Browse

• Abstract

SECTION I

## INTRODUCTION

Photonic crystal fibers (PCFs), which contain axially aligned air channels, represent a very promising and suitable microfluidic platform for accurate detection of physical magnitudes, such as refractive index (RI) [1], [2], [3], biological events [4], [5], and temperature [6], [7] et al. The required liquid volume inside the air channels is much lower than PCF sensors developed for external sensing [8], [9], usually in the order of femtoliter to subnanoliter depending on the infiltrated air channel diameter and the infiltration length. Selective infiltration techniques have enabled a controllable and repeatable fabrication of tunable PCF devices. For instance, by filling a single void in the PCF structure with liquid having a higher RI than the background RI, high sensitivity of temperature response can be achieved by the directional coupler structure [10], [11]. The high RI liquid channel acts as a waveguide and interacts with the core guided modes, resulting in transmission notches at phase matching wavelengths. The temperature sensitivities reported in [10], [11] were 11.61 and 54.3 $\hbox{nm}/^{\circ}\hbox{C}$, and the temperature measurement windows were 1 °C and 1.5 °C, respectively.

Liquid crystal (LC) is a good infiltration candidate due to its thermal and electrical tunability. Previous reports of LC-filled PCF have demonstrated excellent tunable properties for sensing applications such as temperature [12], electric field [13] and hydrostatic pressure [14]. In particular, the temperature sensing is realized by observing the tunable bandgap properties induced by the LC infiltration [15], [16]. To the best of authors' knowledge, this is the first time of reporting selective infiltration of LC in a single void in the PCF structure and realizing the highly temperature sensitive directional coupler. The fabricated device is based on light coupling between the LC-filled channel higher order modes and silica-core guided modes, showing linear temperature response of $-3.90\ \hbox{nm}/^{\circ}\hbox{C}$ within the temperature range from 44 °C to 53 °C. Furthermore, the novelty of our device compared to previous reported directional coupler [10], [11] is as follows. First, we demonstrate experimental and theoretical investigation of the device length dependence on the transmission notch properties. The device exhibits multipeak resonance wavelength region in the transmission notches due to multiple mode couplings with long device length. When the device length is shortened, the multipeak is reduced until a single dip is present resulted from the light coupling between the LC-channel higher order modes and silica core fundamental mode only. Secondly, because of dissimilar dispersion profile of the LC and silica waveguides, the phase matching condition is highly wavelength dependent and the full-width at half-maximum (FWHM) of the transmission dips is narrow, i.e., approximately 2 nm when the device length is short. This is highly desirable for sensing purpose. Thirdly, the temperature sensing range of our device is broader than that of the reported devices [10], [11]. Fourthly, we demonstrate good linearity of the temperature response both experimentally and theoretically. Finally, the selective infiltration technique adopted in this work is simple and efficient compared to femtosecond laser inscription [11]. Only liquid glue is used to obtain the blocking pattern.

SECTION II

## FABRICATION METHOD

The PCF used in the experiment is LMA-25 fiber (NKT Photonics) with air hole diameter of 8.4 $\mu\hbox{m}$, hole-to-hole separation of 16.35 $\mu\hbox{m}$ and outer diameter of 268 $\mu\hbox{m}$ [17]. One air hole in the first ring of voids next to the core is selectively filled with nematic liquid crystal (NLC) 6CHBT (Military University of Technology (MUT), Warsaw, Poland) via capillary effect as shown in Fig. 1(a). The NLC-filled void shows a different color compared to the other voids. The temperature dependence of the refractive indices of 6CHBT from 20 °C to 53 °C is shown in Fig. 1(b). Anisotropic optical properties of the NLC are described by extraordinary and ordinary refractive index of $n_{e}$ and $n_{o}$. In nematic phase, as temperature increases, $n_{o}$ increases and $n_{e}$ decreases. The nematic to isotropic phase transition occurs at 43 °C indicated by the vertical dashed line in Fig. 1(b). In isotropic phase, the refractive index of 6CHBT $n_{iso}$ decreases with rising temperature. The flow process of the NLC induces dominating molecular orientation along the fiber axis [13], $n_{o}$ takes dominating effect in modifying the wave-guiding properties of the PCF [6], [7].

Fig. 1. (a) Microscopic image of the selective infiltration of NLC inside single void of the LMA-25 fiber; (b) the refractive index profile of the NLC, i.e., 6CHBT as a function of temperature.

The PCF and a glass tip made from tapering a standard single-mode fiber (SMF) are micropositioned using the 3-axis translation stage and a microscope. The tapered fiber tip is used to place the adhesive to the PCF end. The liquid glue used in the work is commercially available in stationary shops and commonly used at home or office for paper bonding. When the glue droplet on the tapered fiber touches the PCF, the glue starts flowing on the cross-sectional surface of the PCF and the movement speed depends on its viscosity as well as surface tension. The tapered fiber and the PCF are separated from each other when the glue has covered desirable regions/area. The microscopic images shown in Fig. 2 illustrate different blocking patterns of the voids achieved by using the liquid glue, including the single-hole pattern demonstrated in this work [Fig. 2(e)]. Because of the fluidity, the glue also flows tens of micrometers into the air holes. After the voids are covered by the glue, the PCF is left to set until the glue dries. The blocked side of the PCF is placed into the 6CHBT solution, allowing the liquid crystal to infiltrate the unblocked void(s) via capillary effect. Thereafter, we inspect the other side of the PCF under the microscope and verify the infiltration pattern as shown in Fig. 1(a). One void in the first ring appears white in color because of the NLC and the rest of the voids in darker color are not filled. This confirms that the NLC is filled up to the entire length of the PCF.

Fig. 2. Various blocking patterns using liquid glue. (a) nearly half of the voids are covered; (b) only the center row of voids are open; (c) a pie-shape of voids are open; (d) a triangle-shape of voids are open; (e) a single void is open and used for selective infiltration in this work.

Both sides of the PCF are cleaved and spliced to SMF using a fusion arc splicer shown in Fig. 3. There are a few points that need to be addressed in order to have a good splice for the NLC-filled PCF to SMF. First, it is desirable to have a good cleaved facet of the PCF. Secondly, in order to prevent bubble formation from airgap/NLC in the voids and to minimize the damage of NLC near the splice position due to heat, we adjust the splicer setting and shift the SMF-PCF splice joint from the center with 80 $\mu\hbox{m}$ offset. Consequently, the heat received at the PCF side is reduced. Thirdly, because of the huge outer diameter difference between PCF and SMF, a fully collapsed region is necessary to form a mechanically strong splice between two fibers. Therefore, we re-arc a few times to ensure the air holes are completely collapsed at the PCF side near the splice interface. Nevertheless, the NLC-filled channel is not collapsed and instead it appears in proximity to the splice interface. The single protruding channel toward the splice joint is an evidence of single void NLC infiltration into the PCF. Otherwise multiple protruding channels would be observed. Lastly, it is noted that if the NLC-filled channel extends beyond the joint interface which usually occurs due to excessive heat, a bubble with random shape can be formed during the splicing process and it incurs significant scattering loss. If the aforementioned considerations are addressed appropriately, we are able to fabricate the splice between SMF and NLC-filled PCF as shown in Fig. 3.

Fig. 3. Image of the splice between the SMF and the NLC-filled PCF.
SECTION III

## RESULTS AND ANALYSIS

The selectively NLC-filled-PCF length is about 10.1 cm. The PCF device is placed in an oven and heated. The transmission spectrum of the device is measured and monitored by connecting the SMFs to a broadband amplified spontaneous emission source (1530 nm–1600 nm) and an optical spectrum analyzer. As the temperature increases within nematic phase of the NLC, the transmission spectrum of the device shifts to longer wavelengths as shown in Fig. 4(a) and (b). Around the clearing temperature, i.e., when the NLC transits from nematic phase to isotropic phase, the device spectrum undergoes significant change and is highly unstable as shown in Fig. 4(c). In isotropic phase, the device spectrum is stabilized and shifts to shorter wavelengths as temperature increases as shown in Fig. 4(d). This is in line with the temperature dependence of the NLC refractive index profile, i.e., in nematic phase, $n_{o}$ increases with temperature whereas in isotropic phase, $n_{iso}$ decreases with temperature. The transmission dips are resulted from mode(s) couplings between the NLC-filled channel and the silica core. At all temperatures of measurement, $n_{o}$ or $n_{iso}$ is higher than the background index, i.e., silica. Therefore, it is only possible for higher-order modes of the NLC channel to couple to the fundamental/higher order modes of the silica core at phase matching (resonance) wavelengths $\lambda_{\rm res}$. For simplicity, the temperature dependence of the silica core guided modes is neglected. Consequently, the temperature induced shift of the spectrum can be attributed to the change of effective index of the NLC channel modes.

Fig. 4. Measured transmission spectra in (a) nematic phase, temperature range from 28.8 °C to 36.3 °C, (b) nematic phase, temperature range from 37.4 °C to 40.7 °C, (c) temperature near clearing temperature, transition from nematic phase to isotropic phase, (d) isotropic phase, temperature range from 44 °C to 47.3 °C.

The resonance wavelengths corresponding to the spectral “dips” can be measured as a function of varying temperature. However, it is shown in Fig. 4(a) that the resonant wavelength region exhibits multipeaks, especially at shorter wavelengths. Therefore, the transmission “dip” at longer wavelengths is measured to show the temperature response. The wavelength variation with temperature is plotted in Fig. 5(a). The red-shift of the resonance wavelength is well fitted by a quadratic curve. The refractive index profile of NLC, i.e., $n_{o}$ in nematic phase is also plotted in the inset figure. In isotropic phase of NLC, the device shows better linear response to temperature and the resonance wavelength is blue-shifted with increasing temperature. The linear fitted curve suggests that the temperature sensitivity is $-3.72\ \hbox{nm}/^{\circ}\hbox{C}$ as shown in Fig. 5(b). $n_{iso}$ in isotropic phase refractive index as a function of temperature is plotted in the inset figure.

Fig. 5. Temperature dependence of the resonance wavelength shift, (a) in nematic phase, (b) in isotropic phase.

Next, the NLC-filled PCF is cut back and spliced to SMF. The spectra of the device with different lengths are measured around 45.8 °C to 46. 2 °C, i.e., isotropic phase of NLC, shown in Fig. 6(a)(e). There are a few observations from Fig. 6(a)(e). First of all, as PCF length is reduced, the multipeak in the resonance wavelength region is also reduced. The spectrum of the PCF length of 0.9 cm exhibits a single and sharp notch at around 1576.5 nm. Second, the NLC infiltration into single void does not incur significant propagation loss in the device. Except the sample with PCF length of 1.8 cm, the other samples exhibit insertion loss from 16 dB to 20 dB. As a comparison, we fabricate similar PCF devices without NLC infiltration at lengths 10.1 cm, 6.7 cm, 3.8 cm, and 1.9 cm. The transmission spectra of the modal interference are plotted in Fig. 6(f). The lowest insertion loss is around 6 dB for PCF length of 10.1 cm, and the highest insertion loss is around 15 dB for PCF length of 6.7 cm. The longest PCF device shows minimum insertion loss, indicating that the difference of insertion loss is not a function of fiber length, but influenced by the splice loss. In addition, compared to NLC-filled samples shown in Fig. 6(a)(e), the PCF only devices exhibit lower insertion loss, which indicates that NLC-filled single void channel incurs extra propagation loss. We believe that the large loss exhibited in Fig. 6(d) is due to splicing loss. Thirdly, despite the broad spectral width as well as multipeaks of the resonance wavelength region, the “most lossy” dip wavelength at all PCF lengths is around 1580 nm. Finally, the FWHM of the transmission dips is narrow, i.e., approximately 2 nm when the device length is 0.9 cm. The temperature sensitivity of each sample is measured and plotted in Fig. 7. The device shows good linearity at all PCF lengths, giving temperature sensitivity of −3.72, −3.85, −3.96, −4.09, and $-3.86\ \hbox{nm}/^{\circ}\hbox{C}$ from 42 °C to 53 °C corresponding to the isotropic phase of the NLC. The mean sensitivity of this device is $-3.90\ \hbox{nm}/^{\circ}\hbox{C}$.

Fig. 6. Transmission spectra of the device with different lengths of (a–e) 10.1 cm, 7.8 cm, 6.8 cm, 1.8 cm, and 0.9 cm, respectively. The measurements are taken in the isotropic phase of NLC. (f) the spectrum of the PCF device without NLC infiltration at lengths of 10.1 cm, 6.7 cm, 3.8 cm, and 1.9 cm.
Fig. 7. Dip wavelength of the device with different PCF length shows linear blue-shifts as a function of temperature from 42 °C to 53 °C.

In order to confirm the mode coupling mechanisms as well as the temperature sensitivity, the mode properties of the NLC-filled PCF device are numerically investigated by using commercially available finite difference method software (Mode Solutions, Lumerical). In the simulation, the silica refractive index is assumed to be 1.448. $n_{iso}$ is chosen to be 1.5605 corresponding to the temperature at 46 °C. In coupled mode theory, each core, i.e., the NLC-filled channel and the silica core, is treated as an independent waveguide which is affected by the perturbation of lightwave traveling in the other core. At the phase matching wavelength $\lambda_{\rm res}$, the effective mode indices of both cores are equal, and the maximum power coupling from one core to the other occurs. Because $n_{iso}$ is much higher than that of silica, the phase matching condition can only be satisfied by higher order modes of the NLC-filled channel. In Fig. 8(a), we plot out the calculated dispersion curves for both waveguides. We can observe from the magnified plot Fig. 8(b) that in wavelength region of 1530–1600 nm, NLC-guided modes $\hbox{LP}_{71}$, $\hbox{LP}_{04}$, and $\hbox{LP}_{23}$ modes intersect with silica core guided $\hbox{LP}_{01}$ and $\hbox{LP}_{11}$ modes. Here, LP represents linear polarized. There are a number of intersections where the phase matching condition is satisfied, and the NLC-guided higher order modes can be coupled to both fundamental and second order modes of silica core. We believe that the multipeak resonance wavelength region observed in Figs. 4 and 6 is resulted from the complexity of mode coupling between NLC and silica core guided modes. In addition, we can observe in Fig. 6 that the multipeak resonance wavelength region of all PCF lengths occurs at around 1580 nm. In comparison, as shown in Fig. 8(b), the coupling between the NLC $\hbox{LP}_{71}$ mode and silica core $\hbox{LP}_{01}$ mode occurs at around 1575 nm. Therefore, we track the NLC $\hbox{LP}_{71}$ mode and simulate the device at 48 °C and 53 °C, with $n_{iso}$ is 1.5595 and 1.5573, respectively. The dispersion curves of the tracked modes are plotted in Fig. 9(a). The resonance wavelength $\lambda_{\rm res}$ corresponding to the intersections is measured and plotted in Fig. 9(b). Good linearity is observed with estimated temperature sensitivity of $-3.30\ \hbox{nm}/^{\circ}\hbox{C}$, which is in good agreement with the experimental measurements. In addition, we calculate the coupling length for two mode coupling configurations, i.e., between LC-channel $\hbox{LP}_{71}$ mode and silica core $\hbox{LP}_{01}$ mode, and LC-channel $\hbox{LP}_{23}$ mode and silica core $\hbox{LP}_{11}$ mode, i.e., 5.2 cm and 13 cm, respectively. We believe when device length is longer than 5.2 cm, because of multiple mode couplings occur, the device spectrum exhibits multipeak resonance. As the device length is reduced, the light coupling between LC-channel $\hbox{LP}_{71}$ mode and silica core $\hbox{LP}_{01}$ mode becomes dominating. Therefore, the multipeak is also reduced until only one deep notch is formed as shown in Fig. 6(e).

Fig. 8. (a) NLC-filled channel modes (from $\hbox{LP}_{01}$ to $\hbox{LP}_{23}$) and silica core modes ($\hbox{LP}_{01}$ and $\hbox{LP}_{11}$) at 46 °C, (b) details of NLC-filled channel modes $\hbox{LP}_{71}$, $\hbox{LP}_{04}$, and $\hbox{LP}_{23}$ intersecting the silica core modes $\hbox{LP}_{01}$ and $\hbox{LP}_{11}$.
Fig. 9. (a) The mode coupling between NLC-guided $\hbox{LP}_{71}$ and silica core at different temperatures, i.e., 46 °C, 48 °C, and 53 °C, (b) the resonance wavelength $\lambda\hbox{res}$ is plotted as a function of temperature. The gradient of the linear fitted line is $-3.30\ \hbox{nm}/^{\circ}\hbox{C}$.
SECTION IV

## CONCLUSION

In conclusion, we demonstrate a highly temperature sensitive device based on PCF directional coupler structure both experimentally and theoretically. The device is fabricated by selectively infiltrating NLC into a single void in the PCF structure. The thermally tunable phase matching wavelength between the NLC-channel modes and silica core modes facilitates the highly linear, sensitive temperature response at $-3.90\ \hbox{nm}/^{\circ}\hbox{C}$ within the temperature range from 44 °C to 53 °C.

### ACKNOWLEDGMENT

The authors would like to thank Prof. R. Dabrowski and Dr. E. Nowinowski-Kruszelnicki of Military University of Technology, Warsaw, Poland, for supplying the liquid crystal material.

## Footnotes

This work was supported in part by the A∗STAR-SERC Thematic Strategic Research Program (TSRP) under Grant 102-152-0012, Singapore. Corresponding author: D. J. J. Hu (e-mail: jjhu@i2r.a-star.edu.sg).

## References

No Data Available

## Cited By

No Data Available

None

## Multimedia

No Data Available
This paper appears in:
No Data Available
Issue Date:
No Data Available
On page(s):
No Data Available
ISSN:
None
INSPEC Accession Number:
None
Digital Object Identifier:
None
Date of Current Version:
No Data Available
Date of Original Publication:
No Data Available