Double-clad fibers (DCFs)—originally developed for high-power lasers [1]—are gaining popularity in biomedical imaging and sensing, as they allow the combination of coherent and incoherent imaging through their single-mode core and multi-mode inner cladding regions, respectively. While this combination was first used to obtain speckle-free high-resolution endoscopic images [2], DCFs were rapidly used for dual-modality imaging requiring coherent and incoherent light detection. The combination of optical coherence tomography (OCT) with fluorescence maps, for instance, provides high-resolution imaging based on backscattering with specific molecular contrast [3–6].

In order to minimize coupling losses and improve the robustness of current free-space beam-splitter (or dichroic filter) approaches, all-fiber DCF couplers (DCFCs) can be fabricated using twisting [4], polishing [5], or fusion tapering [7,8]. These techniques typically involve two identical commercially available DCFs. We previously reported such a coupler designed for OCT and fluorescence detection [9] with an achromatic transmission of 90% single-mode light and more than 40% multi-mode light extraction. This previous design was limited to a theoretical equipartition limit of 50%, dictated by the symmetry imposed by the combination of the two identical fibers. For detection of weak signals, however, an ideal DCFC would show a more efficient multi-mode extraction capability over a wide spectral range, without affecting the single-mode transmission.

In this Letter, we present a novel DCFC design exceeding the equipartition limit by exploiting an asymmetry of the fiber diameters in the fused section of the coupler. Figure 1(a) shows a schematic diagram of this asymmetric DCFC (A-DCFC) combining a commercially available DCF with a multi-mode fiber (MMF) having a larger diameter. The coupler was realized with a fusion-tapering technique modified to allow maximum multi-mode signal extraction without affecting the single-mode signal propagation in the core of the DCF. We describe the setup used to characterize the coupler for single-mode and multi-mode propagation over a broad spectrum and provide a comparison with its symmetric counterpart for imaging of a weakly backscattering biological sample.

 

Fig. 1. (a) Schematic diagram of an asymmetric DCF coupler achieved by fusing a DCF (top) with a MMF (bottom). The core, inner cladding, and outer cladding regions are shown in red, white, and blue, respectively. (b) Schematic diagrams (left) and photographs (right) of cross-sections of the DCF. (c) Representative fused-section. (d) The MMF. In (c), one can appreciate the absence of outer cladding, which was removed chemically. Scale bar: 100 μm.

Download Full Size | PDF

The A-DCFC is an all-fiber coupler capable of quasi-lossless transmission of single-mode light propagating through the core as well as efficient extraction of multi-mode light from the inner cladding of a DCF. Excitation light is launched in the core of the DCFC at Port 1 [Fig. 1(a)], which allows for coherent illumination of a sample (Port 2). Backscattered light from the sample is collected by the core and by the inner cladding of the DCF at Port 2. The core signal returns to Port 1 while the inner cladding signal is extracted by the MMF and is sent to Port 3.

Figures 1(b) and 1(d) show schematic diagrams (left) and photographs (right) of the cross sections of the two commercially available fibers used for fabricating the A-DCFC. The DCF is a 125 μm diameter fiber (Nufern SM-9/105/125-20A) having a 9 μm core (numerical aperture (NA) of 0.12, cutoff at 1210 nm) and a 105 μm inner cladding (0.2 NA), while the MMF is a 220 μm diameter coreless multimodal fiber (Nufern MM200/220-22A, 0.22 NA). Figure 1(c) shows a schematic diagram (left) and a photograph (right) of the cross section of the fused section of the coupler.

Higher extraction of the multi-mode signal is achieved by modifying the fusion process. After removing both fibers’ acrylic coating on a span of 30 mm, this zone is etched with hydrofluoric (HF) acid in order to remove the external cladding, which lowers the multimodal signal transfer. The etching rate is tuned in order to have sufficient surface quality before the fusion of both fibers ($0.2\mathrm{\mu m}{\min }^{-1}$ in a 49% HF solution diluted $1:2$ with demineralized water) [10].

The fibers are installed side by side and maintained mechanically with appropriate tension and geometry on a custom fusion-tapering setup. Both fibers are fused with a micro-torch fueled with an oxygen–propane mix in order to heat the glass up to temperatures ranging from 1500°C to 1600°C. The fusion initiates a coalescence of the structure into an ellipse (within $\approx 200\mathrm{s}$), preserving the refractive index profiles over the cross section of the coupler. During the fusion step of the process, the single-mode response is monitored with a broadband source (BBS: Hewlett–Packard, Broadband Light Source, 83437A, 1200–1650 nm) injected in the core of the DCF and sent to an optical spectrum analyzer (OSA: ANDO Electric, AQ6317). Tapering of the coupler is then performed until losses begin to appear on the single-mode signal. The coupler is glued on a glass substrate using low-shrinkage UV curing glue and secured in a stainless-steel tube, which is sealed on both ends with flexible boots.

Two alternate metrics are used to characterize DCFCs. A first metric is the coupling ratio, $\gamma$, defined as

While the single-mode transmission of the A-DCFC is monitored during the fabrication process, the multi-mode transmission is measured on a different setup once the device has been packaged. Figure 2 shows the multi-mode characterization setup consisting of a swept-source laser centered at 1310 nm (20 kHz, 25 mW, 75 nm bandwidth) [11] focused on the cleaved face of a DCF through a diffuser in order to excite a high number of modes. This DCF segment is placed in a mode scrambler to promote uniform excitation of modes. The free end of the DCF is then spliced with Port 1 of the device under test, and the signal is measured at Ports 2 and 4 using a calibrated integrating sphere (Thorlabs, NJ, S145C). The injection power is measured afterward by sectioning the input branch of the coupler to take into account possible loss due to the splice (cutback technique). Figures 2(b) and 2(c) show transmission curves for near-infrared (using the wavelength-swept source and a photodetector) and visible spectra (using a BBS and an OSA).

 

Fig. 2. (a) Characterization setup for the multi-mode transmission. (b) Comparison of multi-mode transmission in the visible range for symmetric (dashed line) and asymmetric (solid line) designs. (c) Comparison of single-mode (red) and multi-mode (blue) transmission for symmetric (dashed line) and asymmetric (solid line) designs in the near infrared range. MS, mode scrambler; PD, photodetector.

Download Full Size | PDF

This A-DCFC transmits more than 95% of the single-mode signal (from Port 1 to Port 2, as well as from Port 2 to Port 1) and more than 75% of the multi-mode signal (from Port 2 to Port 3, as well as from Port 1 to Port 4) over the near-infrared spectral band (1260–1340 nm). In the visible range, the A-DCFC transmits $\approx 72\%$ in either direction. The propagation of visible light through the core, however, is slightly multi-mode (15–20 supported modes) as the cutoff of the DCF has been measured at 1210 nm using the 0.2 dB criterion.

Our A-DCFC compares well with previously reported DCFCs, as these iterations exploited symmetric designs that are subjected to the equipartition limit. In the near-infrared region, multi-mode transmission around 40% was achieved [7,8]. In the visible region, used for fluorescence detection, the A-DCFC nearly doubles the extracted light from the inner cladding. It also shows a lower chromaticity from 450 to 1000 nm, compared with the symmetric version. This can be explained by the additional chemical-etching step in the asymmetric version, exposing the inner cladding and thus facilitating transmission at shorter wavelengths. Single-mode core propagation in the A-DCFC is slightly better than that of the symmetric version, since only a shallow taper is imposed to the structure.

To demonstrate the potential of this new device for biomedical applications, we integrated the A-DCFC in a spectrally encoded (SE) wide-field imaging setup [12], which allows for rapid imaging while adding an additional requirement that the A-DCFC be achromatic. In a SE imaging setup [shown in Fig. 3(a)], the fast-axis scanner is replaced by a diffraction grating, which spreads wavelengths from a broadband source across a sample. The slow axis is achieved using a galvanometer-mounted mirror (Thorlabs, NJ, GVSM001). The polygon-based wavelength-swept laser is coupled to the core of the DCF (Port 1) through a splice with single-mode fiber (Corning, SMF-28). Light travels to the imaging arm (Port 2) and is collimated into an 2.8 mm diameter beam (Thorlabs, NJ, F260APC-C). The wavelengths are then spread using a diffraction grating (Wasatch Photonics, 1004-1, 1145l/mm) and finally focused through a 100 mm focal length achromatic doublet (Thorlabs, NJ, AC254-100-C) on the sample. Backscattered light couples in part coherently through the core and is sent to a photodetector (New Focus, FC-1117) via an optical circulator (Thorlabs, NJ, CIR-1310-50-APC) for single-mode imaging. A larger fraction of the backscattered light from the sample couples back into the inner cladding and is extracted by the A-DCFC (to Port 3) to a photodetector (Thorlabs, NJ, PDA10CS). A custom beam dump is installed on the end of Port 4 to prevent backreflections. Imaging is performed at a rate of $20\text{images}/\mathrm{s}$ over a field of view of $1{\mathrm{cm}}^2$. We acquired 800 by 800 pixel images of a 28-day-old mouse embryo fixed in 4% paraformaldehyde shown in Fig. 3(b).

 

Fig. 3. (a) Diagram of the SE imaging setup. PD, photodetector; BD, beam dump; DG, diffraction grating; C, circulator; galvanometer-mounted mirror (not shown). (b) Picture of the sample: a mouse embryo fixed in 4% paraformaldehyde. (c)–(e) Images of the embryo with the A-DCFC [(c) single-mode (Media 1) and (e) multi-mode (Media 3)] compared with a DCFC [(d) multi-mode (Media 2)]. Scale bar: 1 mm. Gray scale in nanowatt.

Download Full Size | PDF

Figure 3 further compares images acquired in single-mode detection [Fig. 3(c), taken from Media 1] with images acquired in the multi-mode detection using a symmetric DCFC [Fig. 3(d), taken from Media 2] and an A-DCFC [Fig. 3(e), taken from Media 3]. As reported previously, the multi-mode detection allows for speckle reduction, increased signal collection (factor varying from 4 to 10 depending on the sample), and for depth-of-field improvement. In both multi-mode images, we can appreciate the three-dimensional aspect of the sample and more easily identify features, such as paws, eyeballs, and nostrils. While displayed in different intensity scales, the multi-mode image comparison allows appreciating the increased collection efficiency (factor of 1.5–2 depending on the sample) of the A-DCFC for weakly backscattering sample.

The equipartition limit being overcome, we define a new theoretical limit for multi-mode transmission. This can be achieved by comparing the etendues of the fibers used, leading to new guidelines for asymmetric designs. The optical etendue (or throughput), $G$, is commonly defined [13] as follows:

Noting that $\mathrm{NA}=n\sin \alpha$, we finally deduce that the etendue depends on the mode area and NA of the waveguide as

Maximizing the multi-mode power transfer from Port 2 to Port 3 entails maximizing the ratio of $G_3$ (optical etendue at Port 3) over $G_1$ (optical etendue at Port 1), as light coupling into Port 2 will split between Ports 1 and 3 [following the convention of Fig. 1(a)]. Neglecting excess loss, the power transmission is predicted to be

 

Fig. 4. Multi-mode transmission as a function of etendue ratio. The solid line shows the theoretical prediction, while the dots show experimental data points from couplers with different combinations of fibers.

Download Full Size | PDF

In conclusion, we demonstrated an asymmetric DCFC capable of more than 70% multi-mode transmission, while leaving the core signal unperturbed (more than 95% transmission). This design exceeds the previous equipartition limit exploiting the asymmetry in the fiber combination. The coupler was realized using two commercially available fibers, as well as fabricated and characterized using a systematic and detailed approach. To demonstrate the potential of this new device, we integrated the coupler in a SE imaging setup with a low-reflectivity sample. We finally defined a new theoretical multi-mode transmission limit using the concept of optical etendue and predicted the feasibility of an asymptotically perfect device. Several applications can benefit from this improved signal collection such as OCT combined with fluorescence imaging, spectroscopy or Raman scattering.