定义：缩写：LMA光纤。具有相对较大的模式面积并且只有单个横模或者少数模式的光纤。

在有些应用中，需要用到具有大的有效模式面积的光纤（LMA光纤），通常是单模导波。由于面积大给定动率的情况下光强会变小，这种光纤通常具有较低的非线性和较大的损伤阈值，因此适宜于用在放大器中放大强脉冲或者单频信号，或者用在无源光纤中传输被放大的光。标准单模光纤的有效模式面积通常小于100 μm2，然而大模式面积光纤的值为几百甚至几千μm2。 
为了得到大的模式面积，需要大的光纤纤芯。但是，大芯径光纤如果具有多个模式，则不一定具有大的模式面积；基模通常远小于纤芯。 

图1：标准纤芯尺寸（例如，直径为8 μm）的裸光纤（无涂层）与大芯径光纤的对比（直径为50 μm）。 
需要注意的是，模式面积大时，波导色散变弱。因此，大模式面积光纤的色散可改变量很有限。 

设计方案和限制因子 
一种直接的设计大模式面积光纤的方法是减小数值孔径，即对于阶跃折射率光纤，可以减小纤芯和包层间的折射率差。但是有一些限制因素：导波效应变弱，如果光纤中有小的缺陷或者弯曲就会引起很大的损耗。因此，数值孔径通常不能小于0.06。为了在更大模式面积中得到稳定的单模导波，可以采用更精细的设计优化折射率分布，实现模式面积达到1000 μm2。这比普通的单模光纤大一个数量级。 
稀土掺杂光纤中实现大模式面积更加困难。通常需要相对高浓度的掺杂其它粒子，例如用于减小淬灭效应，并且掺杂会提高数值孔径。尽管可以采取方式减小折射率差，控制折射率的准度会下降，这会影响实现非常大的模式面积。 
设计光纤支持少数传播模式可以实现较大的模式面积（参阅多模光纤）。这时基模仍是导波中的主要部分，这样光纤放大器的输出接近于衍射极限。如果光入射条件很严格，并且光纤中存在模式混合，会产生更严格的限制，会降低光束质量，引起波束指向波动。 
为了解决上面的限制问题，发展了很多更加复杂的光纤设计（主要是基于光子晶体光纤）和技术。很多情况下，人为的对高阶模式引入很大的传播色散，使多模光纤保持很好的单模传输。另一个角度是最小化不需要的模式耦合。例如： 

• 可以很大程度的弯曲光纤；根据不同的光纤设计，使高阶模式的弯曲损耗足够大直到高阶模式比基模小很多。从这个角度也可以优化光纤设计。需要注意的是，弯曲不仅会引入损耗，还会减小有效模式面积。尤其是大模式面积的阶跃折射率光纤更加明显。在比较不同类型的光纤时，这一效应需要考虑在内。有些得到大模式面积的光纤设计中不采用弯曲，但是弯曲光纤会极大减小模式面积，还有的设计方法中（例如，采用抛物线型折射率分布），最初模式面积比较小，但是对弯曲几乎不敏感。 
• 所谓的手性耦合芯光纤的中心的纤芯很直，光在其中传播，还有另一个纤芯螺旋环绕在中心的芯上。螺旋芯选择性的与中心的纤芯中的高阶模式耦合，基模基本不受影响。选择性耦合的原理是螺旋性影响传播常数，影响方式为在某一有效波长范围内，只有与高阶模式之间的耦合满足相位匹配，而基模不满足。 
• 在泄漏管道光纤中，纤芯周围是少量的较大的空，造成传播模式选择性的泄漏，这样高阶模式存在很大的传播损耗，而基模不存在。早期这种光纤是由光子晶体光纤制作，固体全玻璃设计也可能得到。 

最好的光纤设计得到的有效模式面积为几千μm2。对模式面积并没有严格的限制，但是得到更大的模式面积更难得到单模传输，并且不能允许大的弯曲程度。增大模式面积的同时不可能不影响单模传输。原因在于，模式传播涉及到衍射和波导效应的平衡，由于模式面积更大时衍射不可避免的会减弱，平衡也变得对外界干扰越来越敏感。 

在采用大模式光纤的高功率光纤激光器和放大器中，热透镜效应会改变模式性质，尤其会减小有效模式面积。 
有些情况下采用多丝纤芯可以减弱折射率控制存在的问题，其中光纤纤芯是由一系列二维排列的丝组成的。由于每根丝都是很弱的导波，整体结构可以实现单模导引。这一概念对于掺铒镱光纤中非常重要。 

近来提出的一个新的方法是先采用长周期光纤布拉格光栅从基模耦合光到高阶模式中，然后使该高阶模式在放大光纤中传播，然后采用另一个光纤布拉格光栅转化成基模。采用高阶模式的优势体现在两个方面：这些模式具有更大的有效模式面积，与其它模式的耦合很弱。耦合到高阶模式及耦合出去过程中的功率损耗很小，可以优化光纤使耦合发生在大的带宽中。但是，由于光强的非常不平均会引起很大的困难。甚至当非线性不是很强的某一区域光纤也会发生损伤，因此这一方案可以解决非线性问题，但是不能解决光纤损伤问题。另外，模场延伸到了包层区域（在双包层光纤中延伸到了内包层），这不利于放大过程。 

另一个新概念是增益导引，非折射率导引单模光纤，是一种活性光纤。其中，未泵浦的光纤不能导波，因为纤芯中的折射率低于包层。如果泵浦光足够强，增益导引可以稳定泄漏模式，使其具有很高的光束质量。由于增大纤芯尺寸会使模式损耗急剧下降，因此当纤芯较大时，中等强度的激光器增益就足够实现正的净增益。高阶模式同样也存在，但是需要更高的增益。当低阶模式增益饱和时，高阶模式不能得到足够的增益，这在激光器中经常发生，但是在高增益放大器中不会。这种方法最大的困难就是实现有效的泵浦。需要注意的是，泵浦光并不被导波，由于掺杂纤芯的折射率结构甚至被驱除出去不能被吸收。

常见困难 
采用大模式光纤的常见问题是这种光纤和标准光纤不匹配。（很多光纤器件只有标准模式尺寸。）当将大模式面积光纤熔接到标准光纤上时，模式面积不匹配会在接头处产生附加功率损耗（参阅词条光纤接头）。这一问题有两种解决方法，但是任意一种都无法完全满足： 

• 可以采用一个锥形光纤作为两根光纤的模式转换器。锥形光纤一端的模式尺寸与大模式面积光纤相同，另一端的则与标准光纤的匹配。但是这时就存在两个接点，每一个接点处都有较低的损耗。这种方法最大的困难就是需要使用锥形光纤，但是锥形光纤很难制备。 
• 实验室中更常用的方法是利用自由空间耦合进或者耦合出大模式面积光纤。激光器谐振腔只由体材料制作，另外还有活性光纤。这样会使实验室装置更加灵活，但是不太适合商业使用，因为这种方法需要非常准确的对准并且对灰尘敏感，尤其是光纤端口附着灰尘的情况。 

Acronym: LMA fiber

Definition: optical fibers with relatively large mode areas and a single transverse mode or only a few modes

More general term: optical fibers

For some applications, it is desirable to use optical fibers with particularly large effective mode areas (LMA fibers) – but often still with single-mode guidance, or at most with few modes.

For a given optical power, large mode areas imply reduced optical intensities. Therefore, such fibers effectively exhibit weaker nonlinear effects and a higher damage threshold. That makes them suitable for, e.g., the amplification of intense pulses or single-frequency signals in fiber amplifiers, or in the case of passive fibers for delivery of such light. Whereas standard single-mode fibers have an effective mode area below 100 μm², large mode area fibers reach values of hundreds or even thousands of μm².

For a large mode area, one requires a large fiber core. However, large-core fibers do not necessarily have large modes, if they are strongly multimode; at least the fundamental mode may then be much smaller than the core.

Note that waveguide dispersion becomes weak for large mode areas. Therefore, there is limited scope for tailoring the chromatic dispersion of large mode area fibers.

Figure 1: Comparison of bare (uncoated) fibers with a standard core size (e.g. 8 μm diameter) and a large core (50 μm diameter).

Design Approaches and Limiting Factors

A straightforward design approach to obtain large mode areas is to decrease the numerical aperture (NA), i.e., to decrease the refractive index difference between the core and the cladding, for a step-index fiber design (see Figure 2). However, there are severe limitations: the guidance (waveguiding) then becomes weak, and significant losses can arise from small imperfections of the fiber or from bending (→ bend losses). Therefore, the numerical aperture should normally not be made smaller than approximately 0.06, although in extreme cases values as low as 0.04 have been used [39].

Figure 2: Effective mode area of a step-index fiber as a function of numerical aperture (NA) for different values of the V number. Large mode areas can in principle be achieved even with V = 2, i.e., with strictly single-mode guidance, but this requires the use of a very small numerical aperture, which implies a critical dependence on refractive index inhomogeneities and bending.

There are additional difficulties in applying this concept to rare-earth-doped fibers. Relatively high concentrations of additional dopants like alumina or phosphorus are often required, e.g. for reducing certain quenching effects, and these dopants often increase the numerical aperture. Even if the refractive index contrast can be reduced in some way, e.g. by also somewhat raising the refractive index around the core, the precision of refractive index control may be decreased, and this affects the ability to realize very large mode areas. Also, in high-power fiber lasers and amplifiers based on large mode area fibers, thermal lensing can lead to changes of mode properties, in particular to a reduction in effective mode area [32].

Somewhat larger mode areas can be achieved with fiber designs supporting a few propagation modes (→ few-mode fibers). It may then still be possible to guide light dominantly in the fundamental mode, so that the output e.g. of a fiber amplifier is close to diffraction-limited [2, 3, 6]. Stringent limitations arise from the more critical launch conditions and from mode mixing in the fiber, which can spoil the beam quality and lead to beam pointing fluctuations [17].

Various more sophisticated fiber designs (partly based on photonic crystal fibers) and techniques have been developed for addressing these challenges. In many cases, one attempts to introduce substantial propagation losses for any higher-order modes, making it more easy to maintain robust single-mode propagation in a multimode fiber. Another important aspect is to minimize unwanted mode coupling, basically by achieving a large enough difference in the propagation constants between the fundamental and higher-order modes.

Some examples for possible design strategies of large mode areas are:

• One may strongly bend the fiber; depending on the fiber design, the induced bend losses may be substantial for higher-order modes at a point where they are not yet significant for the fundamental mode. Fiber designs can be optimized in that respect. Note that bending not only introduces losses, but also can reduce the effective mode area. This is particularly true for large mode area step-index fibers. For a fair comparison of fiber types, this effect definitely has to be taken into account [9]. It turns out that some fiber designs have a large mode area without bending, but a much reduced mode area with bending, whereas there are other designs (e.g. with a parabolic index profile) where the mode area starts with a somewhat smaller value but is much less sensitive to bending.
• So-called chirally coupled core fibers [13, 22, 34] have a straight central core in which the signal propagates, plus another core which is helically wound around the central core. It is possible to obtain strongly selective coupling of the helical core only to higher-order modes of the central core, while leaving the fundamental mode essentially unaffected. The principle of this selective coupling is that the helicity affects the propagation constant in such a way that at least in some limited wavelength range phase matching occurs only for coupling to higher-order modes but not to the fundamental mode.
• Other fiber designs work with other kinds of shunt cores, to which unwanted higher-order guided modes are resonantly coupled if they have a similar effective refractive index.
• In leakage channel fibers [8, 12], the core is surrounded by a small number of large holes, making all propagation modes leaky in a very selective way, such that all higher-order modes but not the fundamental mode experience substantial propagation losses. While earlier versions of such fibers have been made as photonic crystal fibers with air holes, solid all-glass designs are also possible [24].
• Pixelated Bragg fibers are modified versions of so-called Bragg fibers, i.e., fibers with a sequence of high-index rings fulfilling a half-wave stack condition. Here, the high-index rings are made discontinuous (pixelated) such that the propagation losses become high for higher-order modes while remaining low for the fundamental mode [36, 38]. Such fibers can be fabricated with stack-and-draw techniques.
• The problem of refractive index control can in some situations be mitigated by using a multifilament core, where the fiber core is realized as a two-dimensional array of filaments [20]. As each single filament exhibits only weak guidance, the overall structure can exhibit single-mode guidance. This concept is particularly interesting for erbium-ytterbium-doped fibers.
• A relatively novel concept is that of the gain-guided, index-antiguided single-mode fiber [7, 15, 23, 27], which is a type of active fiber. Here, the unpumped fiber is not guiding (even antiguiding), as the refractive index of its core is lower than that of the cladding. For sufficiently strong pumping, however, gain guiding can stabilize a leaky mode with high beam quality. As the losses of such a mode rapidly decrease for increasing core size, a fairly moderate laser gain can be sufficient to achieve propagation with positive net gain if the core is large. Higher-order modes also exist, but would require a significantly higher gain. That level of gain is not reached when the lower-order mode saturates the gain, as it easily happens in a laser, but not necessarily in a high-gain amplifier. The greatest challenge of this concept is efficient pumping. Note that the pump light is not guided and even expelled from the doped core by the index structure and the absorption. A diode-pumped fiber laser has been demonstrated with this concept [23], but novel pump arrangements will have to be developed to allow for efficient operation.

The best fiber designs reach an effective mode area of several thousand μm². There is no strict limit, but designs with larger mode areas exhibit less robust single-mode propagation and often can tolerate only very slight bending. It appears that no kind of design can offer a further substantial expansion of the mode area with robust single-mode propagation. The reason for this is essentially that a mode involves some balance of diffraction and waveguiding, and as diffraction inevitably becomes weaker at larger mode areas, this balance becomes more and more sensitive to any disturbances.

Difficulties with Interfacing

A general problem with the use of large mode are fibers is that these fibers are not compatible with standard fiber components. (Many fiber-optic components are only available with standard mode sizes.) When a large mode area fiber is fusion spliced to a standard fiber, the large mismatch in mode areas leads to an excessive power loss at the joint (see the article on fiber joints). There are two solutions to this problem, which however are both not fully satisfying:

• One can use a tapered fiber as a mode converter between the two fibers. The tapered fiber must be made such that the mode size matches that of the large mode area fiber on one end and that of the standard fiber on the other hand. One then has to do two splices instead of one, but each one can have very low losses. The main difficulty with that method is that a tapered fiber is needed, which may not be easy to obtain.
• A frequently used solution for laboratory experiments is to use free-space coupling from and to the large mode area fiber. A laser resonator may then be made with bulk components only, apart from the active fiber. This leads to flexible laboratory setups, which however are not very suitable for commercial use, as they involve sensitive alignment and are sensitive to dust, particularly to dust deposited on the fiber ends.

Alternative Solution: Using Higher-order Modes

Normally, a large mode area fiber is understood to be a fiber which has a single guided mode for polarization direction or only very few modes. However, one can also obtain a large mode area by utilizing a high-order mode of a multimode fiber [11, 21, 35]. In practice, one can first couple light from the fundamental mode to a particular higher-order mode, using a long-period fiber Bragg grating. Thereafter, light can propagate in that mode e.g. in an amplifying fiber, and finally it can be converted back to the fundamental mode with another fiber Bragg grating.

The advantage of using a higher-order mode is twofold: such modes have larger effective mode areas, and they exhibit a weaker coupling to other modes. The power losses associated with coupling to and from this higher-order mode can be small, and the fiber design can be optimized for a broad bandwidth for this coupling.

However, difficulties with that HOM fiber technique might arise from the very uneven intensity distribution. This can lead to fiber damage even in a regime where the overall nonlinearity is moderately strong, so the approach may solve problems with nonlinearities but not those with damage. Also, the mode field significantly extends into the cladding (the inner cladding in the case of a double-clad fiber), which is not ideal for amplification.

Bibliography