2.1.

Materials

PVA (99% hydrolyzed; $M_w=\mathrm{85,000}$ to $\mathrm{124,000}\mathrm{g}{\mathrm{mol}}^{-1}$), fluorescein isothiocyanate (FITC), pyridine, dibutyltin dilaurate, sodium citrate monobasic, and fluorescein sodium salt were acquired from Sigma-Aldrich (St. Louis, Missouri) and used without further purification. Dimethyl sulfoxide (DMSO) and methanol were purchased from Thermo Scientific (Waltham, Massachusetts) and Spectrum, respectively. Hydrochloric acid (37%) was purchased from Spectrum Chemical (New Brunswick, New Jersey). Optical equipment consisted of light-emitting diode (LED) lamps (M490F2 and M530F1; Thorlabs, Newton, New Jersey), which emit 490- and 530-nm light, respectively. The M490F2 was 350 mA, 1.5 mW (min) with a bandwidth (FWHM) of 23 nm, while the M530F1 was 1000 mA, 4.0 mW (min) with a bandwidth (FWHM) of 33 nm. These lamps along with an S120VC photodiode power sensor, bare fiber adapter, and the PM100USB power and energy meter were all purchased from Thorlabs (Newton, New Jersey). All specifications and technical information are readily available on Thorlabs.com. The NE-1000 programmable syringe pump was purchased from New Era Pump Systems Inc. (Farmingdale, New York), and the syringe needles were custom manufactured by Hamilton (Reno, Nevada).

2.2.

Functionalization of Polyvinyl Alcohol

Functionalizing of PVA with FITC to make fluorescein-labeled PVA (F-PVA) was modified from Kaneo et al.²³ with small alterations and can be seen in Fig. 1. In brief, 608 mg of PVA was dissolved in a beaker containing 16 mL of DMSO and $100\mu \mathrm{L}$ of pyridine. Once fully dissolved, 6 mg of FITC was added to the solution while stirring on a hot plate. The solution was brought to 95°C, and to catalyze the reaction, $100\mu \mathrm{L}$ of DBDT was added. Then, the solution was magnetically stirred at 450 rpm for 2 h. Being light sensitive, the beaker was wrapped in tin foil to protect it. The solution was then poured into a 50-mL centrifuge tube. Then, 35 mL of butanol was added and vigorously shaken using an analog vortex mixer and by hand before being spun at 5000 rpm for 2 min. The resulting supernatant was poured out and disposed of, and the process of adding butanol, spinning, and disposing of the supernatant was repeated five times. The pellet was transferred into a new beaker and protected from light with aluminum foil. The sample was then placed in a vacuum oven at 80°C overnight after which the resulting fluorescein-functionalized PVA (F-PVA) was ready for spinning.

Fig. 1

Attachment of FITC to PVA to create a dye functionalized polymer (F-PVA). The process was carried out using dibutyltin dilaurate as a catalyst in a DMSO at 95°C for 2 h. Experiment is outlined in Sec. 3.2.

2.3.

Preparation and Wet Spinning of Polyvinyl Alcohol and Fluorescein-Labeled Polyvinyl Alcohol

The technique of wet spinning the PVA and F-PVA fibers was modified from Jorfi et al.¹¹ A solution of F-PVA was prepared by adding 0.25 g of F-PVA and 1.35 g of PVA to a vessel containing a $1:4$ solution of $\mathrm{Di}$(Deionized)$H_2\mathrm{O}$ and DMSO. This solution was heated to 95°C and stirred until the PVA was fully dissolved. The mixture was then loaded into a 10-mL syringe equipped with a 25.4-cm 17 gauge (1.07 mm inner diameter) blunt tip needle that was bent into a gradual 90-deg turn. The exact layout is provided in Fig. 2. The glass tube was filled with $-20°\mathrm{C}$ methanol. The syringe was loaded into the syringe pump, which was then set to $2.0\mathrm{mL}/\min$ and gradually lowered until an interface was formed by the methanol and PVA/F-PVA solution. Flow speed was then lowered to $1.5\mathrm{mL}/\min$. Once a consistent stream formed, the syringe pump was cut and the gelatinous fiber was removed from the needle tip and transferred to a bottle of clean $-20°\mathrm{C}$ methanol. Fibers were kept in $-20°\mathrm{C}$ methanol for 24 h and then in methanol at room temperature for 24 h. The fibers were then transferred onto a sheet of Teflon and allowed to dry at room temperature in a dark cabinet overnight. The next day, the fibers were transferred to a 50°C oven overnight. The final drying step was in the 150°C oven for 15 min. These fibers swelled when placed in water, increasing in length and width by $26.8\pm 2.2\%$ and $19.5\pm 2.8\%$, respectively.

Fig. 2

Setup for wet spinning of PVA/F-PVA fibers. A syringe containing the F-PVA solution is placed in a syringe pump and slowly injected into a glass tube filled with $-20°\mathrm{C}$ methanol. The needle was positioned just above the methanol and was lowered until an interface formed. DMSO and water diffuse out of the fiber causing the now insoluble F-PVA to solidify.

SEM images of the fibers sputter coated with palladium show a relatively smooth surface with some striations. Most notable is a long wavy feature that covers almost the entire length of the fiber. It is likely that this feature is either caused by the turbulence when the liquid F-PVA solution meets the stagnant methanol or from some microscopic protrusion on the needle tip that disrupts the stream. These images are shown in Fig. 3 along with the image from an inverse light microscope showing the end of the fiber.

Fig. 3

(a) and (b) SEM images of fiber. The fiber diameter is approximately $180\pm 20\mu \mathrm{m}$. (c) Light microscope image of F-PVA fiber.

3.1.

Optical Properties Perpendicular Setup

Fluorescein is a commonly used florescent dye with absorbance maxima at about 490 nm and emission maxima around 515 nm. Doughty investigated the changes in fluorescein absorbance with changing pH and found that as the pH decreased, absorbance also decreased.^24,25 To control the pH of the fiber, the fiber was submerged in a sodium citrate buffer solution with varying amounts of 0.1 M HCl solution to maintain the pH levels tested. Figure 4 shows the UV–Vis spectra of a solution containing 11 mg of spun F-PVA fiber dissolved in DMSO before and after the addition of 0.1 M HCl. The transmittance through the fibers was obtained using M490F2 and M530F1 LED sources coupled with an S120VC photodiode power sensor. Fibers were setup using two methods as shown, in Fig. 5. The first method has the light source positioned at one end of the fiber and a sensor at the other (linear) end. In the second method, the light source was positioned so that the direction it was emitting light was perpendicular to the optical axis of the fiber. These two methods were investigated because the linear set is best suited for in vivo applications, whereas the perpendicular set is used for in vitro applications.

Fig. 4

UV–Vis absorbance spectra of F-PVA in DMSO and HCl. Eleven milligram of spun F-PVA fiber was dissolved into 1 mL of DMSO. F-PVA at neutral pH (solid line) and F-PVA at in 0.1 M HCl (pH 1) (dashed line).

Fig. 5

(a) Linear setup with the source opposite the detector. (b) Perpendicular setup with the source at a 90-deg angle to direction of the fiber. The linear setup is perfect for in vivo applications, whereas the perpendicular setup is viable for in vitro applications.

Initial tests of the PVA fibers that did not contain any fluorescein yielded results in line with what we expected. The light simply passed through the fiber and did not travel down its length, resulting in minimal transmittance. With the dry F-PVA, the light struck the fluoresceins that then fluoresced transmitting light down the length of the fiber. Once the fibers were soaked for approximately 5 min, the amount of light transmitted down the fiber decreased as expected. Studies have shown that the fiber possesses the ability to return to a dry state within 30 min, but this experiment was not conducted as the goal of the device is to be implanted and then removed.¹² Further reduction in transmittance was observed once HCl was added to the cuvette, and the response was almost immediate (Fig. 6). The reason for this further reduction in transmittance when HCl is added is that as the pH decreases, the absorbance and emission abilities of fluorescein diminish. Light from the LED no longer activates the fluoresceins and simply passes through the fiber.

Fig. 6

Changes in transmittance with F-PVA fiber dry (•), wet (∘), and after HCl addition (pH 1) (▪).

Interestingly, when sodium citrate buffer was used to investigate a stepwise decrease of pH, the fibers acted differently than before. Instead of a significant decrease in transmittance upon wetting, an increase was observed. Studies by other researchers showed that the ionic strength of the solution significantly affects the fluorescent properties of fluorescein.²⁵ Sjöback et al. explains that as ionic strength increases, there is a buildup of the dianionic form of fluorescein causing increased emission intensity. Other than this feature, the fluorescein acted as expected, having very distinguishable points at each pH tested (see Fig. 7).

Fig. 7

Transmittance of F-PVA fiber as a function of pH, in citrate buffer, pH 6.5, 6.0, 5.5, 5.0, 4.5, 4.0, and 3.5 using a 490-nm LED source and the perpendicular setup. As the pH decreased, transmittance decreased as well. Normalized values show the power at each pH as a percentage of the highest value achieved, which was at pH 7 for all fibers tested. $n=3$ and the error bars represent deviation from the mean.

This same setup was used to observe the effect of temperature on the transmittance, and the results are shown in Fig. 8. The fiber and a sodium citrate solution container were placed in a low-temperature precise oven to regulate the temperature between 25°C and 37°C. Before starting the transmission testing, submerged fibers were isothermally held for 5 min. With increasing temperature and constant pH, the temperature range was taken from room temperature to body temperature. A maximum temperature at 37°C was chosen because of intention of in vivo applications. All tests were run regardless of their decreased pH with the exception of pH 7, which was relatively constant. Compared with the data in Fig. 9, the transmittance values begin to plateau at pH 7, so it is believed that the same response occurs in the set of temperature experiments. This is because of the stray light reaching the sensor, but from the data in Fig. 9 showing pH changes with transmittance, the 37°C at pH 7 power values decrease to a level similar to those seen by other pHs.

Fig. 8

Effect of temperature on transmittance of F-PVA fiber with varying pH, in citrate buffer, pH 7, 6, 5, 4, and 3 using a 490-nm LED source. With an increase in temperature, the transmittance decreased.

Fig. 9

Normalized power of F-PVA fibers ($n=4$) using linear setup. The highest variability was seen at pHs 4 to 6, where the sensitivity of fluorescein is greatest. Normalized values show the power at each pH as a percentage of the highest value achieved, which was at pH 3 for all fibers tested. Error bars represent the deviation from the mean.

3.2.

Optical Properties Linear Setup

When the LED and the sensor were setup at opposite ends of the clear fiber, transmittance values were excellent (not shown). The F-PVA fiber showed excellent response to changing pH, as seen in Fig. 9, which shows the normalized power for four F-PVA fibers. When using the F-PVA fiber, we also switched to the 530-nm source because the response was better. This is because the activation of our fluoresceins occurs at a wavelength of about 520 nm, which is much closer to the 530-nm LED than the 490-nm LED. The absorption spectra are shown in Fig. 4. We also observed that with the higher values, the background noise plays a lesser role.

As shown in Fig. 9, for the linear setup, the pH decreases while the transmittance increases, which is the opposite of what was observed in the perpendicular setup, but the reason for this is believed to be the same. As pH decreases, the absorbance and emission abilities of fluorescein decreas, allowing more light to pass through instead of being absorbed. In the linear setup, it passes through the detector, whereas in the perpendicular setup, it passes through as it is dispersed into the surrounding media.

These results were in good agreement with the results of the glass fibers as shown by other researchers. The difference in power as pH decreased was significant with the transmittance from 2080 nW at pH 8 to 212 nW at pH 3. This wide range shows clear differences between each pH value. Dong et al.⁸ developed a sensor with a pH range of 4 to 13, but the power output difference between their maximum and minimum was only 140 nW, a very narrow range. With such a narrow range, the need for a more sensitive detector increases, and the difference between each pH is smaller, making it more difficult to differentiate. A detector with minimal noise is required as the difference in pH could be just a couple tens of nanowatts. Lee et al.²⁶ developed a fiber using multiple sol–gel coatings and a sensing region of the same length used in this paper (5 cm) and observed that from pH 4 to 12, their power decreased to roughly 10% of their initial output, which is exactly the same as observed in our PVA fibers albeit within a narrower pH range (3 to 8). Lee et al. also showed results from microbent fiber sensors in a pH range of 2 to 12.²⁷ Microbending takes advantage of the losses induced by bending of the fibers by sandwiching it between two corrugated plates. When looking at the same region, we found a similar optical power range of approximately $2.5\mu \mathrm{W}$.

One problem with our setup is that the sensitivity between 7.0 and 8.0 is not very high as physiological pH is within this range. Expanding the range of this scheme could be done in multiple ways using silica fibers. The use of dyes whose range is specific to physiological pHs could be investigated as was done by Khan et al.²⁸ who used 1-2(-pyridylazo)-2-naphthol to develop a sensor in the pH range of 7.2 to 8.4. Another problem stems from the variability between runs. Although the trend is consistent with the transmittance increasing with decreasing pH, we observe variability between each test. As shown in Fig. 9, the greatest variability observed between pH 4 and 6 is due to variation in the amount of fluorescein per unit area in the fiber.

When the fiber is spun in the cold methanol, it shrinks over time with the region closest to the spinneret bearing much of the weight of the fiber compared with the lower portion of the fiber toward the bottom of the tube. This could cause the fiber to be narrower at the top than at the bottom and possibly lead to the variation that was observed. The slope of the curve is highest in this region, showing high sensitivity to the change in pH compared with the lower and higher ranges that show much less variability.

Another effect of this can be seen in the actual values that were obtained. The normalized values as shown in Fig. 9 represent that power is decreased by a fairly consistent and repeatable amount with respect to the maximum value seen for each fiber, but the actual values in nW can vary significantly for each pH. Maximum and minimum powers of one fiber were between 2080 and 212 nW, whereas for another fiber, power ranged from 120 to 20 nW. As these fibers were all cut from a single longer strand, the position on the longer strand that each fiber is being taken from does affect the result. However, further testing is needed to confirm this. The normalized power shows that within a certain error, the results are repeatable; therefore, we can predict what the pH is by looking at what percentage decrease is seen.

3.3.

Mechanical Properties

Dynamic mechanical analysis was done using a TA Instrument’s Q-800 with fiber clamp geometry. Testing conditions comprised of a temperature ramp at $7°\mathrm{C}/\min$ from room temperature to 90°C, 1-Hz frequency and a preload force of 0.01 N. After spinning, the fibers were left to dry under ambient conditions. Wet fibers were stored in a bath of DI water and taken out to conduct testing. A submersion clamp was not used for wet testing. Other modes of testing are possible with changing frequencies to absorb more viscoelastic properties; this study focuses on the response of the fiber with temperature change.^29,30 The results showed an initial storage modulus of approximately 4700 MPa, which decreased with increasing temperature passing through an extrapolated glass transition temperature of 55°C and continued to decrease until 95°C, where it began to flatten out with a modulus of $<1200\mathrm{MPa}$ (Fig. 10). Glass transition temperature was extrapolated and found to be 55°C, and the slope continued until 95°C where it began to flatten out with a modulus of $<1200\mathrm{MPa}$. Upon wetting, the fiber swelled and the modulus decreased to about 145 MPa (not shown). The values achieved by the dry polymer fibers were very similar to the values reported by Jorfi et al. and followed a similar profile as the temperature increased. Jorfi et al. and his team found the modulus to be around 7100 MPa when dry and 35 MPa when wet. They also showed that time-dependent measurements of the tensile storage modulus of a single-mode (SM) commercial optical fiber (S405-XP Thorlabs), which is silica core coated fiber with dual acrylate, as reference in the dry state at room temperature and upon addition of water at 37°C.¹² As shown in Fig. 10, the mechanical properties of the conventional SM optical fiber remained unchanged compared with our fabricated PVA fiber. In addition to the dry state, in our fabricated PVA fiber once placed into an aqueous environment, the modulus decreased significantly to about 140 MPa (data not shown here), which is comparable with values obtained in our previous works.

Fig. 10

DMA testing results of a dry F-PVA fiber ($n=4$). Peak storage modulus was about 4000 to 5000 MPa, which declined with increasing temperature to about 1200 MPa at 95°C. A glass transition temperature can be seen at about 55°C. $n=3$, and the error bars represent the deviation from the mean.