1606 OPTICS LETTERS / Vol. 22, No. 21 / November 1, 1997

Generation of tunable 187.9–196-nm radiation in b-Ba2BO4

G. C. Bhar, U. Chatterjee, A. M. Rudra, and P. Kumbhakar

Laser Laboratory, Department of Physics, Burdwan University, Burdwan 713104, India

R. K. Route and R. S. Feigelson

Center for Materials Research, Stanford University, Stanford, California 94305

Received June 23, 1997

Tunable 187.9–196-nm vacuum-ultraviolet radiation was generated at room temperature in a beta bariumborate crystal by sum-frequency mixing of Nd:YAG laser radiation and the second harmonic of a dye laserpumped by the second harmonic of the same Nd:YAG laser. By use of the advantageous noncollinear phase-matching configuration, a peak power of 1.3 kW was obtained at 194 nm with input power densities as low as79 MWycm2 for 1064-nm and 0.65 MWycm2 for 237.3-nm radiation. 1997 Optical Society of America

By use of sum-frequency mixing (SFM) of excimer laserand tunable dye laser radiation, UV wavelengths downto 189 nm have been generated in beta barium borate(BBO),1 and to date this is the lowest wavelengthreported to have been generated in the UV region inthis crystal. By SFM of a commonly available Nd:YAGlaser and second-harmonic generation (SHG) of dyelaser radiation, pumped by the second harmonic ofthe same Nd:YAG laser, we report here generation oftunable deep UV radiation down to 187.9 nm in BBOin a noncollinear phase-matched configuration.

Because of the higher frequency (and hence energy)of deep UV radiation, the damage problem for theoptics (as well as dielectric coatings) involved is moresevere in the UV region than in the visible or theIR region. Thus it is rather diff icult (and hencecostly) to manufacture suitable interference filtersthat can simultaneously transmit better than 60%of the generated deep UV radiation and almost cutoff the input parent beams, one of them being inthe UV region itself. In this respect the noncollinearscheme provides the major advantage of automaticseparation of the generated output from the inputparent beams. Moreover, for a noncollinear anglebetween the input beams that is not too large (butis more than suff icient to separate completely theinteracting beams), there is no appreciable decrease inthe conversion efficiency of the SFM process, indicatingthat the effective interaction length of the crystalremains the same as in the collinear case.2

The experimental setup is shown in the insetof Fig. 1. For a 1064-nm sl1d radiation source, anelectro-optically Q-switched Nd:YAG laser (Spectra-Physics, DCR-11) was used. The dye laser (Spectra-Physics, PDL-2) used in the experiment was pumped bythe second harmonic of the Nd:YAG laser, generated ina temperature-tuned DKDP crystal. Coumarin-480dye was used to generate tunable radiation from 456 to480 nm. For type I SFM both the 1064-nm radiationand the SHG of the dye laser radiation should be verti-cally polarized. So a 90± polarization rotator was usedto change the vertical polarization state of the funda-mental dye beam to horizontal. This dye beam wasthen frequency doubled in a BBO crystal (B1) that was8.3 mm 3 7.3 mm 3 6.4 mm s length 3 height 3 widthd

0146-9592/97/211606-03$10.00/0

type I, u ­ 60± cut (NEC, Japan), and rotated in thevertical plane. The generated SHG UV radiation sl2dthus became vertically polarized and was separatedfrom the fundamental by a dichroic mirror (M1) having30% transmission for l2 and .97% ref lectance for400–500 nm. A Glan prism was placed in the beampath of the 1064-nm radiation to reject its horizontallypolarized component. Thus when the separatedSHG UV and the residual 1064-nm radiation werecombined in a second 5.0 mm 3 6.0 mm 3 5.0 mmtype I, u ­ 65±-cut BBO crystal (B2; Stanford Univer-sity) for SFM, both their polarization states becamevertical as required. Therefore, B2 was rotated inthe horizontal plane, which is more convenient for thenoncollinear phase-matching configuration. The non-collinear angle between the input beams was variedby rotation of mirrors M2 and M3. The variations ofthe absorption coeff icients a [in inverse centimeters

Fig. 1. Absorption characteristics of the two BBO crys-tals used in the experiment. Curve 1, crystal B1; curve 2,crystal B2. The inset shows a schematic of the experimen-tal setup: 1, Nd:YAG beam; 2, dye laser beam; M1, dichroicmirror (with 30% transmission for l2 and .97% ref lectancefor 400–500 nm); M2, M3, mirrors with ref lectance of 100%and 75% at 1064 nm, respectively; BD, beam dumper; P, ex-tradense f lint glass prism; R, 90± rotator for the dye laser.

1997 Optical Society of America

November 1, 1997 / Vol. 22, No. 21 / OPTICS LETTERS 1607

scm21d] for B1 and B2 are shown in Fig. 1, a beingcalculated from the measured transmission data fromUV–visible–near-IR (M/S Hitachi, U-3400) and IR (M/S Shimadzu, IR-470) spectrophotometers. Both of thecrystals were used without any antiref lection coating.A long focal-length lens s f ­ 120 cmd was used to focusthe fundamental dye radiation such that its diameteron both B1 and B2 remained 2 mm. The dye SHGbeam diameter on B2 thus also became ,2 mm. The1064-nm beam was focused and its beam diameter wasadjusted to be ,3 mm at the entrance face of B2, andthe focal points of both the beams were kept well be-yond the crystal. This resulted in a small variation ofthe beam diameter throughout the length of B2. Sincethe energy of the l2 radiation is quite small s,, 1 mJd,ensuring eff icient utilization of the total energy, thediameter of the 1064-nm beam is kept larger while theexternal noncollinear angle sad remains fixed at ,2.5±

throughout the experiment. For the crystal size usedand the input beam diameters, this value of a ensuresno reduction in the effective interaction length andautomatic separation of the generated UV radiationdown to 187.9 nm, which can be visually detected quitewell on either white bond paper or a silicate glassplate. Since the available energy for l2 was quitesmall s,250 mJ d, we used mirror M1 only to ascertainproper overlap of l1 and l2 in B2 and hence to findthe values of phase-matching angle sud for SFM. Formeasurement of conversion efficiency, M1 was takenout to allow both the fundamental dye beam and itsSHG to fall on B2. Since the laser damage thresholdof BBO is very high s. 2 GWycm2d, as expected, thisdid not create any problem as far as crystal damageis concerned, but it ensured the availability of totalenergy of the l2 beam for generation of l3. Moreover,since dye SHG is a collinear process, whereas genera-tion of l3 is noncollinear, l3 is automatically separatedfrom the dye radiation as well, and hence there is nodifficulty in the final adjustment of overlap between l1and l2 in B2 to get maximum conversion at l3.

The phase-matching tuning characteristics for non-collinear SFM are shown in Fig. 2 for a ­ 2.5±; thevariation of u with a for generation of 194-nm radi-ation is shown in the inset. We define u as the inter-nal phase-matching angle that the sum-frequency wavevector makes with the crystal optic axis. We used Sell-meier coefficients of Ref. 3 to calculate the theoreticalvalues of u, which are found to be in reasonably goodagreement with experimental ones down to 187.9 nm.Figure 3 shows variations of energy, measured by aScientech power–energy meter, during SHG and SFMfor different interacting beams throughout their tuningranges. The SHG energies are estimated with4

E2 ­ ElTlT2 tanh2f52.2Leff2deff

2ElTlF y

snold3l2tpr2

lg1/2, (1a)where

F ­ hexps22a2Ld f1 2 2 exps2ad 1 exps22adgya2j ,

a ­ s2al 2 a2dL . (1b)In Eqs. (1a) and (1b) El and E2 are the energy in joulesof the fundamental dye beam sld and its SHG sl2d. Tl

and T2 are the f irst surface transmission of B1 for l

and l2, and al and a2 are the respective absorptioncoefficients (in cm21). L is the actual crystal length;Leff is the effective interaction length in centimeters;Leff , 2rlysinsr2d, where r2 is the walk-off angle of theextraordinarily sed polarized second-harmonic beam(in our SHG system Leff ­ L); deff is the effectivenonlinear coefficient in electrostatic units; no

l is theordinary sod refractive index of B1; and rl is thebeam radius in centimeters for l. t is the laser pulsewidth, which is ,10 ns for all the interacting beams.

Fig. 2. Noncollinear phase-matching characteristics oftunable SFM in BBO for a ­ 2.5±. The inset shows thevariation of phase-matching angle u with a for generationof 194 nm by SFM of 1064 and 237.3 nm. Solid curve,theoretical prediction; circles, the experimental values.

Fig. 3. Energy variation at different wavelengths duringSHG and SFM in two BBO crystals. (a) Variation offundamental dye energy measured in front of B1, (b)variation of energy for l2 measured in front of B2, (c)energy of the generated UV radiation.

1608 OPTICS LETTERS / Vol. 22, No. 21 / November 1, 1997

A typical conversion eff iciency E2yEl of ,12% wasobtained at the peak region of the dye used for an inputenergy of 2 mJ, which amounts to a power densityof 6.4 MWycm2. To estimate the energy E3 of l3 weused4

E3 ­ E2T2T3sl2r32yl3r2

2dtanh2h52.2Leff2deff

2E1T1Fy

fno1no

2ne3sudl2l3tpr1

2gj1/2, (2a)where

F ­ hexps22a3Ld f1 2 2 exps2ad 1 exps22adgya2j ,

a ­ sa1 1 a2 2 a3dL , (2b)

Leff , 2r2ysinsa 1 r3d . (2c)In Eq. (2a), Ei are energy in joules, ni are refrac-tive indices, Ti are first surface transmission, aiare absorption coefficients of B2 in inverse centime-ters, and ri are beam radii for interacting li in cen-timeters. r3 is the walk-off angle of e-polarized l3.In Eq. (2b), to determine Leff we chose the beamradius of l2 because it is smaller between the twoinput beams. The conversion efficiency for SFM isdefined as h ­ E3y

psE1E2d, and we obtained a maxi-

mum value of 0.4% at 194 nm sE3 ­ 13 mJ d withinput energy sE1d at 56 mJ at 1064 nm and 200 mJ sE2dat 237.3 nm. These values correspond to peak powerdensities as low as 79 MWycm2 sP1d for 1064 nm and0.65 MWycm2 sP2d for 237.3 nm. With the same valueof P1, the minimum energy of ,1.2 mJ was obtained at190 nm when P2 ­ 0.3 MWycm2 at 231 nm.

In our system the fundamental dye wavelength canbe changed through a step 0.025-nm. So we got rea-sonably good agreement between the experimental(0.07-nm) and the theoretical (0.05-nm) values of thespectral bandwidth for SHG sDlSHGd for B1. But thecalculated value of the spectral bandwidth for SFMsDlSFMd for B2 is as short as 0.041 nm, and sinceDlSHG . DlSFM, the experimental value of DlSFM waslimited by that of DlSHG, because one has to scan thephase-matching peak in B2 to generate l3, with ra-diation with Dl , DlSFM to measure experimentallyDlSFM. Of course, by placing proper etalons in thebeam path of l2 within B1 and B2, one can experimen-tally measure DlSFM.

In conclusion, we have presented an advantageousscheme for SFM in BBO to generate a tunable deepUV source in the 187.9–196-nm region. Particularlywhen the available energy level is very low (as in our

experiment), a noncollinear configuration seems to bethe best choice. Although BBO is phase matchabledown to 187 nm and shorter (2–4±) in this conf igu-ration, without a very large change in u (and hencein deff ), and B2 is also suitable to accommodate thephase matching, tunability below 187.9 nm becomeslimited by the exponential increase in crystal absorp-tion. However, by cooling the crystal, as the lowestband edge shifts deeper into the UV,5 cooled BBO crys-tal is expected to exhibit some improvement in deepUV tunability yet. Taking the value of the tempera-ture coefficient of the band edge,6 we estimate that thelimiting wavelength of 187.9 nm would shift to 184 nmfor crystal cooling to 2200 ±C. However, only a de-tailed investigation of the effect of inherent impuri-ties incorporated during crystal growth can reveal thesalient features of crystal residual absorption near theband edge. When one considers the very high dam-age threshold of BBO, it seems that an increase ofinput power densities (particularly that of l2 in thisexperiment, which we could not demonstrate becauseof the low pump energy of our YAG system) shouldyield a reasonable amount of energy output (down to187.9 nm) from BBO, which will at least be suff icientfor spectroscopic studies.

G. C. Bhar, U. Chatterjee, A. M. Rudra, and P.Kumbhakar are grateful to Y. Kuwano and H. Kouta ofNEC, Japan, for the BBO crystal (B1) used for SHG ofdye laser radiation in the experiment and thank oneof the reviewers for critical comments. Partial f inan-cial assistance from the authority of the National LaserProgramme, Government of India, is also gratefullyacknowledged.

References

1. W. Muckenheim, P. Lokai, B. Burghardt, and D.Basting, Appl. Phys. B 45, 259 (1988).

2. G. C. Bhar and U. Chatterjee, Jpn. J. Appl. Phys. 29,1103 (1990).

3. D. Eimerl, L. Davis, S. Velsko, E. K. Graham, and A.Zalkin, J. Appl. Phys. 62, 1968 (1987).

4. F. Zernike, in Methods of Experimental Physics, C. L.Tang, ed. (Academic, New York, 1979), Vol. 15, Part Bpp. 143–183.

5. V. Kisand, R. Kink, M. Kink, J. Maksimov, M. Kirm, andI. Martinson, Phys. Scr. 54, 542 (1996).

6. G. Zhang, Y. Yang, and C. Zhang, Appl. Phys. Lett. 53,1019 (1988).