Dispersion considerations in ultrafast CPA systems

A basic assumption underlies many designs for chirped-pulse amplification (CPA) of ultrashort pulses, This assumption is that a Taylor's series expansion of the dispersive delay is well behaved in the sense that each phase order in the expansion produces an effect on the pulse that is significantly smaller than the effect of the previous order, This work investigates this assumption both qualitatively and quantitatively. We show quantitatively that the requirements for achieving sub-20-fs pulses are much more stringent than for 100-fs pulses. We find that when the basic assumption holds, a chirped pulse amplification (CPA) system may be designed by zeroing each order in succession, but that zeroing may not work well for some systems that are not well behaved. For these cases minimizing the overall dispersion becomes necessary. We discuss some common optical components including bulk materials, expanders, and compressors and show that they generally satisfy the basic assumption. We also discuss some optical components or effects that may not satisfy the basic assumption such as dielectric mirrors or etalons, and describe means of limiting the problems associated with them when designing ultrafast CPA systems. Finally, we discuss the problem of optimizing a CPA system in the laboratory, and describe a new polarization-gate (PG) frequency-resolved optical gating (FROG) arrangement that is based on thin-film polarizers and that allows accurate measurements of the phase as well as the intensity with minimal dispersive effects.