Twisted Light in Optical Fibers May Lead to Substantial Capacity Enhancements

Fiber-optic networks that underpin the ongoing data revolution have achieved great strides, offering transmission capacities over a single optical fiber link that have grown at the rate of almost 10 times every four years¹ for several decades spanning the late 1980s to early 2000s. This impressive capacity growth was achieved by exploiting almost every attribute of a photon – these include developing the means to encode information in light’s spectrum (color), polarization, phase, as well as amplitude (Fig. 1a).

Having almost exhausted these degrees of freedom, the corresponding growth in information carrying capacity of a single fiber has dropped significantly over the last decade, even as bandwidth demand is forecast to continue to grow exponentially. To address this upcoming “capacity crunch,” the research community has been actively exploring new technological avenues, and the consensus is to exploit “space,” the one degree of freedom of a photon that has remained relatively unexplored.²

Exploiting the “spatial” degree of freedom of light can mean many things. At its most rudimentary manifestation, a “ribbon” of optical fibers would qualify as offering spatial diversity, but this approach faces critical growth challenges for two reasons – (1) the space required to deploy it increases with required capacity, which becomes challenging in deployment scenarios such as undersea cables; and (2) the cost would scale (almost) linearly with capacity, negating one of the great benefits of optical fiber networks over the last decades, wherein parallelism brought down cost per bit even as capacity demands grew.

Another approach is to embed multiple optical fiber cores (the central higher refractive index region within which light is carried with record low transmission loss) within the glass cladding of a strand of an optical fiber. Such multicore fibers (MCF – Fig. 1b) offer some spatial packing benefits, and, depending on the means by which their signals are amplified at periodic repeaters, they may also offer the parallelism that decreases cost/bit. Recently, Google announced the deployment of 2-core MCFs for their undersea cables³. Ultimately, this approach also faces scaling limitations, primarily because embedding too many fiber cores in a single glass cladding matrix necessarily requires increasing the size of the fiber strand, and too large a cross-section of glass decreases its reliability.

Figure 1. (a) Degrees of freedom of a photon in which information can be encoded. The spatial degree of freedom has been the least explored dimension. (b) Multicore fiber, comprising multiple waveguiding cores in a glass fiber matrix. (c) Multimode fibers feature larger size cores and higher refractive index contrasts (top) enabling supporting many spatial modes, but suffer from mode mixing between the modes leading to signal distortion (bottom). (d) Ring-core fibers (top) designed to carry specific ensembles of modes carrying orbital angular momentum represent a class of fibers in which spatial modes can be transmitted with minimal crosstalk, as seen in the clear spiral outputs (bottom).

The third approach uses multimode fibers (MMF – Fig. 1c, top). Spatial modes in a single MMF core can be distinct data carriers, and represent a substantially more spatially efficient means of scaling capacity than MCFs. But a significant debilitating feature is that spatial modes of a fiber tend to mix with one another. Hence, even if one encoded different streams of data in separate modes at the fiber input, light from all these data streams would have mixed during fiber propagation, resulting in, so-called, “speckle patterns” (Fig. 1c, bottom). To counteract this distortion, multi-in multi-out (MIMO) digital-signal processing (DSP) techniques have been attempted to, after the fact, “disentangle” the modal distortions. While researched⁴ since the early 2000s, the use of MIMO-DSP with MMFs has not yet found commercial acceptance, possibly due to the fact that the complexity, cost and energy consumption of the required DSP chips scale quadratically with the number of channels/modes used.

An ideal solution would be an MMF that behaves like an MCF – i.e. a fiber in which spatial modes did not distort one another, minimizing crosstalk. A little over a decade ago, an experimental demonstration⁵ involving, so-called, Orbital Angular Momentum (OAM) modes in a specially designed optical fiber (Fig. 1d, top), featuring a ring, rather than a central, core of high refractive index, enabled the transmission of as many as four distinct data channels at rates exceeding Tb/s over a km with minimal distortions.

Light carrying OAM (Fig. 1d, bottom) has been an intriguing field of study since its elucidation⁶ in 1992, including properties such as the ability to rotate microscopic particles, the ability to achieve super-resolution imaging (which was the basis of a 2014 Nobel prize), and the means to achieve high-dimensional quantum entanglement. The demonstration of the ability to propagate them in optical fibers led to an explosion of interest, both in studying the fundamental properties of such modes in optical fibers, as well as in their potential applications to scale the information capacity of fibers.

Substantial interest and explorations of OAM fiber modes notwithstanding, the number of stable, unmixed modes has been limited to roughly 12 over km-length scales.⁷ If one were to relax the design criteria to allow for some modal crosstalk (and hence to allow for a limited amount of MIMO-DSP processing), transmission demonstrations of up to 100 km for up to 8 modes have been achieved.⁸ The reason for the lack of further progress in mode counts is related to fundamental constraints imposed on a mode’s propagation properties by the refractive index profile of the fiber core. Total-internal reflection (TIR), the primary mechanism by which light is transported in optical fibers with record low losses, limits the effective index n_eff of a mode (a measure of the refractive index a mode effectively “experiences” within a fiber) to be bound between the highest index of the core, and the lowest index of the fiber cladding. Hence, a fiber with many modes has no option but to “pack” them with ever-closer spacing in their n_eff. Crosstalk between any two modes increases exponentially with closer n_effspacing.

Ergo, increasing mode count necessarily increases crosstalk. In conventional MMFs, this crosstalk is debilitating, completely distorting the signal, whereas the limited success of OAM modes in ring-core fibers arose from the fact that angular momentum conservation rules yield some inherent stability, akin to why a bicycle wheel or spinning top doesn’t tip over when spun rapidly. Even so, the fundamental restriction for conventional fiber modes (including the aforementioned OAM modes) remains – packing too many modes within a fiber cross-section leads to crosstalk.

This conventional wisdom was recently upended in a 2023 article in Science. By forcing light transport in a hitherto forbidden regime (where TIR is not satisfied), fiber modes were found to experience negligible loss as long as they carried very high OAM. Optics textbooks state that light propagation in this regime is too lossy to be practical. But what became evident in the experiment described in the 2023 article is that, if light’s OAM is very high (i.e. it twists at very high rates as it propagates), the beam creates a centrifugal barrier for itself, preventing it from leaking.

This topological confinement effect is illustrated in Fig. 2, showing low OAM modes spreading out and being lost, whereas high OAM modes remaining stuck to the core boundary of the fiber. This centrifugal barrier effect is well-known in other areas of physics – the reason orbiting binary stars do not collapse on to each other due to gravity is because their rotational attribute creates the same centrifugal barrier preventing them from doing so.

Figure 2. Twisted light beams carrying orbital angular momentum that do not satisfy the traditional constraint of total-internal reflection for guiding light in a fiber exhibit a peculiar feature that, as the twist rate increases, they tend to “stick” close to the core-cladding interface of the fiber, effectively transmitting with low loss. The sparsity of such modes ensures that they remain robust and resistant to crosstalk even in a fiber featuring several such modes, enabling mode (and hence channel) count scalability.

The reason this finding is exciting for scaling the information capacity of optical fibers is that these so-called topologically confined modes (TCM) occur rather sparsely in a fiber. Modes neighboring them in n_eff are typically not topologically confined (and hence are lost during propagation, as are most conventional modes not satisfying TIR). Hence, even as one increases the packing density of modes, TCMs remain surprisingly robust – measured crosstalks have been as low as -40 dB/km. As of this writing, the TCM platform has yielded mode counts as high as 100. ¹⁰ Figure 3 is a timeline of low crosstalk mode counts demonstrated over the last decade – as is evident, the advent of TCMs has led to abrupt and dramatic increases in achievable mode counts, with the promise of even greater mode counts in the future.

Figure 3. Development of unmixed-mode count in MMFs. The fibers illustrated here include several designs explored for increasing the information-carrying capacity of optical fibers, and the evident jump in mode counts from 10-20 over the last decade to the 50-100 recently achieved, illustrates the power of topological confinement as a light-guiding principle that promises significant future scalability.

TCMs are rapidly finding applications in the development of Erbium-doped fiber amplifiers that form one of the most critical building blocks of a fiber network –the prospects of developing one amplifier that provides the functionality of 100 conventional amplifiers is intriguing, given the attendant benefits of cost, space and energy consumption. ¹¹ The large channel count of TCMs are also beneficial for developing noise-tolerant, high-dimensional quantum computing schemes. ¹² Even as these exciting applications develop with this new regime of light propagation in fibers, several practical problems need to be addressed.

The loss of these fibers is currently higher than that of conventional single-mode fibers that form the bedrock of the fiber-optic internet. It remains to be seen how low these loss values can be driven down with manufacturing optimizations. The means to introduce the high-OAM TCMs into suitably designed fibers remains a challenge, requiring specialized holographic beam-shaping elements. The principles of designing such elements are well-known but the technology for manufacturing them at cost and scale needs to be determined. In summary, topologically confined modes in optical fibers offer a record information-capacity scaling prospect for tomorrow’s internet, once practical and manufacturing feasibilities are addressed.

This article was written by Siddharth Ramachandran, Professor of Engineering and Researcher, Boston University (Boston, MA). For more information, visit here .