1396
OPTICS LETTERS / Vol. 32, No. 11 / June 1, 2007
Coherent synthesis using carrier-envelope phasecontrolled pulses from a dual-color femtosecond optical parametric oscillator Jinghua Sun,* Barry J. S. Gale, and Derryck T. Reid Ultrafast Optics Group, School of Engineering and Physical Sciences, Heriot-Watt University, Edinburgh EH14 4AS, UK *Corresponding author:
[email protected] Received January 8, 2007; revised March 7, 2007; accepted March 8, 2007; posted March 9, 2007 (Doc. ID 78793); published April 26, 2007 By using a dual-color femtosecond optical parametric oscillator (OPO), a coherent waveform was synthesized from two coresonant near-infrared signal pulses whose center wavelengths had a separation of 100 nm. Immediately after the OPO cavity the pulses had independent carrier-envelope phase-slip frequencies, and synthesis was achieved by shifting these frequencies using an acousto-optic modulator driven by an internally generated difference frequency. Soliton self-frequency shifted pulses from a photonic crystal fiber and a cross-correlation frequency-resolved optical gating (XFROG) measurement were used to analyze the result of the synthesis experiment and revealed that the synthesized waveform was a train of high-contrast 30 fs pulses. © 2007 Optical Society of America OCIS codes: 120.5060, 140.7090, 190.4970, 320.7110.
Coherent synthesis offers a route to subfemtosecond laser pulses but generally requires a spectral bandwidth wider than the emission from a single laser. High-harmonic generation has been used to create phase-coherent synchronized pulses that can be combined to produce isolated pulses with 130 as duration [1], and Raman sideband generation has also been used to synthesize a train of single-cycle optical pulses with 1.6 fs pulse duration [2]. Both techniques require high-energy amplified laser pulses to access the high-order nonlinearities involved; however, coherent waveform synthesis is also possible by controlling the carrier–envelope phases [3–5] and repetition frequencies [6–8] of independent ultrafast laser oscillators, which offers a direct and intuitive way of generating pulses with durations significantly shorter than the parent laser pulses [9]. This kind of coherent synthesis has previously been demonstrated between two independent mode-locked Ti:sapphire lasers [10] and between independent Ti:sapphire and Cr:forsterite lasers [11,12]. With this general concept in mind, synchronously pumped optical parametric oscillators (OPOs) are excellent candidates for coherent synthesis because of their abundant spectral components and the inherent synchronization between the pump pulses and all the parametrically generated pulses. Prior work in this area has indicated routes to controlling the relative carrierenvelope phase-slip (CEPS) frequencies of the pump, signal, and idler pulses in a femtosecond OPO when it was running with an integer ratio between the pump, signal, and idler frequencies [13]. In this Letter, we report direct synthesis from two signal pulses separated by around 100 nm and generated by a dual-color femtosecond optical parametric oscillator [14]. By using direct acousto-optic modulation of the pulses’ CEPS frequencies, and by taking advantage of their intrinsic mutual synchronization, we demonstrate a simple means of coherent synthesis that 0146-9592/07/111396-3/$15.00
avoids the need for complex electronic phase-slip detection and phase-locking loops. In our earlier work [14] we reported a dual-color femtosecond OPO in which the two coresonant signal pulses had different CEPS frequencies, designated here as F1 and F2 for the shorter- and longerwavelength pulses, respectively. An important feature of this OPO was that intracavity mixing processes between the two pulses generated an easily detectable internal beat frequency, Fint = 兩F1 − F2兩. Shifting the frequency of one of the two pulses by their mutual CEPS frequency difference, Fint, is sufficient to create two pulses with identical CEPS frequencies, which can then be combined to create a coherently synthesized waveform. Since the internal beat frequency is automatically generated in the OPO cavity when it is operating in dual-color mode [14], no additional spectral broadening or nonlinear interferometer is needed to obtain the CEPS difference frequency, greatly simplifying the experiment. Figure 1(a) shows the schematic experimental configuration. The dual-color signal spectrum shown in Fig. 1(b) shows some self-phase modulation in the shorter-wavelength pulse. This is slightly different from that reported in [14] because of the higher intracavity power in the present work. Both signal pulses travel together in a single near-infrared beam, and this was directed into a traveling wave acoustooptic modulator (AOM, Gooch & Housgo M111-2JAV1), which was used to shift the CEPS frequencies of signal pulses diffracted by it. We electrically filtered the harmonic of the internal beat frequency detected by an avalanche photodiode at Fint + Frep and frequency-divided this signal to drive the AOM at 共Fint + Frep兲 / 2. In Fig. 1(a) the AOM acts as a spatial separator as well as a frequency shifter for the two coresonant signal pulses because its Bragg angle is related to the optical wavelength. The signal beam propagating forward through the AOM generates the © 2007 Optical Society of America
June 1, 2007 / Vol. 32, No. 11 / OPTICS LETTERS
Fig. 1. (a) Schematic of the experiment. The thin solid (dashed) line represents light diffracted from the AOM at the shorter (longer) wavelength with CEPS frequency F1共F2兲. DM, dichroic mirror; APD, avalanche photodiode. (b) Spectrum of the OPO signal pulses. Table 1. Outputs from the AOM Center Wavelength 共m兲
CEPS
A
1.24 1.33
F1 − Fint / 2 F2 + Fint / 2
0a
B
1.24 1.33
F1 + Fint / 2 F2 − Fint / 2
2Finta
Port
CEPS Difference
1397
periment is straightforward when these pulses have a degree of spectral overlap. In this case an interference experiment is all that is required to confirm that the pulses share the same CEPS frequency. For a moderate degree of spectral overlap, interference fringes can also be observed in a cross-correlation measurement [10] when the cross-correlated pulses are phase coherent; however, when the spectral overlap is insignificant it is impossible to use cross correlation or other interference techniques. This was the situation in our experiment. The remaining method available that would allow us to confirm the success of the frequency-shifting procedure was to beat each frequency-shifted pulse against a common external reference. For this purpose, a 70 cm long photonic crystal fiber (Crystal Fibre NL-2.0-740) was used to generate soliton self-frequency shifted (SSFS) pulses [15] with a tunable center wavelength as long as 1.35 m when pumped at 800 nm. The SSFS spectrum was not broad enough to cover the spectra of the two signal pulses simultaneously, so their beat frequencies were measured separately by varying the launch condition of the photonic crystal fiber to tune the wavelength of the SSFS pulse [Fig. 2(a)]. The SSFS pulses carry the CEPS frequency of the pump, and by beating them against the frequency-shifted signal pulses the pair of outputs from Port A and Port B (see Fig. 1) having the same CEPS frequency was
a
Here, F1 ⬎ F2 was assumed, so Fint = F1 − F2. The actual relationship between them can be detected by beating against a common reference described later.
+1-order Bragg diffracted beam with a positive 共Fint + Frep兲 / 2 frequency shift, and that propagating backwards, after a reflective mirror, experiences a negative frequency shift. The beams diffracted by the AOM were combined by using a dichroic beam splitter with high reflectivity at the shorter signal wavelength and high transmission at the longer signal wavelength. Depending on the choice of beam splitter port (Fig. 1, A or B) we were able to combine the short- and long-wavelength pulses with a CEPS frequency difference of either zero or 2Fint. The combinations of CEPS frequency shifts of each pulse at ports A and B are listed in Table 1. Although in this scheme the absolute frequency shift applied to each pulse is ±共Fint + Frep兲 / 2, the effective CEPS shift is the applied shift moduloFrep which is indistinguishable from a shift of ±Fint / 2. As reported in [14], the second-harmonic of the signal pulses (red) and the sum-frequency mixing between the pump and the idler pulses (yellow) both carry a modulation at the same frequency as the internal beat frequency Fint of the signal pulses. The AOM driving signal can be extracted by detecting any of the visible outputs leaking through any cavity mirror without disturbing the signal outputs, and in practice we used the red second-harmonic output [Fig. 1(a)]. The measured jitter of Fint was less than 500 kHz. Confirming the successful preparation of phasecoherent parent pulses in a coherent synthesis ex-
Fig. 2. (a) Spectrum of the dual-color signal output from the OPO (solid line) and the overlap with the tunable SSFS pulse from the PCF (dashed lines). Before compensation, the CEPS frequencies of the shorter- and longerwavelength signal pulses are F1 and F2, respectively. (b) RF spectra of beat frequencies obtained by beating the signal pulses leaving Port A with SSFS pulses tuned to the same wavelengths as the signal pulses. Upper trace, longwavelength signal pulse beat with a long-wavelength SSFS pulse; lower trace, shorter wavelengths. (c) Similar results for the signal pulses leaving Port B. Comparison with (b) confirms that the difference in the CEPS frequencies of the two coresonant signal pulses is only compensated for those pulses leaving Port A. In (b) and (c), for simplification, the CEPS frequency of the SSFS pulse in the beat frequencies is ignored because the absolute values of F1 and F2 are unimportant.
1398
OPTICS LETTERS / Vol. 32, No. 11 / June 1, 2007
identified. In the experiment, the internal beat was 27.5 MHz, and the RF measurements indicated that the frequency-shifted pulses with the same CEPS frequencies were coupled from Port A [Fig. 2(b)]. Consequently, those pulses leaving Port B showed a CEPS frequency difference of 2Fint or 55 MHz [Fig. 2(c)]. The CEPS-controlled pulses prepared in this way are suitable for coherent synthesis. To understand the result of the synthesis experiment, an XFROG measurement [16] was applied to analyze the phasecoherent pulses from port A, where the total power was about 18 mW. By measuring their sumfrequency mixing spectrum after a BBO crystal, along with their individual spectra, the XFROG retrieval result revealed that the long-wavelength pulse was a soliton-shaped output [nearly sech共t兲 amplitude profile, Fig. 3(a)] and the short-wavelength pulse was chirped and probably dispersive in nature [Fig. 3(b)]. These results agree with the analysis of [14] where we showed that the OPO was working in two modes: the longer-wavelength pulse was resonant in a negative group-delay dispersion regime and the shorter-wavelength pulse in a positive groupdelay dispersion regime. The XFROG data show that by combining the appropriate frequency-shifted OPO pulses with zero delay and an approximately equal power balance we can obtain a train of 30 fs pulses with high contrast between them [Fig. 3(c)]. This condition readily existed at port A of the dichroic mirror (see Fig. 1). By turning the AOM, and so changing the incident angle made by the signal pulses on the acoustic grating, it is easy to balance the powers of the two pulses since the Bragg diffractive efficiencies at a particular incident angle are different for each wavelength. Temporal overlap is achieved trivially by adjusting the delay line in the system. In summary, by shifting the CEPS frequencies of two synchronized pulses from a dual-color OPO we have coherently synthesized an ultrafast waveform that is indistinguishable from a single output and
Fig. 3. XFROG measurement results. (a), (b) Temporal intensities and phases of the short wavelength and long wavelength, respectively. (c) Temporal intensity of the synthesized pulse.
which an XFROG measurement reveals to be a train of 30 fs pulses. Our experiment highlights the challenge of confirming that the synthesized pulse is indeed phase coherent across its entire bandwidth. Using highly nonlinear fiber, it might be possible to use self-phase modulation to spectrally broaden the signal pulses sufficiently to observe a strong beat signal between the pulses whose CEPS frequencies were separated by 2Fint. Comparing this signal with the original internal beat frequency could provide a way of inferring the quality of the synthesis result. We anticipate that the general approach of employing a common resonant cavity to create the parent pulses for a coherent synthesis experiment could be extended to carry out synthesis across a much wider bandwidth than the 100 nm demonstrated here. For example, an intracavity Ti:sapphire-pumped OPO would achieve a high degree of common mode rejection of environmental noise sources and could be used to prepare trains of subfemtosecond pulses by choosing an appropriate integer frequency ratio between the pump and signal pulses. The authors gratefully acknowledge financial support for this project from the UK Engineering and Physical Sciences Research Council and from Coherent Inc. References 1. G. Sansone, E. Benedetti, F. Calegari, C. Vozzi, L. Avaldi, R. Flammin, L. Poletto, P. Villoresi, C. Altucci, R. Velotta, S. Stagira, S. De Silvestri, and M. Nisoli, Science 314, 443 (2006). 2. M. Y. Shverdin, D. R. Walker, D. D. Yavuz, G. Y. Yin, and S. E. Harris, Phys. Rev. Lett. 94, 033904 (2005). 3. H. R. Telle, G. Steinmeyer, A. E. Dunlop, J. Stenger, D. H. Sutter, and U. Keller, Appl. Phys. B: Photophys. Laser Chem. 69, 327 (1999). 4. J. Reichert, R. Holzwarth, Th. Udem, and T. W. Hänsch, Opt. Commun. 172, 59 (1999). 5. S. T. Cundiff and J. Ye, Rev. Mod. Phys. 75, 325 (2003). 6. R. K. Shelton, S. M. Foreman, L.-S. Ma, J. L. Hall, H. C. Kapteyn, M. M. Murnane, M. Notcutt, and J. Ye, Opt. Lett. 27, 312 (2002). 7. Z. Wei, Y. Kobayashi, and K. Torizuka, Appl. Phys. B: Photophys. Laser Chem. 74, SI71 (2002). 8. D. Yoshitomi, Y. Kobayashi, M. Kakehata, H. Takada, and K. Torizuka, Opt. Express 14, 6359 (2006). 9. T. W. Hänsch, Opt. Commun. 80, 71 (1990). 10. R. K. Shelton, L.-S. Ma, H. C. Kapteyn, M. M. Murnane, J. L. Hall, and J. Ye, Science 293, 1286 (2001). 11. A. Bartels, N. R. Newbury, I. Thomann, L. Hollberg, and S. A. Diddams, Opt. Lett. 29, 403 (2004). 12. Y. Kobayashi, D. Yoshitomi, M. Kakehata, H. Takada, and K. Torizukaet, Opt. Lett. 30, 2496 (2005). 13. Y. Kobayashi, H. Takada, M. Kakehata, and K. Torizuka, Opt. Lett. 28, 1377 (2003). 14. J. Sun, B. G. J. Gale, and D. T. Reid, Opt. Lett. 31, 2021 (2006). 15. D. T. Reid, I. G. Cormack, W. J. Wadsworth, J. C. Knight, and P. St. J. Russel, J. Mol. Spectrosc. 49, 757 (2002). 16. D. T. Reid, P. Loza-Alvarez, C. T. A. Brown, T. Beddard, and W. Sibbett, Opt. Lett. 25, 1478 (2000).
