Welcome to the Warren Group

Principles of Pulse Shaping

At the beginning of the 1980s we developed multiple pulse laser sequences with full inter-pulse phase control, and applied these sequences to study gas dynamics; in the mid-1980s we demonstrated that we could produce complex phase and amplitude modulation as well. However, these methods were limited to roughly 5 ns resolution and <1 W peak power, and thus were far too slow for most optical transitions. In 1986 we developed the first method for altering the shape of subpicosecond laser pulses under computer control, using high-speed electro-optic modulators and picosecond voltage waveform generators. We later used variable length rectangular laser pulses (600-1300 fs)

Figure 1. Acousto-optic pulse shaping permits >1000-point resolution for an arbitrarily phase/amplitude modulated pulse.

Figure 2. Changing the RF frequency changes the pulse delay. Thus N sine waves give N pulses with complete phase control.

to probe hot carrier relaxation by photoconductive sampling. We also produced phase and amplitude modulated laser pulses at λ=1.5 µm with 10 ps resolution using a LiNbO3 intensity modulator and a separate phase modulator. Unfortunately, these pulse-shaping capabilities never became widely available in the ultrafast laser community, largely because of technical complexity. An alternative approach is to use frequency selective optical elements (diffraction gratings or prisms) to spatially disperse the large bandwidths associated with such short pulses, then to use spatial modulators to alter the spectrum. Weiner and Heritage made this programmable for the first time in 1989 with liquid crystal arrays.This approach is much simpler than the electro-optic method, but it suffers from a variety of technological limitations (slow update, gaps between pixels, large phase shifts for complex waveforms).

Figure 3. Amplified shaped pulse (200 µJ, 1kHz).

We have recently developed a method for modulating spatially dispersed fs laser pulses using microsecond RF pulses in an acousto-optic modulator (AOM) (Figure 1).An AOM diffracts light when an acoustic wave is present; typical conversion efficiencies are about 50% for 1 Watt of RF power.We apply a shaped RF pulse to the AOM's transducer, which creates an acoustic wave traveling through the crystal; we then save the beam diffracted by the modulator. The transit time of a fs pulse through the crystal is too short for significant acoustic propagation, so the acoustic wave looks like a modulated diffraction grating.

In comparison with prior methods, this approach has the advantages of faster update times, no pixel gaps, high isolation and simple calibration. It is also better suited for use with existing commercial components; in fact, we have patented this approach and licensed this technology to a commercial laser manufacturer, who will incorporate the technology into their amplified laser systems. Figures 2 and 3 show a few simple examples.Generation of amplified pulses with high fidelity is particularly important for nonlinear applications. Figure 3 shows we can shape prior to amplification and retain good fidelity. We have already demonstrated the equivalent of >1000 independently modulated frequency components, far better than any other approach is likely to demonstrate in the foreseeable future. Such high resolution is important for designing frequency-modulated pulses, or to compensate for dispersion and other phase errors with ultrafast laser systems.Ordinary detection devices do not reveal optical phase, but a variety of methods developed in other laboratories permit direct measurement of phase properties. We use these methods to prove that we can actually produce arbitrary phase or frequency modulation, or to correct for instrumental errors. Figure 4 shows a few examples, using a technique called STRUT which shows how the pulse frequency varies with time. Combined with recent developments in other laboratories of new techniques for wavelength conversion, this approach essentially solves the technological issues associated with generating useful waveforms to alter molecular dynamics; the following sections are devoted to shaped pulse applications in spectroscopy and imaging.

Figure 4. Cubic spectral phase (phase shift proportional to ω^3, or equivalently frequency proportional to ω^2) is usually the first nontrivial broadening mechanism for ultrashort laser pulses; it also limits propagation distances in fibers. Such a phase profile gives a characteristic beating pattern, which can be seen in a cross correlation (left). STRUT detection, which gives the time-frequency profile directly, is shown in the center. The parabolic shape of the frequency profile is apparent in the STRUT, which can also be used to detect much more complex waveforms. We have demonstrated roughly ten times this amount of cubic phase generation (and higher order phase generation), which can be used to correct propagation effects. Right: the "smiley face" pulse (positive cubic spectral phase, plus two transform-limited pulses with specific frequencies).


	The Warren Research Group at Duke University Home Research LASERS NMR Publications Personnel Useful Links Contact Principles of Pulse Shaping At the beginning of the 1980s we developed multiple pulse laser sequences with full inter-pulse phase control, and applied these sequences to study gas dynamics; in the mid-1980s we demonstrated that we could produce complex phase and amplitude modulation as well. However, these methods were limited to roughly 5 ns resolution and <1 W peak power, and thus were far too slow for most optical transitions. In 1986 we developed the first method for altering the shape of subpicosecond laser pulses under computer control, using high-speed electro-optic modulators and picosecond voltage waveform generators. We later used variable length rectangular laser pulses (600-1300 fs) Figure 1. Acousto-optic pulse shaping permits >1000-point resolution for an arbitrarily phase/amplitude modulated pulse. Figure 2. Changing the RF frequency changes the pulse delay. Thus N sine waves give N pulses with complete phase control. to probe hot carrier relaxation by photoconductive sampling. We also produced phase and amplitude modulated laser pulses at λ=1.5 µm with 10 ps resolution using a LiNbO3 intensity modulator and a separate phase modulator. Unfortunately, these pulse-shaping capabilities never became widely available in the ultrafast laser community, largely because of technical complexity. An alternative approach is to use frequency selective optical elements (diffraction gratings or prisms) to spatially disperse the large bandwidths associated with such short pulses, then to use spatial modulators to alter the spectrum. Weiner and Heritage made this programmable for the first time in 1989 with liquid crystal arrays.This approach is much simpler than the electro-optic method, but it suffers from a variety of technological limitations (slow update, gaps between pixels, large phase shifts for complex waveforms). Figure 3. Amplified shaped pulse (200 µJ, 1kHz). We have recently developed a method for modulating spatially dispersed fs laser pulses using microsecond RF pulses in an acousto-optic modulator (AOM) (Figure 1).An AOM diffracts light when an acoustic wave is present; typical conversion efficiencies are about 50% for 1 Watt of RF power.We apply a shaped RF pulse to the AOM's transducer, which creates an acoustic wave traveling through the crystal; we then save the beam diffracted by the modulator. The transit time of a fs pulse through the crystal is too short for significant acoustic propagation, so the acoustic wave looks like a modulated diffraction grating. In comparison with prior methods, this approach has the advantages of faster update times, no pixel gaps, high isolation and simple calibration. It is also better suited for use with existing commercial components; in fact, we have patented this approach and licensed this technology to a commercial laser manufacturer, who will incorporate the technology into their amplified laser systems. Figures 2 and 3 show a few simple examples.Generation of amplified pulses with high fidelity is particularly important for nonlinear applications. Figure 3 shows we can shape prior to amplification and retain good fidelity. We have already demonstrated the equivalent of >1000 independently modulated frequency components, far better than any other approach is likely to demonstrate in the foreseeable future. Such high resolution is important for designing frequency-modulated pulses, or to compensate for dispersion and other phase errors with ultrafast laser systems.Ordinary detection devices do not reveal optical phase, but a variety of methods developed in other laboratories permit direct measurement of phase properties. We use these methods to prove that we can actually produce arbitrary phase or frequency modulation, or to correct for instrumental errors. Figure 4 shows a few examples, using a technique called STRUT which shows how the pulse frequency varies with time. Combined with recent developments in other laboratories of new techniques for wavelength conversion, this approach essentially solves the technological issues associated with generating useful waveforms to alter molecular dynamics; the following sections are devoted to shaped pulse applications in spectroscopy and imaging. Figure 4. Cubic spectral phase (phase shift proportional to ω^3, or equivalently frequency proportional to ω^2) is usually the first nontrivial broadening mechanism for ultrashort laser pulses; it also limits propagation distances in fibers. Such a phase profile gives a characteristic beating pattern, which can be seen in a cross correlation (left). STRUT detection, which gives the time-frequency profile directly, is shown in the center. The parabolic shape of the frequency profile is apparent in the STRUT, which can also be used to detect much more complex waveforms. We have demonstrated roughly ten times this amount of cubic phase generation (and higher order phase generation), which can be used to correct propagation effects. Right: the "smiley face" pulse (positive cubic spectral phase, plus two transform-limited pulses with specific frequencies). Website concerns - contact mike.conti@duke.edu