定义：
通过将光注入到谐振腔来使激光器或者光参量振荡器得到更窄带宽辐射。

种子注入技术主要应用于脉冲激光器和光参量振荡器，主要目的是为了得到更窄光学带宽（线宽）的辐射光。光首先从一个种子激光器（通常为连续波工作的单频激光器），在脉冲积累期刚开始时，注入到调Q激光器或者纳秒光参量振荡器中。如果不采用种子光，从属激光器或者OPO会辐射经历一定程度光增益的多个谐振腔模式，而且各模式的功率分布会存在涨落。如果种子光频率接近于从属装置某一个谐振腔模式的振荡频率，那么这一模式会产生振荡，并且功率远高于其它模式，因此占据输出脉冲的主导地位。在这种情况下，辐射带宽相比于无种子光的辐射显著减小，并且脉冲的时间谱会更平滑，因为避免了模拍频。 

种子注入另一个显著的影响是减小了脉冲积累时间。这在采用纳米泵浦光源的OPO中及其重要，其中泵浦功率在脉冲积累时不能被转化，因此减小脉冲积累时间会提高输出脉冲能量和转化效率。在调Q激光器中，减小的脉冲积累时间不会影响功率效率，但是如果合理稳定谐振腔频率（见后面），时间抖动会减小。 

种子注入不需要种子激光器频率与从属激光器的谐振频率严格相等。只要足够的光功率注入到某一个谐振腔模式，在这个前提下频率存在一些不匹配（从属激光器谐振腔自由光谱范围的几分之一）是可以接受的。这是产生的脉冲频率与注入频率有一些偏移，这是和注入锁定不同的地方，其中辐射频率是严格等于入射频率。在种子注入激光器中也可能存在频率锁定，但是这种方法也可以工作在窄的锁定带宽之外。 

可以采用一些方法将从属激光器频率稳定在种子频率处。例如，可以采用抖动技术产生一个误差信号，这一方法通常用于稳定连续光激光器。另外，还可以自动调整从属激光器频率（例如通过移动谐振腔镜），这样使脉冲积累时间最短。还可扫描一些频率段，当探测到共振时打开调Q开关。 

种子自注入 
普通的种子注入方法一个很明显的缺点是需要一个单独的种子激光器（主激光器）。种子自注入是一个改进的方法，不需要采用种子激光器。种子光是由被注入的激光器自身产生，或者至少采用相同的增益介质。 

例如，存在一种双腔装置，包含了一个泡克尔斯盒作为光开关。刚注入泵浦脉冲到增益介质中时，激光谐振腔中包含频率选择器件，会产生损耗并且不能承受高功率，但是可以允许窄线宽的光积累。一旦达到一定的功率，泡克尔斯盒打开，重新配置激光谐振腔，这样光就不能通过敏感的频率选择器件。尽管这时激光器谐振腔对频率不敏感，前一步产生的光这时作为一个种子，因此经历后面的放大过程线宽依然很窄。 

另一个不同的技术是对于同一谐振腔进行预激射，这样谐振腔损耗保持在一个很高的值。尽管在预激射时期谐振腔不是频率选择的，线宽仍被强烈的减小。

Definition: a technique for enforcing narrowband operation of a laser or optical parametric oscillator by injecting light into its resonator

Injection seeding is a technique which is mostly applied to pulsed lasers and optical parametric oscillators, usually with the main goal of achieving emission with a narrower optical bandwidth (linewidth). Essentially it means that light from a seed laser, which is usually a single-frequency laser in continuous-wave operation, is injected into a Q-switched slave laser or into a nanosecond optical parametric oscillator at the beginning of a pulse buildup period. Without that seed light, the slave laser or OPO would normally emit on multiple resonator modes having an optical gain of comparable magnitude, and the power distribution on several modes may fluctuate from pulse to pulse. If the optical frequency of the seed light is close enough to the resonance frequency of one particular resonator mode of the slave device, that mode can start oscillation with a much higher power than all the competing modes, and can therefore strongly dominate in the output pulse. In that way, the emission bandwidth is drastically reduced, compared with unseeded (free-running) emission, and the temporal pulse profile can be smoother, because mode beating is avoided.

Another effect of injection seeding is a significantly reduced pulse build-up time. This is particularly useful in an OPO with a nanosecond pump source, where the pump power can not be converted during the pulse build-up phase: a shorter build-up time increases the output pulse energy and conversion efficiency. For a Q-switched laser, the reduced build-up time will usually not influence the power efficiency, but with proper stabilization of the resonator frequencies (see below), the timing jitter may be reduced.

Injection seeding does not require an exact match of the seed laser frequency and a resonance of the slave laser. It is sufficient that some substantial optical power is injected into one of the resonator modes, and for that purpose some moderate mismatch of frequencies (some fraction of the free spectral range of the slave laser resonator) is acceptable. The frequency of the generated pulse may then deviate from the injected frequency; this is in contrast to the method of injection locking, where the emission frequency exactly equals the injected frequency. Frequency locking may occur in an injected-seeded laser, but the method can work also outside the narrow locking band.

Different methods can be used for the (approximate) stabilization of the slave laser frequency to the seed frequency. An error signal can be generated e.g. with a dithering technique, as also used for the stabilization of continuous-wave lasers. Alternatively, it is possible to adjust the slave laser's frequency automatically (e.g. via moving a resonator mirror) so that the build-up time is minimized. It is also possible to scan some range of frequencies and fire the Q switch as soon as a resonance is detected (ramp-fire technique).

Self-injection Seeding

An obvious disadvantage of the conventional method of injection seeding is that an additional seed laser (master laser) is required. Self-injection seeding is a modified method, where the seed laser is eliminated. The seed light is then produced from the seeded laser itself, or at least using the same gain medium.

For example, a dual-cavity arrangement has been demonstrated, which contains a Pockels cell as an optical switch [3]. In the initial phase, directly after injection of a pump pulse into the gain medium, the laser resonator contains frequency-selective elements, which introduce some losses and could not tolerate a high power level, but allow for the build-up of light with a narrow linewidth. Once a certain power level has been reached, the Pockels cell is switched, and this reconfigures the laser resonator so that the light does no longer go through the sensitive frequency-selective elements. Although the laser resonator is much less frequency-selective in this state, the light generated in the previous phase acts as a seed so that the linewidth remains narrow during subsequent further amplification.

A variant of this technique uses prelasing in the same resonator, when the resonator loss is kept at a relatively high level. Although the resonator is then not particularly frequency-selective in the prelasing phase, the linewidth may be strongly reduced.

Bibliography