evaluation of nd:yag-pumped raman shifter as a broad-spectrum light source

Evaluation of Nd:YAG-pumpedRaman shifter as a broad-spectrum light source

George B. Jarvis, Sam Mathew, and Jonathan E. Kenny

We have examined the utility of a gas-filled, Nd:YAG-laser-pumped Raman shifter as a possiblebroad-spectrum light source. Six to nine new output frequencies with pulse energies above 1 uJ areproduced when a pure-hydrogen or pure-methane Raman shifter is pumped with 40 mJ of second-harmonic, 20 mJ of third-harmonic, or 11 mJ of fourth-harmonic pump pulse energy. Optimum outputoccurs at pressures of approximately 10 atm for the pure-gas experiments. We also report the outputfrequencies and pulse energies of a mixed hydrogen-methane Raman shifter pumped by 20 mJ of thethird harmonic of the laser for various proportions of the two gases at pressures up to nearly 20atm. Depending on composition and pressure, over a dozen new output lines with pulse energies over 1tiJ can be produced. We discuss the nonlinear processes involved, the optimum operating conditions,and the suitability of the source for our application of groundwater monitoring.

Key words: Stimulated Raman scattering, hydrogen-methane mixtures, broad-spectrum light source,nonlinear optics, wavelength conversion, Raman shifting.

1. Introduction

We have been involved in recent years in the develop-ment of field-portable spectroscopic instrumentationusing lasers and fiber optics for in situ monitoring oforganic contaminants in groundwater. Our firstinstrument' used the fourth harmonic of a Nd:YAGlaser at 266 nm as a light source to excite fluorescencein contaminants. The total fluorescence signal ob-tained provided a relative measure of pollution butdid not permit identification or quantification of thesubstance or substances that produced the response.To detect chemical species that do not absorb at 266nm and to develop a second-generation instrumentcapable of identifying and quantifying individualgroundwater constituents using excitation-emissionmatrices,2 we sought a method of generating multipleexcitation wavelengths in the UV and visible regionsof the spectrum.

Important factors in the choice of a light sourceinclude the availability of suitable wavelengths (orthe suitability of available wavelengths), beam diam-

When this work was performed the authors were with theDepartment of Chemistry, Tufts University, Medford, Massachu-setts 02155. G. B. Jarvis is now with Thermetics Detection, Inc.,220 Mill Road, Chelmsford, Massachusetts 08124.

Received 16 August 1993; revised manuscript received 24 Novem-ber 1993.

0003-6935/94/21493809$06.00/0.© 1994 Optical Society of America.

eter and divergence, instantaneous and average power,linewidth, cost, size, ease of operation, and utilityrequirements. For our application, a Raman shifter,a dye laser, and an arc lamp were initially considered.We expected the first two to be superior in terms ofthe potential to be coupled into optical fibers and toprovide high sensitivity, especially in the UV, wheresignificant fiber losses must be overcome. Of thetwo laser-based sources, the Raman shifter is farsimpler and cheaper than the dye laser. The pur-pose of the present study was to determine whetheradequate spectral coverage could be obtained with theRaman shifter, which lacks the continuous tunabilityof the dye laser.

The literature on stimulated Raman scattering(SRS), which was recently reviewed by White,3 con-tains reports of broad spectral coverage through theuse of a Raman shifter in combination with a dyelaser with 1 to 3 dyes," a significant improvementover the 10 to 12 (or more) dyes normally needed by adye laser, along with second-harmonic-generatingcrystals, to cover the UV and visible spectra from200-700 nm. Our application requires dense, butnot continuous, spectral coverage. The observationof combination and difference frequency shifts fromgas mixtures,7'8 along with the availability of severalpump frequencies from a Nd:YAG laser equippedwith harmonic generators, suggested that the dyelaser might be eliminated altogether for this andperhaps a number of other spectroscopic applications.

4938 APPLIED OPTICS / Vol. 33, No. 21 / 20 July 1994

However, little relevant performance data was avail-able in the literature, perhaps because many sharedthe sentiments of one group8 who noted that, withgas-mixture Raman shifters, pump power was distrib-uted into so many lines that obtaining a seriousamount of energy at any one frequency was difficult.(Lamps have somehow survived as spectroscopic lightsources despite their being subject to a similar objec-tion.) We suspected that many of the lines might beabove the threshold for utility in our application, sowe pressed on.

Design considerations for a Raman shifter includethe pump laser properties, the Raman medium, andthe cell design. These are discussed at length else-where.3 We chose to investigate a single-pass Ra-man generator that uses the second, third, and fourthharmonics of a pulsed Nd:YAG laser as pump beamsand also uses a hydrogen-methane mixture as theRaman medium. These harmonics provide pumpfrequencies in the visible, near V, and far UY thatare separated by only a few multiples of the activevibrational frequency of either Raman gas. Hydro-gen, methane, and mixtures of the two gases werechosen because of the simple, strong stimulatedRaman spectrum of each, their miscibility and stabil-ity, their transparency throughout the spectral re-gion of interest, and the favorable values of theirRaman shifts: 4155 cm-' for hydrogen and 2916cm-' for methane. Combinations of these gaseswere expected to yield an acceptable density of reason-ably strong output wavelengths. In this paper wepresent relevant performance data for this system.Further comparison of the various light sources forour application appears in detail elsewhere.9

2. Experiment

Shown in Fig. 1 is a block diagram of the Raman-shifter instrumentation. The pump source is apulsed Nd:YAG laser (Quanta-Ray DCR-11) with a1064-nm fundamental and donut-shaped beam profile.The three harmonics at 532, 355, and 266 nm have6.5-, 5.5-, and 4.5-ns pulse widths, respectively, accord-ing to the manufacturer. The laser beam diameteris 6.4 mm for all harmonics. The pump polarizationwas selected to minimize the reflections from thePellin-Broca prisms.

The gas cell is custom built from type-304 stainlesssteel and is a cylinder 75 cm long with a 1-cm inside

PB1 adA Lase OA

Fig. 1. Block diagram of Raman shifter. The pump beam is oneof the three harmonics (second, third, or fourth) of the Nd-YAGlaser generated by the harmonic generator (HG) and selected bythe first Pellin-Broca prism (PB1). A second Pellin-Broca prism(PB2) disperses the output of the Raman shifter into the opticalattenuator (OA).

diameter. The 2.54-cm diameter, 1-cm-thick fusedsilica end windows have 0-ring-sealed flanges canted20 to provide a means of dumping the strong backreflections from the focused pump beams, to preventdamage to the laser and the harmonic-generatingcrystals. Grade 4 or ultra-high-purity hydrogen andmethane (Linde) was used for all experiments.When the gas composition was changed, the cell wasflushed three times with the new gas or the gasmixture before the final fill.

Two right-angle deflections were used to direct thelaser beam into and out of the Raman cell. Thisgeometry provides a safety measure against damageto the laser in case of window explosion. A pumplaser beam of horizontal polarization was focused intothe center of the Raman cell with a 75-cm-focal-length fused-silica lens. The exiting diverging beam,which consists of pump, Stokes, and antistokes fre-quencies, was approximately collimated with a second75-cm-focal-length lens and directed to a secondPellin-Broca prism positioned near the Brewsterangle and mounted on a rotary stage. The desiredoutput frequency of the Raman shifter was selectedby adjustment of this prism and its alignment withthe entrance to the optical attenuator. The attenua-tor was placed approximately 1 m from the prism toallow separation of adjacent output frequencies.The optical attenuator' 0 allows adjustment of thebeam intensity without a change in the operatingparameters of the laser or Raman shifter. Theattenuation is necessary in our application to preventdamage to the proximal end of the optical fiber for thehighest output pulse energies.

Throughout all experiments, pulse energies of thestimulated Raman output were measured with aspectrally flat joulemeter (Laser Precision RjP-735)and ratiometer (Laser Precision Rj-7200), and (time-averaged) pump powers were measured with a calorim-eter (Laser Precision RT-30). The pump pulse ener-gies used for all experiments were 40, 20, and 11 mJfor 532, 355, and 266 nm, respectively, at a 10-Hzrepetition rate. Transmission losses of the fused-silica optics near 200 nm were not quantified. In thefollowing discussion pressure is given in psi (gauge);the conversion to Pa is as follows:

1.01 x 105 PaP(pascals) = P[psi (gauge)] + 14.7} 14 P

In the first set of experiments, the Raman cell wasfilled to 300 psi (gauge) with pure hydrogen or meth-ane. Because the stimulated Raman spectrum ofeach gas consisted of multiples of only one shift, weidentified the different orders by simply countingfrom the pump on a card placed approximately 2 mfrom the final dispersive prism. The polarization ofeach of the visible stimulated Raman scattering out-put beams was compared to that of the 532- and355-nm pumps with a sheet of Polaroid of knownpolarization. Several 100-shot averages of pulse en-ergy of all stimulated Raman frequencies of detect-

20 July 1994 / Vol. 33, No. 21 / APPLIED OPTICS 4939

able intensity were measured downstream of theoptical attenuator (which was set at minimum attenu-ation) and recorded. The transmission of the disper-sive elements and optical attenuator was approxi-mately 50%. The measurements were repeated for20 psi (gauge) decrements in total gas pressure untilno stimulated Raman scattering was observed.

The second set of experiments was conducted withhydrogen-methane mixtures. We performed a de-tailed study of the output power of the stronger linesas a function of gas composition and total pressure,using only the 355-nm pump. We identified andmeasured Stokes and anti-Stokes output for thefollowing partial pressure ratios R (defined as thehydrogen:methane ratio): 0.5, 0.82, 1.0, 1.25, 1.5,and 2.0, using the method outlined above. Toidentify the lines, a card was placed 2 m from the finaldispersive prism, and the positions of the shifts weremarked as the Raman gas was systematically changedfrom pure hydrogen to pure methane. For a givengas composition, the pulse energy of each outputfrequency was recorded at 25 psi (gauge) intervalsfrom 300 to 0 psi (gauge). We also performed adetailed spectroscopic study of the output from amixed-gas Raman shifter. The attenuated total Ra-man-shifter output from a R = 1.5 mixture at a totalpressure of 375 psi (gauge) was directed into a 0.35-mscanning monochromator (McPherson) that was fit-ted with a photomultiplier tube (Hamamatsu R928)and controlled by a personal computer. The mono-chromator was stepped at 0.2-nm intervals, with abandpass of 0.6 nm, through the spectral regioncorresponding to the range of output frequencies foreach of the three pump frequencies.

3. Results and Discussion

A. Pure Gases

The polarization of the SRS output beams was deter-mined to be the same as that of the input laser beamfor both the Stokes and anti-Stokes scattering, asexpected for the totally symmetric vibrational modesinvolved. The pulse stability of the stimulated Ra-man scattering was quantified as the percent relativestandard deviation of several 100-shot averages ofpulse energy measured with the joulemeter. Therelative standard deviation was 1% or less for eachpump wavelength and between 1 and 5% for the vastmajority (nearly 90%) of stimulated Raman outputlines. For example, for H2 pumped by the secondharmonic, the RSD's of anti-Stokes orders 1-6 at 60psi (gauge) were 0.5, 1.0, 1.1, 2.8, 3.7, and 2.9%,respectively. The dependence of forward-scatteredstimulated Stokes and anti-Stokes Raman outputpulse energies on cell pressure for pure hydrogen isshown in Figs. 2(a)-2(c) for pump wavelengths of 532,355, and 266 nm, respectively. The results for puremethane with the same pump wavelengths are shownin Figs. 2(d)-2(f).

For each pure gas and each pump wavelength,between six and nine new frequencies were generatedwith sufficient intensity to be measured; these in-

cluded up to fifth-order Stokes and sixth-order anti-Stokes lines in individual cases. For the most part,the same lines were observed for both gases at a givenpump wavelength, and corresponding lines of hydro-gen and methane had similar pulse energies (usuallywithin a factor of 2).

All output lines showed some pressure dependence;generally the pulse energy increased to a maximum,then decreased as pressure increased, and the sharp-ness of the maximum increased with increasing order.The hydrogen lines tended to reach slightly highermaximum pulse energies than did the correspondingmethane lines (except with the 532-nm pump), andthese generally occurred at slightly lower pressures.Furthermore, pulse energy generally decreased withincreasing order at all pressures. These trends areconsistent with the results of other pressure-depen-dence studies.5. 81 The first and second Stokes linesshowed the only strong deviations from these generaltrends, with the first order stronger than the secondat low pressures, the curves for the two orderscrossing at intermediate pressures, and both showinga relatively weak pressure dependence beyond thecrossover pressure.

The approximate equality of the pulse energies ofthe first Stokes lines of hydrogen and methane is atfirst surprising, given that the steady-state gaincoefficient,'2 y (also called g or g. in the literature), atany given wavelength and pressure studied here isseveral times larger for hydrogen. (Although theRaman cross section for methane is larger by nearly afactor of 3, the linewidth for the hydrogen transi-tion13 is nearly 20 times narrower, with y propor-tional to the cross section/linewidth.) However,simple steady-state gain expressions ignore depletionof both the pump and the first Stokes line by otherRaman and parametric processes and are not strictlyapplicable here. Although we measured only trans-mitted pump pulse energy for 266-nm pumping (forwhich case it was approximately 20%), we can inferstrong pump depletion in all cases by noting that thetotal photon-conversion efficiency in all measuredStokes and anti-Stokes lines for each gas was over40% at optimum pressures. Given the approxi-mately 50% loss of the dispersing optics and theoptical attenuator, as much as 80% or more of thepump was converted to photons at other frequencies.That the first Stokes line was significantly depletedby the higher-order lines is strongly suggested by theobservation that the second Stokes was often stron-ger.

The pressure dependence of the Raman-shifteroutput lines falls into two general cases, as sketchedabove: (1) the first two Stokes orders, and (2) every-thing else. These pressure dependencies can be ratio-nalized if we assert that direct SRS contributessignificantly to lines in only the first case, and thatparametric four-wave-mixing processes dominate thesecond case. Qualitatively we may predict first andsecond Stokes behavior as a function of pressure byexamining the pressure dependence of the gain coeffi-


30

2900 3400 3900 4400 4900 5400 5900 6400 6900 6?

Wavelength, AWavelength, A

Wavelength, A Wavelength, A

1900 2400 2900 3400 3900 4400 4900 5400 5900 6400 69003 ?

Wavelength, A o 1900 2400 2900 3400 3900 4400 4900 5400 5900 6400 6900 2Wavelength, A

Fig. 2. Log of pulse energy in microjoules versus wavelength and pressure for output from a Raman shifter with pure gas [hydrogen for(a)-(c), and methane for (d)-(f)] as the working fluid. Pump wavelengths are 266 nm for (a) and (d), 355 nm for (b) and (e), and 532 nm for(c) and (f). For the 266-nm case only, the energy of the depleted pump beami is also shown. In the case of the 532-nm data, the secondStokes pulse energies (at 9543 nm in the IR) were measured but were outside the range shown.

cient y, given by

fAVpz= A (1)

where f(X) is a function of the wavelength,12 where pis the density in amagats and is proportional topressure to within 4% over the range studied,' 4 andAv is the bandwidth of the spontaneous Raman linegiven by 13

309AVH 2 (MHz) =-+ 51.8p, (2)

p

for the QI(1) line of hydrogen at 4155 cm-' and

psi (gauge)]. For methane, the first term on theright in Eq. (3) is larger than the second over theentire range studied here, so that the gain increasesrelatively faster with pressure. For both gases, theabsolute increase in the calculated gain coefficient ongoing from 100 to 300 psi (gauge) is approximately thesame, 0.5 x 10-1" m/W. Thus the first Stokes lineincreases rapidly with pressure, then tends to level offor even fall if depleted severely enough by the genera-tion of higher-order lines. In addition to directpumping of the second Stokes line, the first Stokesmay be depleted by parametric processes like thefollowing:

2k- = k- 2 + k, (4)

AvcH4 (MHz) = 8220 + 384p, (3)

for the v, vibrational transition of methane at 2916cm-'. These equations describe a hydrogen gainthat increases slowly with pressure for pressures atwhich the 3 0 9/p term is appreciable, then becomespressure independent for larger pressures [the twoterms contribute equally to the linewidth at p = 24

where k -, k -2, and k are the wave vectors of the firstStokes, second Stokes, and pump, respectively, and

k + k-1 = k- 2 + kj, (5)

where k, is the wave vector of the first anti-Stokesline.'5 16 Both of these processes, it is seen, alsobuild up the second Stokes line. Because the higher-


3.5.3

2.52.

1.51

0.5

order lines disappear eventually at higher pressuresas a result of increasing dispersion, the first Stokesmay actually grow again at high pressure. 5 Thesecond Stokes behavior is analogous to the extentthat it is generated by a true Raman process (i.e., anymismatch between the incoming and scattered pho-ton momenta is taken up by the molecule undergoingthe vibrational transition, so the increase in disper-sion of the gas with pressure has no deleteriouseffect), but its initial growth with pressure is slowerbecause it depends on the buildup of the first Stokes.

Lines belonging to the second case, i.e., generatedprimarily by parametric four-wave mixing, will in-crease with pressure more slowly than the firstStokes line, because their production depends on theintensity of the first Stokes line (and higher-orderlines in general depend on intensities of lower-orderlines). But the phase-matching conditions that mustbe met are more difficult to achieve as pressure isincreased, because of the increase in refractive indexn of the gas, with n2

- I proportional to p. Becausethe dispersion of methane is significantly larger thanthat of hydrogen, this effect becomes more dramaticin methane, and intensities fall off faster when pres-sures increase beyond the optimum pressure. Fur-thermore, because refractive-index differences arelarger for larger frequency shifts, higher-order lineswill fall off faster. Finally, the normal dispersion ofhydrogen and methane means that phase mismatchis a much greater problem for shifts to higher fre-quency than to lower frequency. These trends arewell displayed in the plots of Fig. 2. For example,the fourth, fifth, and sixth anti-Stokes lines of meth-ane pumped at 532 nm disappear completely at only afew pounds per square inch after they pass throughtheir respective maxima.

Although we have not attempted an exhaustivequantitative analysis of the specific Raman and para-metric processes contributing to the intensity of eachline, we have demonstrated that experimental condi-tions may be found that permit approximately equalconversion of pump photons into corresponding SRSlines of pure hydrogen and pure methane. Achiev-ing such a balance in the component gases wouldappear necessary if a mixture of the two is to functionin the desired way, i.e., distributing significant pumpenergy into many lines of both pure gases and intotheir combination lines. Furthermore, the pressure-optimized pulse energies for a given line were remark-ably similar as the pump changed from the second tothe third and fourth harmonics of the Nd:YAG laser,e.g., the first Stokes lines of both gases were always inthe range 1000-4000 p.J, and the first anti-Stokeslines were always between 100 and 400 J. Thisresults largely because the decrease in pump pulseenergy with increasing harmonic or decreasing wave-length is balanced by the increasing Raman crosssection with decreasing wavelength because of theresonance-enhancement effect.'2 This property ofthe medium is desirable in a system intended to be amultiple-wavelength source, because a whole new set

of output lines may be easily accessed if a differentpump harmonic is chosen, which is an easier changeto accomplish than switching to a new medium.

The pressure dependence of stimulated Ramanoutput resulted in a maximum number of Stokes andanti-Stokes wavelengths being generated at low pres-sures. For anti-Stokes scattering, optimized pulseenergies occurred at an average pressure of 125 psi(gauge). The higher-order (n = 3,4 .. .) Stokes pulseenergies were optimized at an average of 114 psi(gauge). Thus, a low-pressure cell at 120 psi(gauge) should produce high pulse energies for alarger number of stimulated Raman scattering wave-lengths. [The first and second Stokes were notincluded in the calculation of these averages becausethey have excellent conversion at all pressures greaterthan 40 psi (gauge).]

B. Gas Mixtures

Previous workers have observed new SRS lines as aresult of combinations of vibrational shifts of thegases.78 These new frequencies occur at sums anddifferences of the hydrogen and methane vibrationalshifts according to the following:

(6)VC = VL + jVH + kvM,

where vc is the frequency of the combination shift, i isthe harmonic of the laser, vL is the fundamentalfrequency of the laser (9395 cm-', according to themanufacturer), vH is the hydrogen vibrational shift(4155 cm-' for the Q1(1) line 7), vM is the methanevibrational shift [approximately 2916-2917 cm-' forthe Q branch of v1 (Ref. 18)], and j and k are integersand may be positive, negative, or zero.

1. Dependence of Pulse Energies on Pressure andCompositionIn Fig. 3, the pressure dependencies of the variousindividual and combination shifts are plotted forhydrogen:methane pressure ratios R = 0.5, 0.82, 1,1.25, and 2, respectively, with 355 nm as the pumpwavelength. (The data from the R = 1.5 mixtureare not shown; qualitatively, they show most of thesame trends, but there are quantitative differencesbecause of minor long-term drift between data sets.)Our data indicate that combination shifts were gener-ated for R between 0.82 and 1.5. Outside this rangeof pressure ratios, only single-component stimulatedRaman scattering was generated. Analysis of Fig. 3indicates the optimum value of R appears to be near1.0 for generating combination shifts with the 355-nmpump wavelength. Duardo and co-workers,7 using aruby laser at 694.3 nm as the pump, reported that therange of R that generated combination shifts was 1.3to 2, with up to 955 psi (gauge) total cell pressure.We suspect the slight difference may be due to thelarger Raman cross section at shorter pump wave-length.

A discussion of the mixture results is facilitated bythe definition of the absolute order of a line as thesum of the absolute values of its j and k values. In


/0.51 H:M -3.

Wave length, A

2700 3200 3700 4200 4700 5200 giWavelength, A

Wavelength, A

0.5 -~.1 . :.§ 252700 3200 \3700 4200 4700 5200 d

2 (0 2Wve th ,-2) L

2~~~~~~aeegh -A.

Fig. 3. Log of pulse energy in microjoules versus wavelength andpressure for the output lines from a Raman shifter pumped at 355nm and operating with various hydrogen:methane mixture ratiosas indicated on the graphs, with the methane fraction increasingdownward. Assignments with the (j, k) notation are also given.Note the displaced wavelength scale for the R = 2.0 plot.

the data set shown in Fig. 3, pulse energies weremeasurable at various gas pressures and composi-tions (a) for all 4 lines with absolute order 1: (j, k) =(0, 1), (0, -1), (1, 0), and (-1, 0); (b) for all 8 lineswith absolute order two: the 4 pure-gas lines,(-1, -1) and (1, 1) at ±7071, and (1, -1) and (-1, 1)at +1239; and (c) for 6 out of 12 lines with absoluteorder 3: the 4 pure-gas lines and (-1, 2) and (1, -2)at 1677. The combination lines conspicuous bytheir absence are (2, -1) and (-2, 1) at ±5394; (1, 2)and (-1, -2) at 9987; and (2, 1) and (-2, -1) at±11 226. We expect that an important parametricprocess in the generation of combination lines (inanalogy to that described by Zubov'5) is

2k = k(j,k) + (-j,-k)2 (7)

where k(jk) is the wave vector of a combination line.Evidence of the importance of this particular process(7) is the fact that pairs of combination lines (j, k 0)given by ±(j, k), which we call parametric pairs, wereeither both observed or both not observed in ourmixture experiments. Furthermore, when observed,members of the pair had roughly similar pulse ener-gies and pressure dependencies. An exception isprovided by the pair (-1, -1) and (1, 1), where theformer is much stronger than the latter. However,in that case we expect a significant contribution bytwo direct pumping processes because of the strengthof the first Stokes line of each gas. Overlooking theexceptional case, we see that parametric pairs ofabsolute order up to 3 were observed only when theshifts from the pump frequency were less than ap-proximately 2000 wave numbers, with increased dis-persion and phase mismatch presumably makingprocess (7) less effective for larger shifts.

We now address the pressure dependence of thecombination lines. Qualitatively, we saw the mostlines at low pressures, approximately 10 atm, in thepure-gas experiments because at higher pressures thehigher-order lines disappeared. In the case of themixtures, the higher-order lines of the pure gases,i.e., (j, 0) or (0, k) rapidly disappear as the second gasis added, and the multitude of output frequenciescomes from combination lines with relatively lowabsolute order, and, as noted above, relatively smalldisplacement from the pump. At low values of R,e.g., 0.82, these combination lines achieve maximumpulse energy at intermediate pressures, followed bydecline at higher pressure. But as R increases to 1.0,their pressure dependence gets flatter, and at R =1.25 the pulse energies are still increasing at thehighest pressures measured in this work.

2. Dispersed Spectra and AssignmentsThe dispersed spectra of the Raman shifter output fora 1.5:1 hydrogen:methane mixture are shown in Fig.4 for 532-, 355-, and 266-nm pump wavelengths fromtop to bottom, respectively. Because no order-sorting filters were used, all peak assignments weremade with the explicit consideration that they mightin fact be first-, second-, or even third-order diffrac-


.F5

-

.1.cai)

cc

1.0 -

0.8-

0.6-

0.4-

0.2 -

0-1.0-

0.8

0.6

0.4

0.2

0.~

1.0

0.8

0.6

0.4

0.2

0

5000 10000

Shift in Vacuum WavenumbersFig. 4. A portion of the dispersed spectra of output from a Raman shifter with a 1.5:1 hydrogen:methane ratio at a total pressure of 375 psi(gauge) for pump wavelengths of 532 nm (top), 355 nm (middle), and 266 nm (bottom). Assignments with the (j, k) scheme for number anddirection of hydrogen (j = 4155 cm-') and methane (k = 2916 cm-') vibrational quanta are shown for the 355-nm data. Additionalassignments involving the hydrogen rotational spacings of 354 and 587 cm-1 and hydrogen v = 0 - 2 transitions, as discussed in Subsection3.B.1, are also shown. Shift is defined as the wave number minus the pump wave number.

tion maxima of the grating of the spectrometer. Ineach of the three spectra, a large number of strongand weak lines are observed at the expected primaryfrequency shifts, i.e., linear combinations of 4155 and2916 cm-'. We have observed the great majority ofall possible lines with an absolute order of 6 or lesswithin the wavelength regions scanned: 27 out of 29with the 532-nm pump, 40 out of 49 with the 355-nmpump, and 22 out of 49 with the 266-nm pump.Several more lines (10 at 532 nm, 10 at 355 nm, and 1at 266 nm) were observed with absolute orders be-tween 7 and 11. Also assigned are a few other linesresulting from the harmonics of the laser fundamen-tal, v = 0 - v = 1 Raman shifts of partiallydeuterated methanes,' 9 and combinations of theseshifts with each other and with the primary frequen-cies. In Fig. 4, in the middle spectrum, which isproduced by 355-nm excitation, (j, k) assignmentsare shown. In addition, several peaks appear to besatellites of (j, k) lines with additional splittingscharacteristic of the pure rotational Raman scatter-ing of hydrogen, namely 354 and 587 cm-' for J =0 - 2 and J = 1 - 3 transitions, respectively.Although from their positions these could also be 0-or S-branch lines of the v = 0 +-> 1 Raman spectrum of

hydrogen, their strengths make the pure rotationalassignment more likely. The Raman cross sectionsfor 0- and S-branch lines of the vibrational transitionare weaker than the strongest Q-branch line Q1(1),20

and the strongest pure rotational lines are strongerthan Q1(1).21 Splittings around 590 cm-' have beenreported in stimulated Raman scattering of hydrogenby other workers622; a broad-spectrum laser sourcebased on enhancing rotational over vibrational SRShas been described recently.23 In the dispersed spec-trum we also observed reasonably strong lines at±8077 cm-', which happens to be the frequency ofthe Q2(1) line of the v = 0 to 2 transition of hydrogen,observed in the Raman spectrum under high sensitiv-ity.24 In our spectrum, the Stokes line at -8077could arise from a one-step Raman transition orsequential 0 -- 1 and 1 * 2 transitions, with thelikelihood of the last transition's being enhanced25 insuch a sample, which has a relatively high v = 1population. But the anti-Stokes line at + 8077, whichis even stronger than the nearby (2, 0) line, could notarise from a true Raman process, because it requires asignificant population of v = 2. It is possible that theStokes line is due at least in part to a true Ramanprocess and that the anti-Stokes is the result of


l | | E | l

parametric four-wave mixing, e.g.,

2k = k 80 77 + k-8077 . (8)

Some lines in the spectrum remain unassigned.

C. Assessment of the Raman-Shifter Excitation Source

The data shown in Fig. 3 indicate that, for mixtureratios of approximately 1:1, typically 14 output lines(including the pump) are available at output pulseenergies above a few microjoules. For example, forR = 1 at a pressure of 225 psi (gauge), output pulseenergies vary from 6 to 1800 pJ for lines varying overa 15 000 cm-' range, with an average spacing of 1183cm-'. Based on our results for the pure gases, onewould expect similar numbers of lines, pulse energies,and frequency ranges with 532- and 266-nm pumpwavelengths (these expectations have been verified inwork reported elsewhere26). The ranges of outputlines for the three pump wavelengths overlap to agreat extent and provide significant coverage of theUV and visible regions of the spectrum.

As noted in Subsections 3.A and 3.B, pressureoptimization of the number of output lines gives quitedifferent results for pure gases and mixtures. Puregases at low pressure give wide but sparse spectralcoverage with high-order Stokes and anti-Stokes lines,whereas mixtures with R = 1 at higher pressure givedenser spectral coverage with low-order and low-displacement combination lines.

For our application-electronic equipment monitor-ing analysis of fluorescent analytes-previous work-ers have used from 16 to 64 discrete excitationwavelengths. Therefore, the performance of the Ra-man shifter appeared to be adequate to our purposes,particularly for the initial laboratory studies thatused a diode-array spectrometer to capture one emis-sion spectrum produced by one excitation line at atime, because the required switching of pump harmon-ics was not objectionable in that case. We achievedgood results in analyzing 2-component mixtures (i.e.,solutions containing 2 fluorescent analytes) using 14different lines and 7-m lengths of optical fiber,9 or 3-and 4-component mixtures using 26 different linesand 0.7-m fibers.26 In each case only the third- andfourth-harmonic pumps were used. Obviously, animportant limitation to the usefulness of a givenoutput line in our application is the attenuation of thefiber, which increases sharply with decreasing wave-length in the UV and increases exponentially withfiber length. Fortunately, the analytical demandson an instrument such as ours would be higher in alaboratory setting (or for an on-site field instrument),where fiber lengths may be quite short and samplesmay be pretreated, e.g., by extraction, to removespecies that might interfere with the fluorescenceanalysis. In true in situ field work, where longerfibers must be used to access the sample, a smallernumber of usable output lines still provides a power-ful capability for screening, if not for absolute identi-fication and quantitation of all fluorescent contami-nants.

The availability of simultaneous output at manywavelengths is an especially important attribute of alaser light source now that CCD's and related detec-tors are making parallel spectroscopy attractive.Few laser systems can produce as wide an array ofoutput lines simultaneously as a laser-Raman shiftercombination. Using a CCD-based detector, we arenow developing a version of our instrument that uses10 excitation lines, all produced with the 266-nmpump. This version results in an instrument withsignificant analytical capability. It has a very rapiddata-acquisition time (- 1 s), which makes it suitablefor monitoring flow systems, and it has no movingparts, which makes it attractive for field work andother demanding operating environments.

This work was supported by the EnvironmentalProtection Agency through a grant to the Center forEnvironmental Management at Tufts University.We thank K. K. Innes and G. Bickel of the StateUniversity of New York at Binghamton for helpfuldiscussions and information.

References1. J. E. Kenny, G. B. Jarvis, W. A. Chudyk, and K. 0. Pohlig,

"Remote laser-induced fluorescence monitoring of ground-water contaminants: prototype field instrument," Anal. In-strum. 16, 423-445 (1987).

2. I. M. Warner, E. R. Davidson, G. D. Christian, "Quantitativeanalyses of multicomponent fluorescence data by the methodsof least squares and nonnegative least sum of errors," Anal.Chem. 49,2155-2159 (1977).

3. J. C. White, "Stimulated Raman scattering," in TunableLasers, Vol. 59 of Topics in Applied Physics, L. F. Mollenauerand J. C. White, eds. (Springer-Verlag, Berlin, 1987), pp.115-207.

4. V. Wilke and W. Schmidt, "Tunable UV radiation by stimu-lated Raman scattering in hydrogen," Appl. Phys. 16, 151-154(1978).

5. V. Wilke and W. Schmidt, "Tunable coherent radiation sourcecovering a spectral range from 185 to 880 nm, " Appl. Phys. 18,177-181 (1979).

6. S. Kawasaki, T. Imasaka, and N. Ishibashi, "Widely tunablelaser system using a single dye solution based on the stimu-lated Raman effect," Opt. Commun. 66, 285-288 (1988).

7. J. A. Duardo, L. J. Nugent, and F. M. Johnson, "Combinationlines in stimulated Raman emission from gas mixtures," J.Chem. Phys. 46, 3585-3591 (1967).

8. T. R. Loree, R. C. Sze, D. L. Barker, and P. B. Scott, "New linesin the UV: SRS of excimer laser wavelengths," IEEE J.Quantum Electron. QE-15, 337-342 (1979).

9. T. A. Taylor, H. Xu, G. B. Jarvis, A. C. Bevilacqua, and J. E.Kenny, "Fluorescence EEM studies of aqueous pollutantsusing stimulated Raman-shifted laser excitation," Anal. In-strum. 21, 141-162 (1993).

10. K. Bennett and R. L. Byer, "Computer-controlled wedged-plate optical variable attenuator," Appl. Opt. 19, 2408-2412(1980).

11. V. Y. Baranov, V. M. Borisov, A. Y. Vinokhodov, Y. B.Kiryukhin, and Y. Y. Stepanov, "Stimulated Raman scatteringof radiation from an electric-discharge pulse-periodic XeCllaser in compressed H2," Sov. J. Quantum Electron. 15,727-729 (1985).

12. W. K. Bischel and G. Black, "Wavelength dependence ofRaman scattering cross sections from 200-600 nm," AIP Conf.Proc. 100, 181-194 (1983).


13. J. J. Ottusch and D. A. Rockwell, "Measurement of Ramangain coefficients of hydrogen, deuterium and methane," IEEEJ. Quantum Electron. 24, 2076-2080 (1988).

14. R. A. Alberty, Physical Chemistry, 6th ed. (Wiley, New York,1983), pp. 12-18.

15. A. Zubov, A. V. Kraiskii, K. A. Prokhorov, M. M. Sushchinskii,and I. K. Shuvalov, "Energy and time characteristics ofstimulated Raman scattering," Sov. Phys. 28, 231-235 (1969).

16. B. V. Zhdanov, L. L. Kulyuk, and S. M. Pershin, "Experimen-tal investigation of mechanisms of parametric generation ofstimulated Raman scattering components," Sov. J. QuantumElectron. 6, 550-553 (1976).

17. A. D. May, J. C. Stryland, and H. L. Welsh, "Raman spectra ofH2 and CH4 at high pressures," J. Chem. Phys. 30, 1099-1100(1959).

18. A. Owyoung, C. W. Patterson, and R. S. McDowell, "CWstimulated Raman gain spectroscopy of the v1 fundamental ofmethane," Chem. Phys. Lett. 59, 156-162 (1978).

19. G. Herzberg, Infrared and Raman Spectra of PolyatomicMolecules, Part 2 of Series on Molecular Spectra and Molecu-lar Structure (Van Nostrand, Princeton, N.J., 1945), pp.239-369.

20. D. E. Jennings, A. Weber, and J. W. Brault, "Raman spectros-

copy of gases with a Fourier transform spectrometer: thespectrum of D2," Appl. Opt. 25, 284-290 (1986).

21. W. R. Fenner, H. A. Hyatt, J. M. Kellam, and S. P. S. Porto,"Raman cross section of some simple gases," J. Opt. Soc. Am.63, 73-77 (1973).

22. T. Imasaka, S. Kawasaki, and N. Ishibashi, "Generation ofmore than 40 laser emission lines from the ultraviolet to thevisible regions by two-color stimulated Raman effect," Appl.Phys. 49,389-392 (1989).

23. S. Yoshikawa, S. Kawasaki, T. Imasaka, and N. Ishibashi,"Equally spaced multifrequency laser emission generated bytwo-color stimulated Rainan effect. Enhancement of rota-tional lines and suppression of vibrational lines," Jpn. J. Appl.Phys. 30, 283-285 (1991).

24. D. K. Veirs and G. M. Rosenblatt, "Raman line positions inmolecular hydrogen, H2, HD, HT, D2, DT, and T2," J. Mol.Spectrosc. 121, 401-419 (1987).

25. G. V. Venkin and G. M. Mikheev, "Stimulated Raman spectros-copy of excited states of the hydrogen molecule," Sov. J.Quantum Electron. 15, 257-259 (1985).

26. J. Kenny, "Monitoring of aqueous pollutants using laser-excited fluorescence array spectroscopy," final rep. assist.agreement CR813481 (Center for Environmental Manage-ment, Tufts University, Medford, Mass., March, 1993).


evaluation of nd:yag-pumped raman shifter as a broad-spectrum light source

Documents