250 OPTICS LETTERS / Vol. 24, No. 4 / February 15, 1999

Modulating and scanning the mode-lock frequency of an800-MHz femtosecond Ti:sapphire laser

J. R. Demers and Frank C. DeLucia

Department of Physics, Ohio State University, 174 West 18th Avenue, Columbus, Ohio 43210

Received December 7, 1998

An 800-MHz self-mode-locked femtosecond Ti:sapphire laser has been developed whose mode-lock frequency canbe robustly scanned and modulated. The laser is based on the three-element design of Ramaswamy-Paye andFujimoto [Opt. Lett. 19, 1756 (1994)]. By translation and modulation of the position of the prismatic outputcoupler, the mode-lock frequency can be reliably scanned more than 1% and modulated at 80 Hz with a deviationof 2.5 kHz without interrupting the mode lock, changing the pulse length, or inducing signif icant amplitudemodulation. An application in tunable high-resolution terahertz spectroscopy is also demonstrated. 1999Optical Society of America

OCIS codes: 140.3410, 140.3590, 220.4880, 320.5550.

Many different systems utilizing Kerr-induced modelocking have been developed since its discovery in1991.1 Like the original system, most of these designsemploy a titanium-doped sapphire (Ti:sapphire) crys-tal for gain and require a pair of prisms for dispersioncompensation. Because of the prism pairs, the cavitylength is fairly long s,1.5 md, and the laser repetitionrate is ,100 MHz. For some applications the funda-mental properties of the overall system depend on thetiming of the optical pulse sequence and the modula-tion of this timing. One application that was recentlydescribed2 is the production of terahertz (THz) radia-tion for spectroscopy by demodulation of the pulsesfrom an 80-MHz femtosecond Ti:sapphire laser in alow-temperature-grown GaAs photoconductive switch.Because system sensitivity is dependent on the den-sity of the demodulated spectrum, a simple meansfor improving system performance is to use a mode-locked laser with a higher repetition rate. In thisLetter we describe modifications to a three-element1-GHz 105-fs mode-locked laser recently developed byRamaswamy-Paye and Fujimoto3,4 that allow the trans-lation and modulation of the laser’s prismatic out-put coupler (POC) necessary for THz spectroscopy, aswell as other electro-optic applications. It is signifi-cant that the resulting system is unusually robustand that no particularly critical alignment criteria areintroduced.

An optical pulse train can be converted into itsfrequency components with a photodiode (demodu-lator) and a spectrum analyzer. This demodulationreveals a comblike structure in frequency that isintimately dependent on the timing characteristicsof the optical pulse train. If T is the time inter-val between pulses (1yrepetition rate) and t is thelength of each pulse, then in frequency space theinterval between successive components (harmonic)is f ­ 1yT and the components extend until a roll-offoccurs at ,1yt. In most other THz experimentsit is the extremely short pulse characteristics offemtosecond Ti:sapphire lasers that are sought, andlittle concern is given to the frequency characteristicsof the optical pulse train.5 – 8 In our experiments,however, a single frequency component from demodu-

0146-9592/99/040250-03$15.00/0

lation is used to interact with a Doppler-broadenedrotational transition typical of high-resolution THzspectroscopy. For this application the characteristicsof the optical pulse train are paramount.

For frequency modulation (FM) spectroscopy it isnecessary to modulate the frequency of the radiation toat least the width of a Doppler-limited rotational tran-sition (,3 parts in 106 of the transition frequency) andto be able to scan sufficiently far to provide continuousspectral coverage. Since changes to the base repeti-tion rate of the laser scale as the harmonic number, atranslation of the cavity length of ,3 parts in 106 pro-duces a change in frequency that is comparable withthe molecular linewidth regardless of which harmonicis employed for spectroscopy. In addition, providingcontinuous spectral coverage above 80 GHz (100th har-monic) requires a translation of 1%. A Fluke PM6666counter with a 0.2-s gate provides better than requiredfrequency measurement. Effective FM spectroscopyalso requires that the frequency modulation not inducesignif icant amplitude modulation (AM).

Figure 1 shows the basic elements of the system.The pulse train from either the original 82-MHz or

Fig. 1. Entire submillimeter spectroscopy experiment,with both the original 82-MHz and the new 800-MHzmode-locked Ti:sapphire lasers. The low-temperature-grown (LTG) GaAs PCS with antenna, the collimatinglenses, the grating, the 1-m-long sample cell, and a liquid3He bolometer are shown. The biasing circuit for the PCSis not shown. PZT’s, piezo tubes.

1999 Optical Society of America

February 15, 1999 / Vol. 24, No. 4 / OPTICS LETTERS 251

the new 800-MHz femtosecond laser is focused ontoa photoconductive switch (PCS). The PCS is inte-grated with a biased microwave antenna that hasbeen sputtered onto the surface of a wafer of low-temperature-grown GaAs, which has a response timeof ,250 fs, making it capable of producing radiation to,2 THz.9 At this point the propagating radiation con-tains all the radiation generated by the PCS. In ourapplication overall system sensitivity is sacrificed inthe limit of too many comb elements, because the noiseassociated with all the components falls simultaneouslyon the detection element, whereas any molecular ab-sorption occurs in only a single component. Thereforea grating is employed to reduce the number of com-ponents passed through the sample and to increasethe signal-to-noise ratio. For a grating resolution of150 GHz the original 80-MHz system allows ,2000harmonics to pass through the sample onto the detec-tion element, whereas the new 800-MHz system allowsonly ,200, decreasing the noise proportionately.

Figure 2 illustrates the modified 1-GHz laser, whichwas pumped by 3.8 W of power from a Spectra-Physics2060 argon-ion laser and produced from 200 to 300 mWof power with pulse lengths of 200 to 300 fs. Self-modelocking was often observed but occasionally required aslight mechanical perturbation to the intracavity lens.A New Focus picomotor was employed for translatingthe POC stage (i.e., coarsely tuning the repetition rate),while modulation required that an extremely low-massand stable mount be designed for the POC. This re-quirement was complicated by the 40± angle of theprism face to the optical axis and by the equally off-axis output beam. A pistonlike adapter was designedthat held the POC at a constant angle to, but allowedmovement along, the optical axis. An aluminum pis-ton attached to a 2.54-cm-diameter 2.54-cm-long piezotube slid inside a polished stainless-steel cylinder.

Of primary interest was how well the POC couldbe translated and modulated without breaking ordestabilizing mode lock. From a stable mode lock at798.275 MHz the POC was translated without observ-able instability until it hopped out of mode lock at808.1 MHz. This result corresponds to translation ofthe POC by approximately 1.2% of the cavity length,or ,2 mm. After loss, mode lock could be regainedby slight adjustment of the POC, which suggests thatthe reason mode lock was lost was slight misalign-ment during translation. Next, a 300-V 80-Hz modu-lation was applied to the piezo tube. A 500-V dc offsetwas required for prevention of distortion of the 80-Hzsine modulation. Figure 3(a) displays the fundamen-tal at 800 MHz and Fig. 3(b) the broadening inducedby the applied modulation. These measurements wereobtained by illumination of a 1-ns photodiode andrecording of the results on a Hewlett-Packard 8555Aspectrum analyzer. As can be seen from Fig. 3(b),the modulation resulted in a frequency deviation of2.5 kHz at fundamental, or a 1-MHz frequency devia-tion at 300 GHz. This deviation corresponds to theabove-mentioned deviation of ,3 parts in 106 requiredfor FM spectroscopy of rotational transitions.

It was hoped that this FM could be obtained witha minimum of induced AM. To measure the AM it

was necessary to compare the overall intensity of theTi:sapphire laser with the intensity of the AM inducedby modulation of the POC. First, the intensity of thelaser beam was reduced with a 2.0 neutral-densityfilter (1y100th) for prevention of photodiode satura-tion. As before, the modulation voltage was appliedto the piezo tube. Measuring the peak-to-peak volt-age on the chopped beam (18.9 V) and comparing itwith the unchopped signal with a clearly induced 80-Hzripple (0.55 V) revealed an AM of 3%. Next, the pulselength was measured with a modif ied Spectra-Physics409 autocorrelator, and the bandwidth was measuredwith a calibrated monochromator and a photodiodearray. This system currently produces 260-fs pulseswith FWHM Gaussian bandwidth of 5.5 nm. The

Fig. 2. Three-element Ti:sapphire laser with the modi-fied POC.

Fig. 3. (a) First harmonic of the unmodulated 800-MHzlaser. (b) 2.5-kHz deviation on the fundamental of the800-MHz laser, caused by a 300-V 80-Hz modulation with a500-V dc offset.

252 OPTICS LETTERS / Vol. 24, No. 4 / February 15, 1999

Fig. 4. Autocorrelation of the 800-MHz laser during thestage denoted in the text. Each trace has been separatedin amplitude to make the traces resolvable: (a) unmodu-lated; (b) 300 V, 80 Hz, 500 V dc; (c) 300 V, 80 Hz, 500-Vdc, with 1-mm translation.

Fig. 5. The submillimeter spectrometer illustrated inFig. 1 was employed to measure four rotational transitionsin methyl f luoride above 800 GHz. The actual transitionsare labeled in the figure.

time–bandwidth product for these pulses is ,0.75, in-dicating a slight chirp. The initial autocorrelation andthe measurements made during modulation and after a1-mm translation are displayed in Fig. 4 and have beenseparated in amplitude so that all three can be clearlyresolved. There is no observable change in the 260-fspulse length observed in the autocorrelation.

As a final test this new 800-MHz laser system wasemployed to perform Doppler-limited submillimeterspectroscopy. As illustrated in Fig. 1, radiation fromthe low-temperature-grown GaAs PCS is collimatedwith a Tef lon lens, strikes a grating, is passed througha 1-m sample cell containing 4 mTorr of methyl f luo-ride, and is focused onto a liquid 3He silicon bolometer.The laser cavity is tuned so that a harmonic of the repe-tition frequency is near a transition of interest, andthen the POC is modulated and translated. Signal re-covery is achieved with a Stanford Research SystemsSR510 lock-in amplifier. Figure 5 shows the recorded

165-MHz scan achieved by translation of the POC by,40 mm. Clearly evident are the K ­ 0, 1, 2, 3 com-ponents of the J ­ 15 ! 16 transition at ,816.1 GHz(1017th harmonic). The signal-to-noise ratio is ,10but can be improved by a factor of 40 with bettersubmillimeter beam collimation. Although the 10-cmgrating has a theoretical bandwidth of ,4 GHz, withthe current optics a bandwidth of more than 150 GHzis present. This excess bandwidth decreases sensitiv-ity and also results in the unintentional interaction ofother harmonics with rotational transitions. For ex-ample, Fig. 5 also shows a line at 816.074 GHz thatcorresponds to the K ­ 3 component of the 18 ! 19transition interacting with the 1207th harmonic. Thistransition actually occurs at 968.570 GHz.

In summary, we have described a FM technique fora 1-GHz mode-locked laser and measured its modu-lation characteristics. We have demonstrated me-chanical scanning over a range that allows continuouscoverage above 80 GHz without loss of mode lock orobservable variation in pulse width. We have illus-trated modulation at frequencies and depths necessaryfor FM microwave spectroscopy with less than 3%induced AM. We have also shown the applicationof this modulated laser in a particularly demandingapplication, high-resolution THz spectroscopy.

We thank the U.S. Army Research Office for supportof this work and B. E. Bouma for advice during laserconstruction. We also thank Linn Van Woerkom andTom Goyette for assistance and advice. J. R. Demers’se-mail address is jrdemers@pacif ic.mps.ohio-state.edu.

References

1. D. E. Spence, P. N. Kean, and W. Sibbett, Opt. Lett. 16,42 (1991).

2. T. M. Goyette, W. Guo, F. C. De Lucia, J. C. Swartz,H. O. Everitt, B. D. Guenther, and E. R. Brown, Appl.Phys. Lett. 67, 3810 (1995).

3. B. E. Bouma and J. G. Fujimoto, Opt. Lett. 21, 134(1996).

4. M. Ramaswamy-Paye and J. G. Fujimoto, Opt. Lett. 19,1756 (1994).

5. M. T. Asaki, C.-P. Huang, D. Garvey, J. Zhou,H. C. Kapteyn, and M. M. Murnane, Opt. Lett. 18, 977(1993).

6. I. P. Christov, J. Zhou, J. Peatross, A. Rundquist, M. M.Murnane, and H. C. Kapteyn, Phys. Rev. Lett. 77, 1743(1996).

7. R. R. Jones, D. You, and P. H. Bucksbaum, Phys. Rev.Lett. 70, 1236 (1993).

8. C. R. Jones, H. Kosai, J. M. Dutta, M. J. Peters, W. Guo,and F. C. De Lucia, Appl. Phys. Lett. 67, 1483 (1995).

9. E. R. Brown, K. A. McIntosh, K. B. Nichols, and C. L.Dennis, Appl. Phys. Lett. 66, 285 (1995).