastable Carbon Dioxide (C02) Lasers," in Proc. Discharge Containing a He-Ne Mixture," Phys. Rev. Lett. 6, SP/E: Laser Research and Development in the Northeast 709, 106 (1961). 36 (1986). 8. TS. Jaseja, A Javan, and e.H. Townes, "Frequency Stability 29. TR. Gurski, "Laser Frequency Stability Limitations on a of He-Ne Masers and Measuremenrs of Length," Phys. Rev. 10.6 J1.m Laser Radar,"}. Opt. Soc. Am. 66, 1133A (1976). Lett. 10, 165 (1963). 30. e. Freed and A lavan, "Standing-Wave Saturation Reso­ 9. C.H. Townes, "Some Applications of Optical and Infrared nances in the CO2 10.6-J1. Transitions Observed in a Low­ Masers" in Advances in Quantum Electronics, ed. ].R. Singer Pressure Room-Temperature Absorber Gas," App!' Phys. Lett. (Columbia University Press, New York, 1961), p. 11. 17, 53 (1970). 10. L. Mandel and E. Wolf, eds., Selected Papers on Coherence and 31. A Javan and e. Freed, "Method of Stabilizing a Gas Laser," Fluctuations ofLight (Dover Publications, New York, 1970). U.S. Patenr No. 3686585, issued 22 Aug. 1972. 11. ].A Bellisio, e. Freed, and H.A Haus, "Noise Measuremenrs 32. e. Freed and R.G. O'Donnell, "Advances in CO2 Laser Sta­ on He-Ne Laser Oscillators," Appl. Phys. Lett. 4,5 (1964). bilizarion Using the 4.3 J1.m Fluorescence Technique," 12. e. Freed and H.A Haus, "Measuremenr ofAmplitude Noise Metrologia 13,151 (1977). in Optical Cavity Masers," App!. Phys. Lett. 6, 85 (1965). 33. e. Freed, "Frequency Stabilization of CO2 Lasers," in Proc. 13. e. Freed and H.A. Haus, "Photocurrenr Spectrum and Pho­ 29th Annual Symp. on Frequency Control, Atlantic City, Nj, toelectron Counrs Produced by a Gaseous Laser," Phys. Rev. 28-30 May 1975. 141,287 (1966). 34. K.L. SooHoo, C. Freed, J.E. Thomas, and H.A. Haus, "Line­ 14. C. Freed and H.A Haus, "Photoelectron Statistics Produced Cenrer Stabilized CO2 Lasers as Secondaty Frequency Stan­ by a Laser Operating below the Threshold of Oscillation," dards: Determination of Pressure Shins and Other Errors," Phys. Rev. Lett. 15,943 (1965). IEEE J Quant. Electron. QE-21, 1159 (1985). 15. e. Freed and H.A. Haus, "Amplitude Noise in Gas Lasers 35. e. Freed, L.e. Bradley and R.G. O'Donnell, "Absolute Fre­ below and above the Threshold of Oscillation," in Physics of quencies of Lasing Transitions in Seven CO2 Isotopic Spe­ Quantum Electronics Conference Proceedings, eds. P.L. Kelley, cies," IEEE}. Quant. Electron. QE-16, 1195 (1980). B. Lax, and P.E. Tannenwald (McGraw-Hill, 1966), pp. 36. L.e. Bradley, K.L. SooHoo, and e. Freed, "Absolure Fre­ 715-724. quencies of Lasing Transitions in Nine CO2 Isotopic Spe­ 16. e. Freed and H.A. Haus, "Photoelectron Statistics Produced cies," IEEE] Quant. Electron. QE-22, 234 (1986). by a Laser Operating below and above the Threshold of 37. RS. Eng. H. Kildal, J.e. Mikkelsen, and D.L. Spears, "De­ Oscillation," IEEE] Quant. Electron. QE-2, 190 (1966). termination ofAbsolute Frequencies of 12C 160 and 13C l60 17. e. Freed, "Stability Measuremenrs of CO2-N2-He Lasers at Laser Lines," Appl. Phys. Lett. 24, 231 (1974). 10.6 pm Wavelength," IEEE}. Quant. Electron. QE-3, 203 38. e. Freed, "Status of CO2 Isotopic Laser Technology," Proc. (1967). Inti. Con! on Lasers, New Orleans, 15-19 Dec. 1980. 18. e. Freed, "Designs and Experimenrs Relating to Stable Gas 39. e. Freed, "Status of CO2 Isotope Lasers and Their Applica­ Lasers," in Proc. Frequency Standards and Metrology Seminar, tions in Tunable Laser Spectroscopy," IEEE] Quant. Elec­ National Research Council of Canada and Universite Laval tron. QE-18, 1220 (1982). Quebec, Canada, 30Aug.-l Sept. 1971, p. 226. 40. D.L. Spears and e. Freed, "HgCdTe Varactor Photodiode 19. C. Freed, "Sealed-Off Operation ofStable CO Lasers," Appl. Detection of CW CO2 Laser Beats beyond 60 GHz," Appl. Phys. Lett. 18,458 (1971). Phys. Lett. 23, 445 (1973). 20. e. Freed and H.A. Haus, "Lamb Dip in CO Lasers," IEEE] 41. e. Freed, ].W. Bielinski, and W. Lo, "Programmable, Sec­ Quantum Electron. QE-9, 219 (1973). ondary Frequency Standard Based Infrared Synthesizer Using 21. AL. Betz, M.A. Johnson, R.A. Mclaren, and E.e. Sutton, Tunable Lead-Salt Diode Lasers," in Proc. SP/E: Tunable Di­ "Heterodyne Detection of CO2 Emission Lines and Wind ode Laser Development and Spectroscopy Applications 438, 119 Velocities in the Atmosphere ofVenus," Astrophys.]. 208, pt. (1983). 2, L141 (15 Sept. 1976). 42. e. Freed, "Programmable, Secondary Frequency Standards 22. M.A. Johnson, AL. Betz, R.A. Mclaren, and e.H. Townes, Based Infrared Synthesizers for High Resolution Spectrosco­ "Nonthermal 10 Micron CO2 Emission Lines in the Atmo­ py," in Methods ofLaser Spectroscopy, eds. Y. Prior, A Ben­ spheres ofMars and Venus," Astrophys.}. 208, pt. 2, L145 (15 Reuven, and M. Rosenbluh (Plenum Press, New York, 1986), Sept. 1976). pp.151-161.

VOLUME 3. NUMBER 3, 1990 THE LINCOLN LABORATORY JOURNAL 499 .... - • FREED

Ultrastable CO2 Lasers

CHARLES FREED was born in Budapesr, Hungary, in 1926. He received a B.E.E. degree from New York University in 1952, and S.M. and E.E. degrees in elecrrical engineering from MIT in 1954 and 1958, respecrively. From 1952 ro 1958 Charles was employed by rhe Research Laborarory of Elecrronics ar MIT as a research assisranr and sraff member. He larer worked ar Raytheon Company's Spencer Laborarory, Special Microwave Devices Operarion, and Research Division. He joined Lincoln Laborarory as a staffmember in February 1962, and became a member of me senior sraff in 1978. Since 1969 he has also been a lecturer in rhe Deparrmem of Elecrrical Engineering and Computer Science ar MIT. He has wrirren numerous rechnical papers on a variety of ropics, including microwave noise on elecrron beams; varacror diode paramerric amplifiers; amplitude noise and phoroelecuon starisrics of He-Ne lasers; me frequency srability, design, and Output characrerisrics ofsealed-off srable CO2 and CO lasers; long-rerm srabilization ofCO, and CO lasers; precision ­ hererodyne calibrarion; varacror phorodiode derecrion; me dererminarion ofCO,-isorope laser rransirions; and CO2 laser amplifiers. Charles is a member ofTau Bera Pi, Era Kappa Nu, and Sigma Xi. In 1979 he was elecred a Fellow of me IEEE wim me following ciration: "For comribution ro gas lasers and rhe pioneering develop­ mem of ultrasrable lasers."

500 THE LINCOLN LABORATORY JOURNAL VOLUME 3. NUMBER 3. 1990