3.1.

Probe Miniaturization

To achieve extreme miniaturization, we sought for a probe design allowing controllable size reduction with HF wet etching. We opted for a widely used probe configuration, an assembly of SM, CL, and GRIN fibers,⁵ with an addition of a short CL fiber component at the distal end [Fig. 1(a)]. This key modification is crucial for controlling the postetching morphology and performance of the probes because it protects the GRIN fiber core from nonuniform etching. Prior miniaturization, the probes were $125\mu \mathrm{m}$ in diameter [Fig. 1(b)], composed of a doped amorphous silica core ($5.6\mu \mathrm{m}$ for SM and $62.5\mu \mathrm{m}$ for GRIN), and a nondoped amorphous silica cladding [Fig. 1(a)]. Note that HF is capable of etching nondoped amorphous silica isotropically but etches doped amorphous silica nonuniformly and with much faster rate. Since the cladding of the SM and GRIN fibers and the entire CL fiber are made of nondoped amorphous silica, controlled size reduction of the probe diameter can be achieved by tuning the etching time [Fig. 1(b)]. Note that we observed no deterioration of the protective CL layer for etching down to $12\mu \mathrm{m}$ in thickness (Fig. 2). To illustrate the importance of the CL protective layer, we also etched probes that had no such layer (i.e., conventional design). Contrary to the etching-enabling approach, the GRIN fiber core is exposed to HF. As a result, the morphology of the etched probe’s tip becomes nonuniform and hardly predictable [Fig. 1(c), top probe]. Note that the diameter of the GRIN fiber core determines the miniaturization limit, which is equal to $62.5\mu \mathrm{m}$ for the GIF625 fiber used in this work. Figure 1(c) (bottom probe) illustrates the dramatic effect of HF overetching on the integrity of the GRIN fiber component (i.e., core deterioration). These results demonstrate that the morphology of the miniaturized probes remains unaffected compared to their thicker counterparts if (i) an etching-enabling design is employed and (ii) the miniaturization limit, imposed by the GRIN core diameter, is respected.

Fig. 2

Optical microscopy images of an OCT probe (a) before and (b) after etching.

Note that HF etching has been shown to increase the surface roughness of the etched fiber.¹¹ This effect has been attributed to the deposition of the etching products on the surface of the fiber. For long etching time, especially when the core is reached, micrometric scale defects on the fiber surface can be formed.¹¹ With overetched probes being an exception [Fig. 1(c)], we did not observe such defects for the etching conditions used in this study. However, in some occasions, we observed isolated defects around the splicing points. The miniaturization process has been found to be highly reproducible as the etching rate variability was negligible. Although the miniaturized probes presented increased suppleness compared to their thicker counterparts, no special handling equipment was necessary to perform the measurements presented in this article.

3.2.

Probe Design and Optical Validation

We routinely fabricated miniaturized probes by varying the CL and GRIN fiber length, to evaluate experimentally whether the miniaturization process affects their lensing performance. We used previously published analytic expressions for the working distance (WD) and beam waist (BW) of the probes¹² as guide for the selection of the GRIN and CL fiber lengths. The MATLAB code used for the calculations can be found freely online and executed through Code Ocean.¹³ We designed three different probes, a slightly diverging probe (probe 1; before etching: ${\mathrm{CL}}_{\mathrm{be}1}=0\mu \mathrm{m}$, ${\mathrm{GRIN}}_{\mathrm{be}1}=271\mu \mathrm{m}$, protective ${\mathrm{CLp}}_{\mathrm{be}1}=77\mu \mathrm{m}$, ${\mathrm{OD}}_{\mathrm{be}1}=125\mu \mathrm{m}$), a short WD probe (probe 2; before etching: ${\mathrm{CL}}_{\mathrm{be}2}=272\mu \mathrm{m}$, ${\mathrm{GRIN}}_{\mathrm{be}2}=706\mu \mathrm{m}$, protective ${\mathrm{CLp}}_{\mathrm{be}2}=112\mu \mathrm{m}$, ${\mathrm{OD}}_{\mathrm{be}2}=125\mu \mathrm{m}$), and long WD probe (probe 3; before etching: ${\mathrm{CL}}_{\mathrm{be}3}=272\mu \mathrm{m}$, ${\mathrm{GRIN}}_{\mathrm{be}3}=678\mu \mathrm{m}$, protective ${\mathrm{CLp}}_{\mathrm{be}3}=88\mu \mathrm{m}$, ${\mathrm{OD}}_{\mathrm{be}3}=125\mu \mathrm{m}$).

We etched the probes following the protocol given at the experimental part. The etching process affected only the diameter of the probes and the length of protective CL, while the splicing points remained intact [Figs. 1(b) and 2]. We determined the axial etching rate to be $1.406\mu \mathrm{m}/\min$ for all fiber components forming the probe, while we calculated an etching rate of $1.875\mu \mathrm{m}/\min$ for the longitudinal axis of the protective CL.

Figure 3 shows the beam profiling measurements for the miniaturized probes and Table 1 summarizes their predicted and experimentally measured optical properties. With paraxial Gaussian beam fitting, we measured the experimental WD in air for probe 2 and probe 3 at 250 and $500\mu \mathrm{m}$, respectively. These values are in good agreement with the theoretical ones, namely $230\mu \mathrm{m}$ for probe 2 and $449\mu \mathrm{m}$ for probe 3. We observed slightly astigmatic exiting beams for both probes [see insets in [Fig. 3(a)]. We also observed this effect in nonetched probes due to imperfect splicing points. We obtained consistent measurements with side-viewing beam profiling in liquid [Fig. 3(b)]. The WD for probe 2 and probe 3 was measured to be 304 and $605\mu \mathrm{m}$, respectively. The corresponding theoretical values were $306\mu \mathrm{m}$ for probe 2 and $597\mu \mathrm{m}$ for probe 3. The theoretical and experimentally validated BW and depth of field (DF) values can also be found in Table 1, where a relatively good agreement can be appreciated. Note that the differences between predicted and experimental values lie within the usual range observed for nonetched probes. We consider the imperfect splicing process as the main cause for inaccuracies. In fact, we observed slight thickening of the probes around the splicing points.

Fig. 3

Experimental exiting beam profiles for three different ultrathin probes in air. (a) The two insets show experimental beam intensity profiles at focus. (b) An indicative exiting beam profile acquired by immersing probe 2 in an infrared fluorescent dye solution.

Table 1

Theoretical and experimental properties of miniaturized probes.

These results indicate that the miniaturization process does not alter the optical performance of the probes. In fact, the focusing ability of the probes remains predictable using analytic expressions.^9,12 Although cladding etching does not change probes’ optical performance, the shortening of the protective CL shifts their WD. Therefore, the anticipated longitudinal etching rate must be considered during the designing step to attain the desired final WD.

3.3.

Comparative OCT Performance

We further investigated the optical performance of the miniaturized probes by performing comparative OCT measurements. Figure 4 shows a typical A-scan of a microscope slide placed at a distance of $600\mu \mathrm{m}$ in front of miniaturized probe 3. We acquired SNR measurements by varying the distance between the probe and the microscope slide from 100 to $900\mu \mathrm{m}$. We observed a maximum SNR for $400\mu \mathrm{m}$ ($\sim 64\mathrm{dB}$), which is consistent with the beam profile measurements obtained with the same probe [Fig. 3(a)]. To evaluate whether the miniaturization process affects the sensitivity of a given probe, we compared SNR measurements prior and after etching. Note that the etching step slightly shifts the WD of a focusing probe at a distance proportional to the protective CL shortening. To avoid confusion with the interrogated effect (i.e., splicing point and/or fiber deterioration), we used probes with slightly diverging output beams for these experiments. SNR measurements demonstrated that the miniaturization process had a negligible effect (i.e., $\sim 1\mathrm{dB}$ losses) on the performance of the probes (Fig. 5). We used a moderate sensitivity OCT system (85 dB) for these comparative measurements; the use of a high-sensitivity OCT machine ($>100\mathrm{dB}$) could increase the precision in determining losses.

Fig. 4

Representative A-scan (solid line) and SNR measurements (dashed line) acquired with an ultrathin focusing probe by varying its distance in respect to a glass microscope slide. The peak around 250 μm has been identified as an artifact of our setup as it was also present without connecting the probes.

Fig. 5

Comparison of performance for an individual probe prior (dotted line) and after etching (dashed line). The averaged difference in dB for three individual probes is also shown (solid line). The error bar indicates the standard deviation. Measurements were acquired by varying the probe distance in respect to a glass microscope slide. SNR measurements correspond to its front side (i.e., air/glass interface).

We also performed comparative B-scan imaging of a Scotch tape roll and tissue using the probes prior and after miniaturization. Given that intraoperative guidance of intralocular tools (e.g., subretinal injection cannulas) is a promising field of applications of miniaturized probes,^14–16 we chose the porcine retina as a tissue validation sample. The probe-sample distance was kept at $300\mu \mathrm{m}$ for those measurements and no averaging was applied. The reader may refer to Sec. 2 for B-scan measurements details. In accordance with the quantitative A-scan results, the acquired B-scans indicate untraceable differences in the sensitivity of the probes (Fig. 6). Both porcine retina B-scans reveal a hyper-reflective zone at the back of the retina that can be linked to the ellipsoid zones of cones and rods and the interdigitation zone/retinal pigment epithelium complex.¹⁷ Nevertheless, individual layers cannot be clearly identified within the hyper-reflective zone. The use of the miniaturized probes with application-tailored OCT setups/machines as well as the use of averaging and signal processing strategies can facilitate applications demanding high imaging resolution.

Fig. 6

Representative B-scans of a Scotch tape roll and porcine retina acquired with a forward imaging probe [(a) and (c)] before and [(b) and (d)] after etching-enabled miniaturization.

3.4.

Fabrication and Performance Considerations for Miniaturized Probes

Overall, the miniaturization technique presented in this work requires two additional steps compared to the well-established technique of preparing $125\text{-}\mu \mathrm{m}$ GRIN-based fiber probes. The first step consists in splicing and cleaving the protective CL fiber component. The second step consists of immersing the probe in an HF solution. There is a negligible cost for the materials used in these steps. However, there is a need for some additional equipment, such as an HF-resistant container, protective gloves and glasses, and a chemical fume hood. We estimate the total cost for a 10 cm probe to 4 USD. The additional fabrication turnaround time is $\sim 30\min$ ($\sim 5\min$ for the splicing/cleaving steps and $\sim 25\min$ for the etching step) and can be significantly reduced by etching multiple probes at once. There is virtually no fiber length limitation for the miniaturization process, assuming that both the probe and spliced SM fiber fit in the HF-resistant container. Given that ultrathin ($<125\mu \mathrm{m}$) fiber handling is challenging, miniaturization of the exact required length is recommended. Note that we used inexpensive ($\sim 1100$ USD) telecom splicing/cleaving equipment in this study, which might have compromised the quality of the probes’ splicing points. The use of specialized fiber processing research workstations is recommended as it can minimize sensitivity losses. Furthermore, given that we opted for a common path configuration, there was a cost in sensitivity due to relatively low reference power. Indicative probe reference power values as a percentage of the output power ranged from 0.023% (probe 1) to 0.043% (probe used in Fig. 6) in air and from 0.00092% (probe 1) to 0.0022% (probe used in Fig. 6) for probes immersed in water. The implementation of a reference arm and/or common path technique with fixed reference power¹⁴ can further increase the sensitivity performance of the miniaturized probes.