Improving SERS Diagnostics with Hollow-Core Negative Curvature Fibers

When surface-enhanced Raman spectroscopy (SERS) is applied in a medical diagnostics context where an endoscopy is involved-such as in the lung-the performance of the technique depends in part on the performance of the optical fibers used in the endoscopic probe. Fibers are needed that reduce or ideally eliminate the background Raman signal and that minimize signal attenuation, or loss, as the signal is transported to the detector. To be most useful, these fibers must have a small cross section. Hollow-core negative curvature fibers show promise for delivering on those performance goals.

Novel optical probes, such as those using hollow-core negative curvature fibers, can serve in a variety of applications, particularly in medical diagnostics. These developments can be particularly valuable for improving the capabilities of Raman sensing, including surface-enhanced Raman spectroscopy (SERS). In the past, one of the limitations of conventional solid-core fibers for fiber-optic Raman probes was the high Raman background signal generated in the silica cores of these fibers. This limitation led to the use of hollow-core fibers. However, previous hollow-core fibers have had various characteristics that make them challenging to fabricate and use.

Recently, negative-curvature hollow-core optical fibers have been developed to address the limitations of earlier hollow-core fibers. As part of the Proteus project (www.proteus.ac.uk), Jonathan Knight and his colleagues at the University of Bath (UK) have been working to develop a negative-curvature hollow-core optical fiber for use with SERS probes in medical diagnostics, such as in the lung. For such use, the fibers must generate no (or nearly no) background Raman signal-even for a probe that may be 2 m in length-and efficiently collect the Raman signal generated by the SERS probes back to the dichroic or detector. In this interview, Knight and researcher Stephanos Yerolatsitis talk about the development of these fibers. Yerolatsitis will present on this work at the SPIE Photonics West conference in January 2018.

First, can you briefly explain the basic structure of a hollow-core fiber generally, and of a negative curvature hollow-core fiber specifically?

There are a few different types of hollow-core fibers. What they all have in common is that the light in such fibers is guided in an air core, rather than in a glass core as in all conventional optical fibers. To trap the light in a hollow core, the core needs to be surrounded by a silica microstructure that reflects the light back into the core. Different hollow-core fibers reflect using different physical mechanisms and different shapes of microstructure. In a negative curvature hollow-core fiber (NCF) the surrounding structure is quite simple and consists only of a single ring of touching ice cream cone–shaped silica shapes, filled with air or another gas. Each of the "ice-cream" boundaries acts as a Fabry-Perot resonator that reflects the light back into the core in specific wavelength ranges. These ranges are called the guiding bands. An NCF has several bands. Each of those can extend over several hundreds of nanometers of wavelength. Knowing the parameters of the fiber, such as most importantly the wall thickness, we can determine the wavelengths of the guiding bands of the fiber.

Why or how does negative curvature improve the optical performance of hollow-core fibers?

The "negative" or "inverted" curvature of the thin silica walls surrounding the core (which forms the surface of the ice cream in the cone) of a hollow-core fiber can reduce the fiber attenuation. One reason for this effect is that the negative curvature keeps the guided light away from points where the silica strands meet one another. Those meeting points or junctions can serve to increase the attenuation.

One important aspect of the performance of these fibers is mode attenuation. What is mode attenuation, and how is it related to core diameter? Does this relationship depend on the spectral range?

Light can only escape from the fiber core when it gets to the edges. For a given reflection coefficient, for a larger core the light will travel further between reflections and the fiber will have lower losses. By increasing the core diameter the mode attenuation is therefore reduced. Actually, for other reasons, the attenuation decreases even faster than one might initially expect. However, the fiber becomes more sensitive to bend losses at the same time. There can be a balance between these two effects: bending loss and the mode attenuation.

How does the core-wall thickness of the fiber affect attenuation?

The core-wall thickness determines both the wavelengths of the low attenuation bands and also their spectral width. Thinner walls would produce the broadest transmission bands. They also produce the least sensitivity to nonuniformity in the wall thickness measured across the fiber section.

How is attenuation affected by the length of the fiber? What can be done to address this challenge?

The attenuation itself is not affected by the length. Instead, the attenuation is usually measured in units of decibels per kilometer (dB/km). An additional consideration is the uniformity along the length of the fiber. Nonuniformity along the fiber length on almost any length scale would have the potential to increase the attenuation.

How does the material that a fiber is made of-fused silica or soft glass-affect the fiber's performance?

The ultimate performance as measured in terms of several key parameters is closely related to the overlap of the guided light with the glass, with lower overlap giving better results. So far, we have demonstrated overlaps of just one part in 10,000, but this figure may yet be further improved by changes to fiber design or by the use of different materials. We fully expect that consequent performance improvements, including further reduction of Raman background generation, will follow.

It is challenging to scale down fabrication of fibers without increasing imperfections that lead to greater attenuation or loss. What approaches can one take to address this challenge?

The uniformity during the fiber draw is one important challenge. One way to address this challenge is to have a small draw-down ratio between the different fabrication stages. This approach allows one to have a better control over the core-wall curvature and the final design of the fiber. However a smaller draw-down ratio means that one would fabricate fewer meters of the desired fiber. A second challenge is that for smaller fiber microstructures (as required for short wavelengths), surface tension forces become stronger and can become overwhelming. The fiber is like a whole lot of bubbles, but each bubble has holes and so the gas in the holes can leak out along the fiber length. Use of differential pressurization in the preform during the fiber draw is an essential part of fabricating good quality fibers for short wavelengths.

How does the performance of negative-curvature hollow-core fibers in terms of macro-bending loss compare to that of other types of fibers?

Bending loss in hollow fibers differs significantly from the loss mechanisms familiar from conventional fibers. We have a good understanding of one form of bend loss-resonant bend loss, which occurs at certain bend radii when guided light is resonant with the cladding structures. The radii at which this occurs is strongly affected by the design of the fiber and is thus linked to other characteristics like attenuation and single-modedness. Both our group and others have demonstrated fibers with critical bend radii that are sufficiently small for most applications, typically a couple of centimeters.

How does the performance of negative-curvature hollow-core fibers compare to that of other types of fibers in terms of dispersion and nonlinearity?

Compared to standard step-index fibers, in NCFs the interaction between the mode and the silica surrounding cladding is very low. This means that the nonlinearity is far lower than in conventional fibers. Exactly how low will depend both on the fiber design (as that determines the degree of overlap), on what gas fills the fiber, and on exactly what sort of nonlinearity is being investigated. For the most common nonlinearity caused by the Kerr effect, a good starting point would be to consider that the nonlinearity could be 10,000 times lower than in a conventional fiber. When that is combined with the fact that the group velocity dispersion of the hollow fiber always crosses zero within each guiding band, and is typically low over most of each band, this fiber design becomes an obvious choice for delivery of high-power ultrashort picosecond or femtosecond pulses. There is literally no other sort of fiber that can deliver these pulses as the combined effects of self phase modulation and dispersion tear the pulses apart in conventional fiber after just a few millimeters or less. We anticipate that the work we are reporting here on SERS based on optical fibers will in future be extended to other processes where useful spectral information can be gained through fiber using nonlinear processes, based on ultrashort pulse delivery.

A primary goal in your work on this new fiber was to reduce the background Raman signal. How big is the problem of background Raman in SERS signals being used in medical diagnostics? How low do we need to get that background?

In medical diagnostics, the generated background overwhelms any weak signal collected from the distal end of the system. One reason for this effect is that the light propagates through fiber lengths of meters but the sample of interest may only be a few hundred micrometers thick, giving far less opportunity to generate signal. Several ways to overcome this problem have been investigated. For example, the background can be separated using a different fiber for excitation and collection, or a filter can be used at the fiber distal end to remove the fiber-generated Raman before the signal is generated. In order to make a versatile device that can be used for medical and in vivo experiments we would like to use a single fiber and have as small a distal end as possible. That enables us to use the system in all sorts of places where a complex system using filters, for example, cannot be deployed. There is no rule of the thumb for how low the background needs to be. Different samples will generate stronger or weaker SERS signals. Ideally we would like the background to be nonexistent. Reducing the background by 1000 times can be really beneficial for medical diagnostics.

What in particular have you done differently with your latest fiber to reduce the background Raman signal?

We have been examining different fiber designs and ways to minimize the background. One important aspect for minimizing the background is the single-mode performance of the fiber. Higher-order modes will interact more with the silica cladding. For such a fiber it is important to excite only the fundamental mode.

What results have you seen so far in terms of the new fiber's ability to reduce the background Raman signal? Have you reached the limit of how much you can reduce that background signal?

We see a huge reduction but we haven't reached the limit yet. We are examining different fiber designs to minimize the background but also maintain the desired collection efficiency. Coupling light very efficiently into the core of such fiber requires care in design as well as in experiment. Any slight misalignment results in light propagating through silica and hence the generated background is much bigger.

You have developed this fiber for use with SERS probes for use in the lung, to diagnose lung infections and diseases. Were there any particular considerations for that intended application of the fiber that guided your work on it?

The work on this fiber is part of an interdisciplinary project called Proteus, funded by the UK's Engineering and Physical Sciences Research Council and led by the University of Edinburgh in Scotland. In order to optimize the potential of our system for use in the human body, we have decided to strictly limit the size of the fiber probe. Our focus within the project is on the lung, and so the fiber probe would be one part of a multifunctional system to be deployed using a bronchoscope. While we are developing the fiber, our collaborators in Edinburgh are developing SERS probes for specific applications. They have developed SERS probes that have different responses for different pH levels. In order to distinguish this difference between the responses it was crucial to minimize the generated silica background from the fiber.

Is this fiber ready for use in its intended application? Or is more development needed?

The fiber and setup are ready to be used for Raman spectroscopy but there is a lot of work to be done in terms of endoscopic use of that fiber. For example, a more portable and robust system will be desirable for such applications.

What are your next steps in this work?

We are looking into different fiber designs to further reduce the generated silica background. We are also trying to increase the signal collection efficiency of our system. We further need to design an optical system that is robust and mobile, and investigate the use of a range of SERS probes for different measurands.

Jonathan Knight

Jonathan Knight is a professor of physics at the University of Bath in the United Kingdom.

Stephanos Yerolatsitis

Stephanos Yerolatsitis is a post-doctoral researcher at the University of Bath.