15 May 2006
A new method of phase and amplitude control delivers near-transform limited pulses, even after passing through complex beam delivery systems such as multiphoton excitation microscopes. Steve Butcher and colleagues from Coherent explain.
Ultrafast laser pulses present unique challenges and offer unique opportunities to study structures and processes in biology, physics and photochemistry. As femtosecond pulsewidths are pushed to ever shorter durations, the performance of the laser system and the overall experiment are often limited by subtle phase and amplitude effects.
Phase errors arise from group velocity dispersion (GVD) induced by virtually every optic within the laser and beam delivery system. Until recently, commercial and user-built tools to counteract these phase delays have been cumbersome to implement and/or limited in their scope of success. Now, a closed-loop method provides complete turn-key control of pulse phase over the entire spectral bandwidth of any ultrafast laser.
This turnkey method can easily be used to correct phase effects from an entire experimental system - including downstream beam delivery optics such as those found in multiphoton excitation (MPE) microscopes. The same pulse measurement and shaping technology can be configured to arbitrarily shape the pulse's amplitude in order to correct for gain narrowing or gain saturation, to produce a shorter transform limit, or to select or block specific wavelength segments. This pulse control method can also be used to produce custom temporal shapes, including double pulses.
The transform limit and GVD
In a mode-locked Ti:sapphire laser oscillator, interference between the longitudinal cavity modes yields a single pulse circulating in the laser cavity. When there is no phase delay between the modes, the result is a so-called transform-limited pulsewidth. Spectral bandwidth and temporal pulsewidth are related through a Fourier transform: a wider spectral bandwidth gives more longitudinal modes that interfere and result in a shorter transform limit (see figure 1).
But in real world ultrafast laser systems, this transform limit is rarely reached. Even if it is reached within the laser, the pulse is chirped (stretched) to some degree by subsequent beam delivery optics before it reaches the interaction zone of the experiment as a result of GVD.
GVD refers to the fact that the velocity of light in any medium is a function of wavelength. Materials such as glass lenses, multilayer optical coatings and laser crystals are all dispersive and can impart significant GVD. Most of these materials exhibit positive dispersion where longer wavelengths travel at a higher velocity than shorter wavelengths. For this reason, a high-performance laser oscillator contains some form of optical arrangement that provides negative dispersion in an attempt to counteract intracavity GVD effects.
The problem can be even more complex in a system that includes an amplifier. Here, the oscillator pulse is deliberately chirped by orders of magnitude before it is injected into the amplifier. This avoids high peak-powers that could damage the optics inside of the amplifier. The amplified exit pulse is then re-compressed. Both the stretcher and the compressor usually consist of a pair of matched diffraction gratings. In principle, the compressor perfectly restores the original shape of the pulse. However, in practice, undesirable phase (and amplitude) modifications can occur within the amplifier, where every optic imparts its own specific dispersion profile to the pulse. The end result is that the group velocity delay can have a complex, nonlinear dependence on wavelength.
An amplifier can also affect the amplitude (spectral profile) of an ultrafast pulse due to gain narrowing or gain saturation. Gain narrowing occurs because the amplifier gain is peaked at its centre wavelength. Every time the pulse passes through the Ti:sapphire crystal, the centre is amplified more than the wings of the pulse, making it successively narrower. In addition to modifying the spectral profile, gain narrowing lengthens the pulsewidth that is obtained after re-compression, because of the transform relationship between spectral bandwidth and pulsewidth.
Controlling phase with an SLM
The ideal phase correction tool would be a flexible, programmable method that adjusts phase as a function of wavelength. This can be accomplished by using a programmable phase mask in conjunction with a zero dispersion stretcher. This type of stretcher laterally spreads the pulse as a function of wavelength and then recombines it in a way that does not chirp the pulse. In the middle of the stretcher, at the so-called Fourier plane, the beam is parallel (collimated) with a linear mapping of wavelength to lateral displacement (see figure 2).
A linear liquid crystal array, often called a spatial light modulator (SLM), can then be used as a programmable phase mask at this location. Increasing the voltage at any of the SLM pixels retards the phase of light passing through that pixel. For an array with 128 pixels, the spectral bandwidth of the laser pulse is divided into 128 equal segments and the phase of each of these wavelength segments can be arbitrarily changed by adjusting the voltage to that pixel. This type of phase correction tool is normally inserted between the oscillator and the amplifier, so that any losses it introduces have minimal effects on the final system power.
This approach can provide a flexible and rigorous correction of phase distortion but it is potentially very difficult to implement as 128 different voltage values are required. In principle it is possible to use a pulse diagnostic instrument such as a FROG (Frequency Resolved Optical Gating) or SPIDER (Spectral Phase Interferometry for Direct Electric-field Reconstruction) to first measure the phase characteristics of the pulse as a function of wavelength. These parameters are then used to calculate phase correction parameters and then actual voltage values for the liquid crystal array.
But even for a skilled laser operator familiar with these calculations, this process would require two or three manual iterations to get an optimum result: a compressed pulse at < 1.1 times the transform limit. While rigorous and successful, this is not a solution that could be widely used.
Fortunately, a new method of phase and amplitude characterization has been developed that has led to an integrated, turnkey pulse measurement and correction solution based on a liquid crystal array. This method, pioneered by Marcos Dantus and co-workers at Michigan State University, US, uses a measurement tool called MIIPS (Multiphoton Intrapulse Interference Phase Scan). Using this approach, the researchers have demonstrated pulses of less than 1.001 times the transform limit.
The basic principles of MIIPS are fairly straightforward. Importantly, MIIPS is a single beam technique that does not require an interferometer. Instead, it uses the already described liquid crystal phase mask approach to impose a known (reference) phase delay function on the laterally dispersed ultrafast beam. The overall phase delay of the beam as a function of wavelength is a combination of the existing phase delay due to the laser optics and this applied reference function.
The re-compressed beam is then sampled using a beamsplitter or mirror and directed into a thin nonlinear crystal designed for second harmonic generation (SHG). In practice a 0.05 mm-thick beta barium borate crystal is used to enable broadband phase matching. This is followed by a spectrometer and CCD detector that measures the SHG intensity as a function of wavelength. This approach yields second derivative data about the spectral phase, which is then integrated by the software to retrieve the original spectral phase of the beam.
Coherent has now exclusively licensed this technology and will launch a turnkey phase control system based on this approach at CLEO 2006 in Long Beach, California, US. This new product, called Silhouette, consists of the phase correction box and a separate fibre-coupled detector.
The integrated correction box contains a low-loss, adjustable bandwidth stretcher and the liquid crystal SLM. The compact detector head contains the thin SHG crystal, which is then fibre-coupled to a spectrometer. System software and a dedicated laptop controller automatically acquire the SHG intensity readings and iteratively adjust the liquid crystal voltages to create the desired phase profile across the laser's spectral bandwidth. The uncorrected and corrected phase information can also be directly accessed by the operator.
For researchers who also want amplitude control as a function of wavelength, a second version will be available in which the single LCD array is replaced with a double array that incorporates a polarizer. Here, each of the 128 elements can also act as a variable amplitude attenuator. The transmitted intensity is controlled by the voltage to the two LCD arrays.
In most instances, the optimum location for the pulse shaping system will be between the oscillator and amplifier (see figure 3). But the fibre-coupled detector head means that it can be inserted at any arbitrary location in the laser system or delivery optics, and the phase and/or amplitude will be re-shaped at that point. In applications where the pulse characteristics of the experimental set-up may change with time, for example in a poorly temperature-controlled environment, the system can re-sample the phase and/or amplitude and quickly make any necessary corrections.
This approach will allow laser operators with limited expertise to take even a poorly corrected ultrafast laser set-up and do two things. First, phase errors can be corrected to achieve a transform-limited pulse. Second, the spectral profile can be broadened so that this transform limit is even shorter than is possible without pulse-shaping.
For more sophisticated users, the system provides complete access to the uncorrected and corrected phase and amplitude information. In addition, this system can be used to arbitrarily modify the phase and amplitude profile of ultrafast pulses to create some other interesting possibilities. This includes the ability to switch the wavelength of a broadband pulse to selectively excite one of the multiple fluorophores in a MPE microscope. The user can also create a double pulse profile with the pulse pair closely spaced in time, for coherent studies on fast processes such as wave packet evolution in real-time molecular dynamics. For telecom applications the pulse-shaper enables the generation of a multiwavelength spectrum for WDM or high repetition rate pulse bursts for OTDM.
Ultrafast laser builders and users have long struggled to obtain the elusive transform-limited output pulse with varying degrees of success. Even with highly corrected systems that approach (i.e. < 1.5 times) the transform limit, the downstream beam delivery optics generally stretch the pulse.
A new integrated approach based on MIIPS allows any commercial or user-built laser system to be interrogated and modified at any point within the entire experimental set-up, including at the focus of an MPE microscope. What's more, the same tool can be used to programme subtle effects into the beam for novel ultrafast experiments.