定义：
一种可以避免额外非线性畸变和光损伤、可以实现很好光强的，将脉冲啁啾后再放大的技术。

在超短脉冲的放大器中，光的峰值强度会变得非常高，因此可能会产生脉冲的非线性脉冲畸变，甚至会对增益介质和其他元件造成损伤。在这种情况下可以利用啁啾脉冲放大（CPA）来避免以上问题。这一技术最初是在雷达中发明的，后来被引用到光放大器中： 

在通过放大器介质之前，利用强色散元件（展宽器stretcher，如光栅、长光纤）将脉冲啁啾从而其在时间上被拉伸至更长的脉宽，这也就降低了脉冲的峰值功率，因此上述的问题也就得以避免。在放大之后，利用一个压缩器来补偿色散并对脉冲脉宽进行压缩，该压缩器具有与之前的展宽器相反的色散。由于在压缩器上的峰值功率很高，因此也需要增大在压缩器上的光束直径。对于峰值功率极大的设备，通常需要1m以上的光束直径。

  图1：啁啾脉冲放大器中的脉冲时域形状的演变。 

目录

1. 从GW到TW再到PW瓦
2. 展宽器和压缩器
3. 光纤 VS 固体放大器
4. 光学参量啁啾脉冲放大
5. 一个简单的变种
6. CPA的另一种应用
7. 其他注意点

从GW到TW再到PW瓦 
啁啾脉冲放大的方法使得台式放大器可以产生mJ的能量的飞秒脉冲，相应的峰值功率也达到了数TW（1 TW = 1012W，相当于1000大核电站的输出）。而如果要产生更高功率的超短脉冲，其放大器系统则通常包括几个再生放大器和多次通过的放大器，这些放大器大多是基于钛蓝宝石晶体的。这样的放大器可用于高次谐波产生，它们的峰值功率甚至高达PW（1 PW = 1000 TW = 1015 W）的峰值功率。另外，也可以使用光学参量放大器来实现（见下文）。 

展宽器和压缩器 
展宽器和压缩器的很多方面都需要注意： 

1. 在基于光纤的放大器系统中，利用展宽器将脉冲延伸后放大是十分重要的，因为可以减少光纤放大器中的非线性效应。对于不同的输入光谱所需的用于展宽的色散量是相差很大的。通过许多手段都可以实现显著的展宽和压缩，如：衍射光栅对，光纤布拉格光栅和体布拉格光栅。特里西压缩器（Treacy compressor）可以提供反常色散，马丁内斯型光栅设置（光栅之间的有两个透镜）可以提供正常色散。 
2. 对于高能量的系统，压缩器需要耐受高峰值功率的脉冲，且不能引入非线性失真。光栅压缩器在这方面具有良好的特性，因为可以通过增大光束直径来避免以上问题，而使用光纤布喇格光栅时则需要将脉冲能量显著在远低于1μJ。体布拉格光栅允许更高的能量，但却难以具有大的厚度和长度。 
3. 光损耗，特别是在压缩器中，可能会相当大。例如，当利用衍射光栅作为压缩器时，在光栅上的四次衍射极易引起的约50％的损耗。利用电子束光刻技术制备的透射光栅则可以显著的减少损耗，现有的技术下已经实现了单次衍射损耗仅约3％的透射光栅每反射（至少为一个偏振方向而言）。另一种用于减小压缩器损耗的方法是反常啁啾脉冲放大（downchirped-pulse amplification），其利用具有反常色散的展宽器，这样压缩器可以是具有正常色散的玻璃块。 
4. 压缩后的脉冲质量取决于展宽器和压缩器之间的色散匹配的良好与否，并且还受到放大器中残余的非线性效应和放大器自身色散的影响。光栅压缩器会引入大量的高阶色散，因此很难与光纤的色散匹配，但是利用订制的非线性啁啾光纤光栅则很容易匹配。 
5. 展宽器和压缩器的总的色散量也与初始脉冲的带宽有关，带宽不能太小，否则很难展宽，也不能太大，因为太大的带宽会引入高阶色散的匹配问题和增益窄化问题。因此，CPA系统的工作的最佳脉宽在20 飞秒到几百飞秒之间。 

根据所需的性能和其他要求，需要利用合适的展宽器和压缩器才能实现最佳的解决方案。 

光纤 VS 固体放大器 
啁啾脉冲放大的概念也适用于光纤放大器。由于长光纤的高非线性，CPA只能在相对低的脉冲能量下实现，即便利用效果十分显著的展宽器，脉冲能量仍然局限于约10mJ。然而，其平均功率则可以高达几十瓦甚至大于100瓦。因此基于光纤的CPA系统更适合于高脉冲重复率的高平均功率的情况。这种系统中的光纤应该具有以下特点：单位长度的高增益，保偏（强双折射）等。 

全光纤的CPA系统也是可能的，但是其脉冲能量则受到严格的限制。因此，压缩器还通常是由空间光学元件构成。在将来，有可能利用体布拉格光栅取代衍射光栅来实现高脉冲能量的系统。 

光学参量啁啾脉冲放大 
CPA也适用于光学参量放大器，这种技术被称为光学参量啁啾放大（OPCPA），详情参阅光学参量啁啾脉冲放大一文。 

一个简单的变种 
CPA的简单变种实现的光纤放大器不需要特定的展宽器，而是利用光线自身带来的啁啾进行啁啾放大。这在正常色散的光纤中，利用抛物线型的脉冲即可以实现。除了不需要展宽器，这种方法的另一个优点是其得到的啁啾是非常接近线性的。更多内容请参阅抛物线脉冲一文。 

CPA的另一种应用 
CPA的概念也可用于半导体光放大器（SOA）。在这种情况下，拉伸的脉冲的持续时间是可以远远超出放大器的载流子寿命的。因此所述放大器的放大能力不再受到低饱和能量的限制：在放大展宽脉冲的过程中，放大器可以补充能量。此时CPA不再是未避免克尔非线性和光学损伤，而是增加被提取能量。然而，与基于离子掺杂增益介质的放大器相比，这个能量仍然较低。 

其他注意点 
如上所述，CPA并不能用于脉宽较大（几皮秒或更长）的脉冲，因为将这个脉冲进一步展宽需要展宽器具有非常大量的色散。在这种情况下，分裂脉冲放大的技术可以成为一种选择。

Acronym: CPA

Definition: a technique for amplifying pulses to very high optical intensities while avoiding excessive nonlinear pulse distortions or optical damage

In amplifiers for ultrashort optical pulses, the optical peak intensities that occur can become very high, so that detrimental nonlinear pulse distortion or even destruction of the gain medium or of some other optical element may occur. This can be effectively prevented by employing the method of chirped-pulse amplification (CPA), which was originally developed in the context of radar technology, but later applied to optical amplifiers [4]:

• Before passing through the amplifier medium, the pulses are chirped and temporally stretched to a much longer duration by means of a strongly dispersive element (the pulse stretcher, e.g. a grating pair or a long fiber).
• Now, the pulses are sent through an optical amplifier. The long pulse duration reduces the peak power to a level where the above-mentioned detrimental effects in the amplifier gain medium are avoided.
• After the gain medium, a dispersive compressor is used, i.e., an element with opposite dispersion (typically a grating pair), which removes the chirp and temporally compresses the pulses to a duration similar to the input pulse duration. As the peak power becomes very high at the compressor, the beam diameter on the compressor grating has to be large. For the most powerful devices, a beam diameter of the order of 1 m is required.

From Gigawatts to Terawatts and Petawatts

Figure 1: Evolution of the temporal pulse shape in a chirped-pulse amplifier.

The method of chirped-pulse amplification has allowed the construction of table-top amplifiers which can generate pulses with millijoule energies and femtosecond durations, leading to peak powers of several terawatts. (1 TW = 10¹² W, corresponding to the electric output of 1000 large nuclear power stations).

For the highest peak powers in ultrashort pulses, amplifier systems consisting of several regenerative and/or multipass amplifiers are used, which are mostly based on titanium–sapphire crystals. Such amplifiers can be used e.g. for high harmonic generation in gas jets. Large-scale facilities even reach peak powers in the petawatt range (1 PW = 1000 TW = 10¹⁵ W).

It is also possible to use optical parametric amplifiers, leading to the concept of optical parametric chirped-pulse amplification (see below).

Nobel Prize in Physics 2018

In October 2018, the Nobel Prize in physics has been awarded with one half to Arthur Ashkin for work on optical tweezers and the other half jointly to Gérard Mourou and Donna Strickland [31]. The latter two have pioneered the method of chirp-pulse amplification. That was recognized as an essential contribution for the development of laser devices with which one can produce enormously high optical intensities.

Stretcher and Compressor

Several aspects of dispersive stretchers and compressors can be of crucial importance:

• Particularly in fiber-based systems (see below), it is important to obtain very long stretched pulses, with durations e.g. of the order of 1 ns, in order to minimize nonlinear effects in the fiber amplifier. This requires a large difference of group delay within the optical spectrum of the input pulses. Strong stretching and compression is possible e.g. with pairs of diffraction gratings, or with fiber Bragg gratings and volume Bragg gratings. A simple Treacy compressor [1] can provide anomalous dispersion, whereas a Martínez-type setup containing a telescope with two lenses between the gratings [3] can provide normal dispersion.
• At least for high-energy systems, the compressor needs to tolerate high optical peak powers without introducing nonlinear distortions. Grating compressors (with bulk-optical diffraction gratings) are good in this respect, since they can be made for operation with large mode areas, while the use of fiber Bragg gratings limits the pulse energy to well below 1 μJ. Volume Bragg gratings allow for relatively high energies, but are difficult to fabricate with large thickness and length.
• The optical losses, which are relevant particularly in the compressor, can be substantial. For example, when ordinary diffraction gratings are used for the compressor, the four reflections on gratings can easily cause a loss of ≈ 50%. In order not to lose half of the output power, special transmission gratings, fabricated with electron beam lithography, have been developed with losses of only ≈ 3% or even less per reflection (at least for one polarization direction). Another approach for reducing the compressor losses is downchirped-pulse amplification, where (e.g. for the 1-μm spectral region) the stretcher has anomalous dispersion, so that the compressor can be a simple glass block with normal dispersion.
• The quality of recompression depends on a good match between the dispersive properties of stretcher and compressor, and may also be spoiled by residual nonlinear effects and possibly by chromatic dispersion in the amplifier. Grating compressors introduce substantial amounts of higher-order dispersion, which is difficult to match with dispersive fibers, but tailored nonlinearly chirped fiber Bragg gratings can be a solution.
• The total amount of chromatic dispersion required in the stretcher and the compressor also depends on the initial pulse bandwidth, which should thus not be too small. On the other hand, very large bandwidths introduce problems with matching higher-order dispersion, and gain narrowing in the amplifier is another issue. Therefore, CPA systems work best for pulse durations between roughly 20 fs and a few hundred femtoseconds.

Depending on the performance required and on other requirements, different types of stretchers and compressors can present the best solution. The highest performance is achieved if at least the compressor is made with bulk diffraction gratings.

Fiber Versus Bulk Amplifiers

The concept of chirped-pulse amplification is also applied to fiber amplifiers. Due to the inherently high nonlinearity of long fibers, CPA has to be applied already for relatively low pulse energies, and even with strong temporal stretching of the pulses, the achievable pulse energies remain limited to roughly 10 mJ. (Stretching the pulses to more than a few nanoseconds is not practical, and that pulse duration combined with a few megawatts of peak power, which is limited by catastrophic self-focusing, leads to the order of 10 mJ of energy.) However, high average powers of tens of watts or even more than 100 W can be generated [12, 15]. Fiber-based CPA systems are therefore most suitable for high pulse repetition rates combined with high average powers. The fibers used for such systems should be optimized in various respects; they should have features such as a high gain per unit length, polarization-maintaining properties (strong birefringence) and core-less end caps.

All-fiber solutions are possible, but very limited in terms of pulse energy. Therefore, at least the compressor is often made with bulk-optical components. In the future, it may become possible to replace bulky diffraction gratings with volume Bragg gratings while still achieving high pulse energies.

Fiber-based CPA systems also interesting in combination with coherent beam combining. There are plans to combine the outputs of a large number of such amplifiers systems for reaching much higher pulse energies.

See also the article on fiber lasers versus bulk lasers, touching upon various aspects which also apply to CPA systems.

Optical Parametric Chirped-pulse Amplification

The CPA concept is also applied to optical parametric amplifiers, and is then called optical parametric chirped-pulse amplification (OPCPA [8]). The article on optical parametric chirped-pulse amplification contains more details.

A Simple Variant

A simple variant of CPA can be realized with a fiber amplifier where the chirp of the pulses is automatically generated in the fiber, rather than with a pulse stretcher before the amplifier. This can occur in fibers with normal chromatic dispersion, where parabolic pulses are formed. Apart from not requiring a stretcher, an advantage of this method is that the chirp obtained is very close to linear. See the article on parabolic pulses for more details.

CPA with Semiconductor Amplifiers

The CPA concept can also be utilized for semiconductor optical amplifiers (SOAs) [14]. In that case, the duration of the stretched pulses can be well beyond the carrier lifetime of the amplifier. As a consequence, the achievable energy is no longer limited by the low saturation power of such amplifiers: the energy stored in the amplifier can replenished during amplification of the stretched pulse. This means that the main purpose of CPA is in that case not avoiding effects of Kerr nonlinearity and optical damage, but increasing the extractable energy. However, this energy is still low compared with that from amplifiers based on ion-doped gain media.

CPA in the Picosecond Regime

As mentioned above, CPA may not be practical for pulses with relatively long durations (several picoseconds or longer), since very large amounts of chromatic dispersion would be required in the stretcher and compressor. In this regime, the technique of divided-pulse amplification may be an interesting alternative.

Bibliography