The ability to sense heat, cold, and touch is essential for the human being to survive and interact with the world. As a counterpart of our skin, skin-like tactile sensors with temperature and pressure sensing abilities [1–3] play an essential role in next-generation healthcare [4–6], robotics [7,8], and human–machine interfaces [9,10]. Recently, the rapid development of electronic tactile sensors [11–15] triggered the research of optical tactile sensors due to their attractive advantages, such as high sensitivity, fast response, and immunity to electromagnetic interference [16]. For example, Zhao et al. reported stretchable optical waveguides (3 mm in diameter) for strain sensing in a prosthetic hand [17]. Wang et al. developed a fiber optic mattress for noninvasive monitoring of vital signs based on a silica optical fiber (0.125 mm in diameter) Mach–Zehnder interferometer [18]. Guo et al. demonstrated an optical sensor that can simultaneously detect and distinguish temperature and strain using a polydimethylsiloxane (PDMS) fiber (0.5 mm in diameter) doped with upconversion nanoparticles [19]. Although some fiber optic sensors have demonstrated dual/multiple sensing abilities [20–24], it is difficult to assemble a compact wearable sensor using silica optical fibers or the above-mentioned polymer fibers for healthcare or robotic applications due to the lack of flexibility, large size, or expensive signal processing instruments.

Optical microfibers, with diameters close to or below the wavelength of visible or near-infrared light, offer widely tailorable optical confinement and evanescent fields, which is very attractive to optical sensing on the micrometer and nanometer scales [25–27]. To date, optical microfiber tactile sensors have been attracting increasing research interest due to their possibilities of realizing miniaturized fiber optic sensors with a small footprint, high sensitivity, fast response, high flexibility, and low optical power consumption [28–31]. In our previous works, highly sensitive temperature [32] and pressure [33,34] sensors were developed by embedding microfiber into a thin layer of PDMS film. Because temperature and pressure stimuli often mix together, it is challenging to detect temperature and pressure simultaneously by simply recording the output optical intensity.

In this work, we demonstrate a single optical microfiber enabled tactile sensor for simultaneous temperature and pressure measurement. Typically, a biconical tapered microfiber with a specially designed diameter (e.g., 2.0 μm) is fabricated and embedded in a thin layer of PDMS film based on theoretical simulation. For temperature sensing, shifting a high-order mode cutoff wavelength in a short-wavelength region (e.g., 510–610 nm) is used, as the refractive index of the PDMS layer that determines the cutoff wavelength of the high-order mode is a function of temperature, while the spectrum in the long-wavelength region (e.g., 700–1000 nm) shows a negligible change. When exerting external pressure on the sensor, the peak wavelength representing the high-order mode cutoff has no change, but the pressure caused bending loss leads to a significant output intensity change in the long-wavelength region. Thus, the cross-talk issue can be addressed by choosing two wavelength regions for temperature and pressure measurement. The sensor demonstrates high sensitivity, fast temporal response, and excellent reversibility. As a proof of concept, highly sensitive wrist pulse together with skin temperature measurement is realized, which is critical to the accurate analysis of pulse waveforms.

Human skin has different types of tactile receptors (e.g., Ruffini ending, Pacinian corpuscles, and free nerve ending) that can feel mechanical and temperature stimuli [35–37]. In contrast to electrical multiple-parameter tactile sensors integrating individual sensing elements [11], optical sensors offer more options for signal retrieval from optical intensity, spectrum, phase, and polarization [38]. Mimicking the tactile sensing function of human skin, a single microfiber enabled optical sensor that can simultaneously detect and distinguish temperature and pressure is developed by taking advantage of microfiber’s engineerable optical properties, as shown in Fig. 1(a). Briefly, we choose PDMS to embed a biconical tapered silica microfiber due to its low refractive index (∼1.40 of PDMS versus ∼1.46 of silica), large thermo-optic coefficient (−10−4°C−1) [39], low Young’s modulus (EP=750kPa), and biocompatibility. The flexible PDMS film does not only convert temperature stimuli into the change of refractive index contrast, resulting in a cutoff wavelength shift of a particular high-order mode, as shown in Fig. 1(b), but also transduces pressure stimuli into deformation of the microfiber with high fidelity, leading to a bending-induced drop in output intensity, as demonstrated in Fig. 1(c). By optimizing the diameter of the microfiber, it is possible to choose two wavelengths to realize simultaneous temperature and pressure sensing with negligible cross talk. Experimentally, the excellent mechanical property, the high flexibility, and conformal attachment capability, as shown in Fig. 1(d), Fig. 1(e), and Fig. 1(f), respectively, make the sensor suitable for skin physiological monitoring and robotics applications.

The waveguiding modes supported by microfiber can be characterized by normalized propagation constant V=2πan12−n22/λ, where λ is the propagation wavelength, a is the radius of the microfiber, n1 and n2 are the refractive index of the core and cladding of microfiber, respectively. The core and cladding layers of the microfiber sensors are formed by silica and PDMS, respectively. When the temperature increases, the refractive index of PDMS decreases much faster than that of silica as it has a large thermo-optic coefficient (−10−4°C−1). The increase in refractive index contrast results in a redshift of the cutoff wavelength of a particular high-order mode. Thus, the supported high-order modes depend entirely on temperature, the radius of the microfiber, and the propagation wavelength. To achieve the temperature measurement in a desired wavelength range, precise control of the microfiber diameter is needed. Typically, diameter-dependent mode effective refractive indices of microfiber packaged by PDMS at the wavelength of 532 nm are shown in Fig. 2(a), where the refractive indices of the core silica and cladding PDMS are 1.461 and 1.397, respectively, at 25°C. The microfiber diameter is chosen as 2 μm because the high-order EH21 mode can be easily excited. Thus, the cutoff of this mode will introduce an abrupt intensity change in the transmission spectrum.

The microfiber used in the device is tapered from a multimode fiber (SX62.5, Corning) by a home-built fiber-puller system that consists of a hydrogen flame torch, two motorized translation stages, and optical measurement components. During the pulling process, a continuous-wave laser at 785 nm is coupled into the multimode fiber to excite certain high-order modes. From the cutoff of high-order modes (transmission intensity abrupt drops), microfiber with a desired diameter can be precisely fabricated. The target diameter of the microfiber is set at 2 μm so that only one high-order EH21 mode is terminated around the wavelength of 532 nm, and the interference between degenerate modes is avoided, as shown in Fig. 2(a). Once the microfiber is fabricated, it is placed on a 300 μm thick PDMS film and then covered by another layer of degassed PDMS to form a film with a total thickness of 600 μm. The whole structure is then cured at 80°C for 15 min.

Broadband light from a tungsten halide lamp (SLS201L, Thorlabs) is coupled into the sensor. The transmitted light is recorded by a spectrometer (Maya2000 Pro, Ocean Optics). The sensor is placed on a hot plate (IKA-Ret Basic) for temperature control. A tip integrated with force gauge is mounted on a three-dimensional translation stage with differential actuators for fine pressure adjustment.

To investigate the temperature sensing performance, the sensor is put on a hot plate with a controlled temperature ranging from 25°C to 95°C, and the transmitted spectra are recorded by a spectrometer. Figure 2(b) gives transmission spectra recorded at different temperatures when the diameter of the microfiber is chosen as 2 μm. We find that the peak corresponding to the cutoff wavelength of a higher-order mode shows an obvious redshift, while the spectra have no change in the long-wavelength range (e.g., 700–1000 nm). To verify that this higher-order mode is indeed the EH21 mode as predicted in Fig. 2(a), the output intensity profiles of the fiber are measured after passing through a 600 nm short-pass filter and a 550 nm long-pass filter. Subtracting the profile at 25°C from the one at 95°C results in a donut-shaped intensity distribution, which coincides with the simulated profile of the EH21 mode, as shown in the insets of Fig. 2(b). As shown in Fig. 2(c), the data derived from Fig. 2(b) are fitted by λ=485.5+1.205T, which results in a temperature sensitivity as S=Δλ/ΔT=1.205nm/°C, where λ, T, Δλ, and ΔT are peak wavelength, temperature, the shift of peak wavelength, and the change of temperature, respectively. To test its reversibility, the sensor is tested by alternatively changing the temperature of the hot plate from 40°C to 80°C. After 20 circles, the shift of peak wavelengths shows excellent repeatability demonstrated in Fig. 2(d), indicating the reliability of the sensor for potential applications in health monitoring and robotics.

To measure the responses of the sensor to static pressure, a motorized force tester equipped with a digital force gauge is used to provide an external pressure. Figure 3(a) plots the spectra recorded at different pressures controlled by the force tester. The output intensity experiences a gradual decrease in the wavelength range from 700 nm to 1000 nm with the external pressure varying from 0 kPa to 22 kPa. We find that the peak intensity at 515 nm shows a noticeable decrease, while the peak wavelength corresponding to the cutoff of the high-order EH21 mode has negligible change, as demonstrated in Fig. 3(b). Figure 3(c) shows that the pressure sensitivity increases with the increase of the wavelength from 700 nm to 900 nm, as a result of the increasing fractional evanescent fields. For the wavelength of 900 nm, the intensity shows a nearly linear response to the pressure in a range of 6 kPa to 20 kPa, achieving a sensitivity of 4.5%/kPa by defining sensitivity as S=ΔI/ΔP derived by I=0.9961−1.610×10−3P−3.480×10−3P2+8.968×10−5P3, where I, P, ΔI, and ΔP are the normalized intensity, pressure, the change of normalized intensity, and the change of the pressure, respectively. The coincidence of data points between loading and unloading processes verifies the reliability of the sensor. The sensor also exhibits excellent stability by alternatively changing from 8 kPa to 20 kPa for 20 circles, as shown in Fig. 3(d). Benefiting from the robust encapsulation of the microfiber in the PDMS, no noticeable performance degradation is observed.

To test the response of our sensor to dynamic pressure, a home-built vibration test platform that consists of a vibration actuator, a power amplifier, and an oscilloscope is used to record the output signals. As shown in Figs. 3(e) and 3(f), the clear distinguishability up to 1000 Hz and excellent reversibility of the peak intensity over 1000 cycles indicate a fast temporal response and long-term stability, offering a possibility for real-time monitoring of pressure-related signals, e.g., wrist pulse with high accuracy.

To illustrate simultaneous temperature and pressure sensing, we carry out a series of measurements at different temperature and pressure conditions. The experimental results are shown in Table 1, where T and P are temperature and pressure, respectively. Here the measured pressure values are directly interpolated from the fitting curve at 900 nm shown in Fig. 3(c). The overall error resulted from single parameter fitting as <1% for temperature measurement and <2% for pressure measurement, respectively, indicating the sensor based on such a simple structure can measure temperature and pressure simultaneously with negligible cross talk.Table 1.

Comparison between Experimental Setting Values and Measurement Results

The artery pulse waveform monitoring is a safe, noninvasive means of performing the cardiovascular problem analysis [40], while the skin temperature variation will influence the artery pulse waveforms because heat and cold changes affect the blood vessels vasodilation and vasoconstriction, respectively [41]. Therefore, to realize an accurate analysis of pulse waveforms, it is necessary to measure temperature together with pulse monitoring. Here we demonstrate simultaneous skin temperature and wrist pulse monitoring capabilities by attaching our sensor to a healthy volunteer’s wrist, as shown in Fig. 4(a). The inset of Fig. 4(a) demonstrates the blood vessel undergoes vasoconstriction and vasodilation under cold and warm conditions. Obviously, the cold conditions will result in blood vessel vasoconstriction, and vice versa. The sensor can readily read out wrist pulse difference before and after exercise by utilizing the pressure sensing function, and the wrist pulses before and after exercise are 69 beats/min and 101 beats/min, respectively, as illustrated in Fig. 4(b). The three typical peaks, such as the percussion wave (P-wave), tidal wave (T-wave), and diastolic wave (D-wave), are clearly visible, as shown in Fig. 4(c). Figure 4(d) describes the sensor’s temperature response before and after exercise. The measured transmission peak presents a 1.3 nm wavelength redshift after exercise, corresponding to a temperature change of 1.1°C. For comparison, the thermal images obtained by an infrared camera show 1.1°C temperature change in wrist via exercise, as demonstrated in the inset of Fig. 4(d), exhibiting its consistency with the spectral response curve. In addition, the artery pulse pressure illustrates the different waveforms at cold (22.5°C) and warm (40.0°C) conditions, and the round-trip time (TR) decreases with an increase of the temperature, as shown in Fig. 4(e). The sensor with temperature and pressure sensing capabilities shows its great application potential in healthcare.

In summary, a simple optical microfiber tactile sensor is designed and fabricated for simultaneous detection of temperature and pressure by tailoring the waveguiding properties. Typically, an optical microfiber with 2 μm diameter is embedded into a thin film of PDMS to sense the thermal and mechanical stimuli. In this case, the shift of a high-order mode cutoff wavelength in the short-wavelength range is employed for temperature measurement, while the intensity change in the long-wavelength range is used for pressure sensing. Benefiting from the high thermo-optic coefficient of PDMS and considerable fractional power guided outside the microfiber, the sensor achieves high sensitivity in both temperature and pressure sensing. Furthermore, the sensor shows fast temporal response and excellent reversibility due to being without parasitic electrical effects and flexible packaging. As a proof-of-concept application, wrist pulse waveforms together with skin temperature are recorded simultaneously, which is critical to the cardiovascular problem analysis. Owing to the seamless connection with standard optical fiber, we believe our sensor is especially useful in healthcare, robotics, and human–machine interface applications.