定义：
自发辐射被放大的过程。

在增益比较大的激光介质中，自发辐射发光可以被放大到很高的功率。这些放大的辐射可以应用于需要低时间相干性、高空间相干性的领域（下面有介绍）。当工作在激光阈值以下时，激光器中也会发生这种现象。 

由于辐射发生在各个方向，当增益介质具有很大的宽高比时，ASE的方向性会很强。考虑极限情形，对于光纤激光器或者光纤放大器来说，在光纤中的ASE比无定向的辐射发光要强很多。 

在激光器中，尤其是高增益的放大器中，自发辐射放大是需要消除的效应。它限制了单级光纤放大器的增益因子为40-50dB。要获得更高的增益则需要采用多级放大，各级之间由滤波器、法拉第隔离器和光学调制器（光开关）间隔开。
在有些光纤激光器中，如果在其它波长处增益足够产生强的ASE的话，那么在特定频率则不能产生激光。
可以通过优化激光器设计来解决这个问题，例如通过优化光纤长度、掺杂程度等。并且在不需要的波长处产生的ASE也可以通过使用一些特殊光纤（例如光子晶体光纤）来消除，这些光纤在特定的光谱区域意外传播损耗很大。
在体激光器中也存在类似的问题，例如Nd:YAG激光器的工作波长为946nm，在1064nm处存在强的ASE抑制946nm激光强度。 

尽管在一个光纤放大器中放大的自发辐射没有强到能从中获取很多能量，但是它对放大信号产生很大的噪声。光纤放大器的噪声系数是受限于ASE效应的。需要指出的是，准三能级增益介质中的ASE效应是强于四能级介质的。 

图1：在前向泵浦的掺镱光纤放大器中的前向ASE光谱和后向ASE光谱，是采用RP光子学中的RP光纤功率计算软件计算出来的。尤其在准三能级增益介质中，ASE光谱与自发辐射光谱有明显的不同，并且它强烈依赖于传播方向。 

如图1所示，稀土掺杂光纤放大器输出的ASE光谱与光纤纵向辐射的荧光差别很大。这是因为放大和重吸收过程是依赖于波长的。（后者仅在准三能级增益介质中存在。）另外，光纤放大器中辐射的前向和后向ASE光功率也是不同的。
一般来说，在与泵浦光相反方向的ASE更强一些。最后一点，ASE光谱形状依赖于泵浦光强度（如图2所示）。 

图2：在图1中相同的光纤放大器中产生的背向ASE光谱，采用了不同的泵浦功率值。随着泵浦功率的增加，光谱朝着短波方向移动（增益系数迅速变大）并且谱线变得更窄。 

当ASE在诸如光纤等波导结构中产生时，存在许多有意思的现象。例如，单模光纤纤芯中的ASE强度与波导参数无关（如纤芯直径和数值孔径）。不同于一般的情况，ASE功率仅仅依赖于模式数量，而与某一模式的特性无关（传输损耗除外）。
自发辐射的捕获率由全反射条件来决定，仅仅在多模波导的情况下适宜考虑这一参数，而对于单模或者很少模式的情况则不适宜考虑。（具体可参照2007-08-06的Photonic Spotlight）从这个角度看很多光纤放大器模型都是错误的。 

ASE有时也被称为超辐射，而产生超辐射的光源也对应的被称为超辐射光源（即ASE光源或者白光光源）。这些宽带光源（但是通常具有空间相干性）在一些领域具有很重要的应用。 

在自由电子激光器章节中，自放大自发辐射（SASE）这一名词被提到了好多次。自放大是为了强调放大同时发生在产生自发辐射的同一装置中。 

另外还有一个现象与ASE相关，但是在物理上它们又有非常重要的差别，那就是超荧光。 

自发辐射放大的相干特性 
ASE的时间相干性很弱。在时间域中，相关电场存在非常快的随机涨落，也就是说，它具有非常短的相干时间和相干长度。在频率域，这对应的是大的带宽。这一带宽可以与增益带宽具有相同的量级，但是它通常远远小于增益带宽，尤其是在高增益的情况。 
放大的自发辐射的空间相干性对环境的依赖很强。一个小的激光晶体收到具有比较大的模式面积的光泵浦时产生的ASE的空间相干性很低，因为有效激光增益值在不同的方向是类似的。
然而在稀土掺杂的单模激光器中产生的ASE则具有很好的空间相干性。也就说如果采用合适的聚焦光学手段（低象差），这种ASE可以聚焦于一个非常小的点上。 

自发辐射放大是不可避免的么？ 
由于ASE来源于量子效应（自发辐射），是不可避免的放大器噪声。任何对相位不敏感的放大器中的额外噪声与理想四能级激光放大器信号中对应的ASE强度可以比拟。
然而，如果采用准三能级放大器系统或者放大器中存在额外的传输损耗时（来源于吸收或散射），ASE可以大于量子力学极限。具体细节可参阅放大器噪声的相关文章。

Acronym: ASE

Definition: a process where spontaneously emitted radiation (luminescence) is amplified

Alternative term: superluminescence

In a laser gain medium with large gain, the luminescence (most often fluorescence) from spontaneous emission can be amplified to high power levels. This amplified luminescence may be used in applications where light with low temporal coherence but good spatial coherence (see below) is required. It also occurs in lasers, even when operated below the laser threshold.

Whereas luminescence originally goes in all spatial directions, ASE can be strongly directional for gain media with a large aspect ratio. As an extreme case, consider a fiber laser or fiber amplifier, where ASE propagating along the fiber can be much more powerful than the omnidirectional luminescent emission.

Particularly in high-gain amplifiers, amplified spontaneous emission is usually an unwanted effect. It tends to limit the gain achievable in a single stage of a fiber amplifier to the order of 40–50 dB. Higher gain values are possible e.g. for amplification of pulses, if several amplifier stages are used, which are separated by filters, Faraday isolators, and/or optical modulators (switches). Particularly in some fiber lasers, ASE can prevent lasing at extreme wavelengths, if the gain at other wavelengths is high enough for generating strong ASE. Such problems can often be overcome by optimizing the overall laser design, with special attention to fiber length, doping level and the like, and ASE at unwanted wavelengths may be suppressed with certain fiber designs (e.g. photonic crystal fibers) exhibiting high propagation losses outside the desired spectral region. Similar challenges arise in the context of some bulk lasers, e.g., Nd:YAG lasers operating at 946 nm, where strong ASE at 1064 nm can suppress 946-nm lasing.

Even if amplified spontaneous emission in an amplifier is not strong enough to extract significant power, it can contribute significantly to the noise of the amplified signal. The noise figure of a laser amplifier can be considered to be limited by ASE. Note that for quasi-three-level laser gain media this ASE effect is stronger than for four-level media.

Figure 1: ASE spectra of forward and backward ASE in a forward-pumped ytterbium-doped fiber amplifier, calculated with the RP Fiber Power software. Particularly in quasi-three-level gain media, ASE spectra can strongly differ from the spectrum of spontaneous emission, and can be substantially dependent on the propagation direction.

As shown in Figure 1, the spectrum of ASE in the output of a rare-earth-doped fiber amplifier can differ strongly from that of the fluorescence leaving the fiber in a transverse direction. This is because of wavelength-dependent amplification and reabsorption. (The latter occurs only for quasi-three-level laser gain media.) Furthermore, ASE powers emitted in forward and backward direction in a fiber amplifier can differ. Usually, ASE is stronger in the direction opposite to that of pumping. Finally, the spectral shape of ASE can depend on the pump intensity level (see Figure 2).

Figure 2: Spectra of backward ASE in the same fiber amplifier as that of Figure 1, calculated for different pump power levels. With increasing power, the spectrum shifts toward shorter wavelengths (where the gain grows more quickly) and becomes narrower.

Interesting aspects come into play when ASE occurs in waveguide structures such as fibers. For example, the amount of ASE into a single-mode core does not depend on the detailed waveguide parameters (e.g., on the core diameter and numerical aperture). Contrary to a common belief, the ASE power depends only on the number of modes, not on particular mode properties (except for propagation losses). Considerations involving some “capture fraction” for spontaneous emission, which is limited by the condition of total internal reflection, are appropriate only for strongly multimode waveguides, but not for few-mode or single-mode situations. (See the Photonics Spotlight of 2007-08-06 for details.) Some fiber amplifier models are wrong in this respect.

ASE is sometimes also called superluminescence, and correspondingly there is the term superluminescent sources (also ASE sources or white light sources) for light sources emitting superluminescent radiation. Such broadband (but usually spatially coherent) sources are useful for a number of applications.

In lasers, ASE is usually quite weak, thus not extracting substantial amounts of optical power, since the laser gain is typically not as large as the gain in a fiber amplifier, for example. It usually does not have a significant influence on the laser efficiency. However, it contributes to laser noise.

In the context of free electron lasers, the term self-amplified spontaneous emission (SASE) is common. (The addition of “self” emphasizes that the amplification occurs in the same device which produced the spontaneous emission.) Such devices do not have a laser resonator, but generate only strong ASE; the ASE is considered as the laser output, rather than a phenomenon competing with it.

A phenomenon related to ASE, but with important physical differences, is superfluorescence.

Coherence Properties of Amplified Spontaneous Emission

The temporal coherence of ASE can be very low, if its bandwidth is large. In the time domain, the associated electric field exhibits fast random fluctuations, i.e., it has a rather short coherence time and coherence length. In the frequency domain, this corresponds to a large optical bandwidth. That bandwidth can be of the order of the gain bandwidth, but particularly in high-gain situations it is often substantially smaller than that.

The spatial coherence of ASE depends strongly on the circumstances. ASE from a short laser crystal, which is pumped with a large mode area, can have quite low spatial coherence, as the effective laser gain in substantially different propagation directions can be similar. On the other hand, ASE from a rare-earth-doped single-mode fiber can exhibit essentially perfect spatial coherence. This implies that such ASE can be well focused to a very small spot if suitable focusing optics (with low chromatic aberrations) are used.

Is Amplified Spontaneous Emission Unavoidable?

As ASE ultimately arises from a quantum effect (spontaneous emission), it is an unavoidable amplifier noise effect. Any phase-insensitive amplifier must at least add as much excess noise to a signal as corresponds to ASE in an ideal four-level laser amplifier. However, ASE may be above that quantum-mechanical limit, e.g. if one uses a (quasi)-three-level amplifier system and/or there are additional propagation losses (from absorption or scattering) within the amplifier. See the article on amplifier noise for more details.

Bibliography