High-Pressure Sensors Based on Laser-Manufactured Sintered Silicon Carbide

: In this work Sintered Silicon Carbide (S-SiC) samples have been used to fabricate ﬁber-optic-coupled pressure sensors. The sensor structure reproduces a low-ﬁnesse Fabry–Perot (FP) interferometer. Laser manufacturing of cylindrical S-SiC samples was performed to deﬁne the thin membrane geometry of sensors. FP cavity is deﬁned by the end-face of a single mode ﬁber and the S-SiC diaphragm surface. Hence, pressure is evaluated by measuring the cavity depth by a dedicated optoelectronic system coupled to the single mode ﬁber. Exploiting the excellent properties of S-SiC, in terms of high hardness, low thermal expansion, and high thermal conductivity, realized devices have been characterized up to 20 MPa. Experimental results demonstrate that produced sensors exhibit a non-linearity around ± 0.6%F.S. and a high input dynamics. The all-optic sensing system proposed in this work would represent a good alternative to conventional solutions based on piezoelectric e ﬀ ects, overcoming the drawback related to electromagnetic interference on the acquired signals. In addition, the mechanical characteristics of S-SiC allow the use of the sensor in both automotive and aerospace hostile environments as pressure monitors in combustion engines.


Introduction
Laser material processing represents a powerful tool in the modern manufacturing industry. Laser cutting is the most common industrial application of lasers, as well as welding, drilling, and marking processes, which have reached the right maturity to be accepted as standard tools in modern industry. Moreover, in recent years laser processing of materials also opened new perspective in additive manufacturing and micro-fabrication [1]. This trend is driven by the development of new applications as well as the increasing adoption of advanced ceramics in already established applications due to superior material and performance properties. However, it is worth noting that these properties are also the ones that do not allow obtaining accurate geometries using conventional machining techniques in a simple, cheap, and time-effective way [2]. In this context, lasers appear to be a promising technology with respect to ordinary material removal methods, because the machining operation is carried out without any contact between the laser system and the part, therefore ensuring the elimination of cutting forces, tool wear, and machine vibration [3]. In fact, the process efficiency depends on the thermal and optical properties of the material, allowing also hard or brittle materials, such as ceramics, to be machined. Moreover, lasers can be used for different applications, from the ablation of advanced ceramics [4], to the hardening of structural materials [5], to the joining of hybrid sensors able to operate in high-temperature [26,27] and MEMS-based sensors and actuators [28,29], finding application in several fields, from automotive to aerospace, from healthcare to consumer electronics [30,31]. Indeed, as aforementioned, SiC represents the elective material for the fabrication of mechanical components able to work under severe conditions. Pressure sensors able to work in harsh environments are fundamental for combustion chambers and gas turbine engines monitoring [32]. Crystalline-SiC (c-SiC) piezoresistive and capacitive pressure sensors, as well as c-SiC diaphragms on silicon have been reported in the literature [26,33]. Moreover, polycrystalline SiC (poly-SiC) diaphragms on poly-SiC substrates for pressure sensors realization also demonstrated their effectiveness in high temperature conditions [34,35]. In our research, S-SiC membranes, fabricated by means of laser treatment able to engrave cylindrical samples, have been used for pressure sensors fabrication. The technological steps necessary for the realization of the sensitive elements are mainly reduced to the laser treatment of S-SiC, overcoming difficulties related to complex etching processing of c-SiC, as well as lowering the costs related to the pristine material growth.
Guaranteeing immunity to electromagnetic interference, fiber-optic pressure sensors [36] and Fabry-Perot (FP) interferometric sensors [37][38][39] have been proposed to operate in hostile and aggressive media. The feasibility of a low-finesse FP pressure sensor has been demonstrated for thin diamond membranes [39]. Adopting the same optoelectronic detection set-up, in this work thick cylindrical samples of S-SiC ceramics have been used to fabricate fiber-optic FP pressure sensors. A fiber laser apparatus was employed to ablate S-SiC creating a hole and leaving a flat diaphragm with defined thickness. The FP cavity is represented by the end-face of a single mode fiber (SMF) and the S-SiC diaphragm surface. Deforming the S-SiC membrane, the external pressure is measured evaluating the cavity length by means of the optoelectronic system coupled to the SMF and able to acquire the light interference signal. Sensor pressure sensitivity and full-scale range can be properly designed adjusting diaphragm diameter and thickness. The proposed structure would represent a valid alternative for the fabrication of pressure sensors able to work in the harsh environments the S-SiC ceramics sustains.

Sintered SiC Laser Processing
Sintered SiC samples (BC-4 SiC, Bettini Textile Division, LC, Italy) used in this work were obtained through dry molding process of high-purity SiC powder. All cylindrical samples, see Figure 1, have a diameter of 5 mm and 2 mm in height. Table 1 summarizes their main characteristics. work in harsh environments are fundamental for combustion chambers and gas turbine engines monitoring [32]. Crystalline-SiC (c-SiC) piezoresistive and capacitive pressure sensors, as well as c-SiC diaphragms on silicon have been reported in the literature [26,33]. Moreover, polycrystalline SiC (poly-SiC) diaphragms on poly-SiC substrates for pressure sensors realization also demonstrated their effectiveness in high temperature conditions [34,35]. In our research, S-SiC membranes, fabricated by means of laser treatment able to engrave cylindrical samples, have been used for pressure sensors fabrication. The technological steps necessary for the realization of the sensitive elements are mainly reduced to the laser treatment of S-SiC, overcoming difficulties related to complex etching processing of c-SiC, as well as lowering the costs related to the pristine material growth.
Guaranteeing immunity to electromagnetic interference, fiber-optic pressure sensors [36] and Fabry-Perot (FP) interferometric sensors [37][38][39] have been proposed to operate in hostile and aggressive media. The feasibility of a low-finesse FP pressure sensor has been demonstrated for thin diamond membranes [39]. Adopting the same optoelectronic detection set-up, in this work thick cylindrical samples of S-SiC ceramics have been used to fabricate fiber-optic FP pressure sensors. A fiber laser apparatus was employed to ablate S-SiC creating a hole and leaving a flat diaphragm with defined thickness. The FP cavity is represented by the end-face of a single mode fiber (SMF) and the S-SiC diaphragm surface. Deforming the S-SiC membrane, the external pressure is measured evaluating the cavity length by means of the optoelectronic system coupled to the SMF and able to acquire the light interference signal. Sensor pressure sensitivity and full-scale range can be properly designed adjusting diaphragm diameter and thickness. The proposed structure would represent a valid alternative for the fabrication of pressure sensors able to work in the harsh environments the S-SiC ceramics sustains.

Sintered SiC Laser Processing
Sintered SiC samples (BC-4 SiC, Bettini Textile Division, LC, Italy) used in this work were obtained through dry molding process of high-purity SiC powder. All cylindrical samples, see Figure  1, have a diameter of 5 mm and 2 mm in height. Table 1 summarizes their main characteristics.   , with a galvanometric scanning head (by LASIT SpA, NA, Italy) equipped with a flat-focusing F-Theta lens (by LINOS, QIOPTIQ, Goettingen, Germany) with focal length of 160 mm has been used for S-SiC processing. Table 2 summarizes the main characteristics of the laser-treatment apparatus. Preliminary tests were conducted at scanning speeds between 500 and 2000 mm/s, at maximum light power (minimum repetition rate of 30 kHz). In order to evaluate the influence of the process on the manufacturing, three different scanning strategies have been verified: LINE, NET and CROSS, whose meaning is clarified with the scheme reported in Figure 2. High-resolution images of treated samples by scanning electron microscopy (SEM) are illustrated in Figure 3. The smoothest surface was obtained by means of the CROSS strategy at a scan-speed of 1000 mm/s. The other strategies show surfaces with higher levels of damage as well as a higher roughness. Indeed, high scanning speeds (2000 mm/s) do not allow to obtain a uniform surface remaining visible the effect induced by each laser-pulse. Conversely, at lower scan-speed smoother surfaces are gained thanks to a better overlapping of laser pulses.
Based on the preliminary investigation, S-SiC samples were laser treated according to the process parameters' combination reported in Table 3. Laser processing was performed in order to ablate each S-SiC cylinder realizing the desired diaphragm diameter and thickness. A picture of an example of laser-processed sample is reported in Figure 4a. In this sample, a 3 mm in diameter hole was realized up a depth of 1.9 mm, hence a membrane 100 µm thick, with a removal rate of about 3.1 mg/min, which is to be considered for micromachining operations [20,40]. Moreover, a digital microscope (Hirox KH8700, River Edge, NJ, USA) was used to analyze the morphology of laser-treated regions. The 2D profile reconstruction of treated sample is shown in Figure 4a  Based on the preliminary investigation, S-SiC samples were laser treated according to the process parameters' combination reported in Table 3. Laser processing was performed in order to ablate each S-SiC cylinder realizing the desired diaphragm diameter and thickness. A picture of an example of laser-processed sample is reported in Figure 4a. In this sample, a 3 mm in diameter hole was realized up a depth of 1.9 mm, hence a membrane 100 µm thick, with a removal rate of about 3.1 mg/min, which is to be considered for micromachining operations [20,40]. Moreover, a digital microscope (Hirox KH8700, River Edge, NJ, USA) was used to analyze the morphology of lasertreated regions. The 2D profile reconstruction of treated sample is shown in Figure 4a (bottom). Based on the preliminary investigation, S-SiC samples were laser treated according to the process parameters' combination reported in Table 3. Laser processing was performed in order to ablate each S-SiC cylinder realizing the desired diaphragm diameter and thickness. A picture of an example of laser-processed sample is reported in Figure 4a. In this sample, a 3 mm in diameter hole was realized up a depth of 1.9 mm, hence a membrane 100 µm thick, with a removal rate of about 3.1 mg/min, which is to be considered for micromachining operations [20,40]. Moreover, a digital microscope (Hirox KH8700, River Edge, NJ, USA) was used to analyze the morphology of lasertreated regions. The 2D profile reconstruction of treated sample is shown in Figure 4a (bottom).  Number of repetition 1 50-75 -Laser fluence 2 ~20 J cm −2 1 Fifty repetitions were used for obtaining a 150 µm thick membrane, while 75 for a 100 µm thick membrane. 2 Defined as the ratio between the laser pulse energy and the laser beam spot area (see Table 2).

Low-Finesse Fabry-Perot Pressure-Sensor Structure and Measurement Set-Up
The schematic configuration of the fiber-optic Fabry-Perot pressure sensor is illustrated in Figure 5a. A single-mode fiber (SMF, SMPF0215-FC, with a mode field diameter = 10 µm @ 1550 nm; a cladding diameter = 125 µm; a coating diameter = 245 µm, by Thorlabs Inc., Newton, NJ, USA), is coupled to the laser-formed S-SiC membrane. FP cavity is represented by the fiber end-face and the inner surface of S-SiC diaphragm. As described in the previous section, S-SiC membrane is obtained by means of laser-ablation process of pristine cylindrical samples. As depicted in Figure 5a, laser light propagating into the SMF is partially reflected at the SMF facet (red arrow, A), whereas light transmitted at fiber output is reflected by the SiC diaphragm surface (orange arrow, B). Reflected beams propagating back into the fiber will generate an interference spectrum.

Low-Finesse Fabry-Perot Pressure-Sensor Structure and Measurement Set-Up
The schematic configuration of the fiber-optic Fabry-Perot pressure sensor is illustrated in Figure 5a. A single-mode fiber (SMF, SMPF0215-FC, with a mode field diameter = 10 µm @ 1550 nm; a cladding diameter = 125 µm; a coating diameter = 245 µm, by Thorlabs Inc., Newton, NJ, USA), is coupled to the laser-formed S-SiC membrane. FP cavity is represented by the fiber end-face and the inner surface of S-SiC diaphragm. As described in the previous section, S-SiC membrane is obtained by means of laser-ablation process of pristine cylindrical samples. As depicted in Figure 5a, laser light propagating into the SMF is partially reflected at the SMF facet (red arrow, A), whereas light transmitted at fiber output is reflected by the SiC diaphragm surface (orange arrow, B). Reflected beams propagating back into the fiber will generate an interference spectrum.
Laser-engraved S-SiC sample was fixed with epoxy to a 3 4 inches full brass male cap holder used for mounting in a high-pressure-chamber used for sensor characterization. To assure SMF-membrane alignment, the brass holder was machined to create a through-pass hole (Ø 1.8 mm) and a 0.5 mm deep engrave for sample housing. The SMF, terminated with a glass ferrule (1.8 mm in diameter), was then inserted into the though-pass hole and positioned in front of the diaphragm surface. Finally, fiber was fixed to the brass sensor holder with epoxy. Appl. Sci. 2020, 10, x FOR PEER REVIEW 7 of 13 (a) (b) Laser-engraved S-SiC sample was fixed with epoxy to a ¾ inches full brass male cap holder used for mounting in a high-pressure-chamber used for sensor characterization. To assure SMF-membrane alignment, the brass holder was machined to create a through-pass hole (Ø 1.8 mm) and a 0.5 mm deep engrave for sample housing. The SMF, terminated with a glass ferrule (1.8 mm in diameter), was then inserted into the though-pass hole and positioned in front of the diaphragm surface. Finally, fiber was fixed to the brass sensor holder with epoxy.
For samples with a hole diameter around 2 mm, the ferrule was inserted until it touched the side walls of the engraved S-SiC cylinder. As shown in Figure 4b, due to the V-groove shape of the lasertreated cylinder, SMF face-end membrane distance was around 100 µm, assuring a good collection of reflected beams. Conversely, for samples with hole larger than ferrule diameter a different approach has been adopted. In this case, ferrule-terminated fiber was inserted into the brass holder/S-SiC system until it touched the surface of the membrane. Then, with a micrometric handling system, the fiber head was removed by 100 µm and finally fixed with epoxy resin. An example of mounted SMF-terminated S-SiC sample is reported in Figure 5b.
The experimental arrangement for characterizing the S-SiC pressure sensor is shown in Figure  6. The optoelectronic system is composed of a fiber-coupled laser diode operating at 1550 nm, an InGaAs photo-diode, and a 3-ports optical circulator (LPSC-1550-FC, FGA01FC, and 6015-3, respectively, by Thorlabs Inc., Newton, NJ, USA). Laser light enters in port 1 of the circulator and is emitted toward S-SiC sensor whose SMF is connected to port 2. Reflected light, hence interference signal, exits from port 3 and it is detected by the fiber-coupled photodiode. A current-to-voltage converter (PDA200C, Thorlabs Inc., Newton, NJ, USA) and a digital voltmeter (Keithley 2700 series, Tektronix Inc., Beaverton, OR, USA) are used for interference light intensity acquisition. As described in [39], for sensor characterizations, a hydraulic table top test pump is used to impinge a stationary pressure on sensor membrane and measuring the inlet pressure with a digital test gauge connected to the pump reference port. For samples with a hole diameter around 2 mm, the ferrule was inserted until it touched the side walls of the engraved S-SiC cylinder. As shown in Figure 4b, due to the V-groove shape of the laser-treated cylinder, SMF face-end membrane distance was around 100 µm, assuring a good collection of reflected beams. Conversely, for samples with hole larger than ferrule diameter a different approach has been adopted. In this case, ferrule-terminated fiber was inserted into the brass holder/S-SiC system until it touched the surface of the membrane. Then, with a micrometric handling system, the fiber head was removed by 100 µm and finally fixed with epoxy resin. An example of mounted SMF-terminated S-SiC sample is reported in Figure 5b.
The experimental arrangement for characterizing the S-SiC pressure sensor is shown in Figure 6. The optoelectronic system is composed of a fiber-coupled laser diode operating at 1550 nm, an InGaAs photo-diode, and a 3-ports optical circulator (LPSC-1550-FC, FGA01FC, and 6015-3, respectively, by Thorlabs Inc., Newton, NJ, USA). Laser light enters in port 1 of the circulator and is emitted toward S-SiC sensor whose SMF is connected to port 2. Reflected light, hence interference signal, exits from port 3 and it is detected by the fiber-coupled photodiode. A current-to-voltage converter (PDA200C, Thorlabs Inc., Newton, NJ, USA) and a digital voltmeter (Keithley 2700 series, Tektronix Inc., Beaverton, OR, USA) are used for interference light intensity acquisition. As described in [39], for sensor characterizations, a hydraulic table top test pump is used to impinge a stationary pressure on sensor membrane and measuring the inlet pressure with a digital test gauge connected to the pump reference port.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 8 of 13 Figure 6. Picture of the measurement set-up used for S-SiC sensors characterizations at different values of inlet static pressure. On the right, a detail of the optoelectronic system used for interference signal measurement (see ref. [39] for more details).

Results and Discussion
Several S-SiC membranes have been fabricated with thickness ranging from 100 µm to 300 µm and diameter from 2.0 to 3.5 mm. Good results have been gained for membrane diameters around (2.4 ± 0.05) mm. Figure 7 shows experimental results for a (160 ± 10) µm thick membrane (S1 sample).

Results and Discussion
Several S-SiC membranes have been fabricated with thickness ranging from 100 µm to 300 µm and diameter from 2.0 to 3.5 mm. Good results have been gained for membrane diameters around (2.4 ± 0.05) mm. Figure 7 shows experimental results for a (160 ± 10) µm thick membrane (S1 sample). Figure 6. Picture of the measurement set-up used for S-SiC sensors characterizations at different values of inlet static pressure. On the right, a detail of the optoelectronic system used for interference signal measurement (see ref. [39] for more details).

Results and Discussion
Several S-SiC membranes have been fabricated with thickness ranging from 100 µm to 300 µm and diameter from 2.0 to 3.5 mm. Good results have been gained for membrane diameters around (2.4 ± 0.05) mm. Figure 7 shows experimental results for a (160 ± 10) µm thick membrane (S1 sample). As shown in Figure 7a, in the 0-10 MPa range photocurrent amplitude displays the expected behavior of an interference signal given by [41]: where d is the cavity length and λ (= 1550 nm) the wavelength of light emitted by laser-diode, whereas A and B are the intensities of light reflected at the fiber end-face and the inner diagram surface, As shown in Figure 7a, in the 0-10 MPa range photocurrent amplitude displays the expected behavior of an interference signal given by [41]: where d is the cavity length and λ (= 1550 nm) the wavelength of light emitted by laser-diode, whereas A and B are the intensities of light reflected at the fiber end-face and the inner diagram surface, respectively (see Figure 5a). The observed signal periodicity corresponds to a λ/2 diaphragm shift, i.e., 775 nm. Continuous red line represents best fit of data according to Equation (1). Up to a pressure of 1.5 MPa, a monotonic behavior of the photocurrent signal is observed, confirming the feasibility of the fabricated sensing element (see Figure 7b). In the 0-1 MPa range, the sensor shows a full scale range non-linearity lower than 4%. In order to evaluate sensor performances in terms of non-linearity and repeatability, according to Equation (1), fit of data reported in Figure 7 has been used to compile a look-up table (LUT) relating the photocurrent amplitude to the diaphragm deflection. Due to the non-monotonic nature of Equation (1), repeated measurements were performed increasing the pressure from 0 to about 10 MPa. Results are summarized in Figure 8a, where each point refers to pressure values recorded in correspondence of the photocurrent threshold values (i.e., membrane displacement) saved in the LUT. Obtained results displayed almost the same values at each test within the ±20 kPa (error bars for data in the figure are within the symbols). The sensor displays a good linearity, with a slope 0.163 nm/kPa (i.e., 16.3 nm/bar). In order to quantify the linearity error, the residuals between experimental and estimated values were evaluated by linear fit of data of Figure 8a. Results, reported in Figure 8b, depict that errors are lower than ±20 nm, i.e., ±0.6% on the investigated full scale range.
Equation (1), repeated measurements were performed increasing the pressure from 0 to about 10 MPa. Results are summarized in Figure 8a, where each point refers to pressure values recorded in correspondence of the photocurrent threshold values (i.e., membrane displacement) saved in the LUT. Obtained results displayed almost the same values at each test within the ±20 kPa (error bars for data in the figure are within the symbols). The sensor displays a good linearity, with a slope 0.163 nm/kPa (i.e., 16.3 nm/bar). In order to quantify the linearity error, the residuals between experimental and estimated values were evaluated by linear fit of data of Figure 8a. Results, reported in Figure 8b, depict that errors are lower than ±20 nm, i.e., ±0.6% on the investigated full scale range. The same characterization has been performed for a (260 ± 10) µm thick diaphragm (sample S2). Figure 9 summarizes the experimental results where membrane deflection Δd was evaluated at voltages corresponding to a phase shift of interference signal of 0°, 45°, 90°, 135°, and 180°. As expected, sensor sensitivity of 0.038 nm/kPa is lower than that found for the S1 thinner membrane. Moreover, in the investigated 0-30 MPa range, non-linearity is around 3% of the full scale. The same characterization has been performed for a (260 ± 10) µm thick diaphragm (sample S2). Figure 9 summarizes the experimental results where membrane deflection ∆d was evaluated at voltages corresponding to a phase shift of interference signal of 0 • , 45 • , 90 • , 135 • , and 180 • . As expected, sensor sensitivity of 0.038 nm/kPa is lower than that found for the S1 thinner membrane. Moreover, in the investigated 0-30 MPa range, non-linearity is around 3% of the full scale.
Appl. Sci. 2020, 10, x FOR PEER REVIEW 10 of 13 Figure 9. Diaphragm deflection as a function of pressure for a membrane 260 µm thick.
As depicted in Figure 5a, diaphragm is compressed towards the fiber end-face by the applied pressure. Being the optical fiber core diameter much lower than the diaphragm diameter, optical interference is mainly induced by the center deflection of membrane. Assuming a uniform thickness t for the membrane, under a uniformly distributed pressure difference ΔP between the two membrane surfaces, the deflection of the circular sensor diaphragm, with a radius r, at the center position is given by [42]: Figure 9. Diaphragm deflection as a function of pressure for a membrane 260 µm thick.
As depicted in Figure 5a, diaphragm is compressed towards the fiber end-face by the applied pressure. Being the optical fiber core diameter much lower than the diaphragm diameter, optical interference is mainly induced by the center deflection of membrane. Assuming a uniform thickness t for the membrane, under a uniformly distributed pressure difference ∆P between the two membrane surfaces, the deflection of the circular sensor diaphragm, with a radius r, at the center position is given by [42]: where E and ν are Young's modulus and Poisson's ratio for S-SiC, respectively. Experimental results reported in Figures 8 and 9 allow to evaluate an effective value for E/(1-ν 2 ) around 600 GPa. In particular, taking into account errors on diameter and thickness evaluation, a (610 ± 160) GPa value is found for S1 membrane, whereas a (570 ± 110) GPa value is obtained for S2 sample. In both the two cases, Young's modulus would be between the one declared for S-SiC sample used in this work (see Table 1) and what reported for crystalline silicon carbide [14], stating the existence of a phase change induced by laser treatment performed to engrave the specimen.

Conclusions
Pressure sensors are widely used in various areas. In addition, it is often necessary to have pressure sensors made of materials suitable for use in hostile environments for particular applications. Due to its peculiar characteristics SiC is a suitable material for the realization of pressure sensors to be used also in hostile areas. In this work, fiber optic coupled pressure sensors were manufactured and characterized using S-SiC membranes with diameter from 2.0 to 3.5 mm and thickness ranging from 100 µm to 300 µm, obtained through laser ablation of uncontaminated samples. Experimental results show good performance for membrane with diameter around (2.4 ± 0.05) mm. In particular, for the membrane thickness (160 ± 10) µm the experimental results show a monotonic change in the acquired amplitudes ∆I ph in the range 0-1 MPa, with a full scale non-linearity around 4%. A further characterization of the sensor has been performed in a wider pressure range (0-10 MPa) and the acquired photocurrent trend shows the expected sinusoidal trend of an interference signal. Moreover, the sensor shows good linearity (0.163 nm/kPa, i.e., 16.3 nm/bar), with linearity errors of less than ±20 nm, i.e., ±0.6% on the full scale investigated. For a second sample (S2 sample) with a (260 ± 10) µm thick diaphragm, the same characterization was carried out. In this case, non-linearity of about 3% of full scale was observed in the range 0-30 MPa and the sensitivity of the sensor (0.038 nm/kPa) is lower than that of the first sample, as expected. Moreover, experimental results allowed to estimate a Young's modulus between the value measured for pristine S-SiC sample and the one reported for crystalline SiC, stating a possible phase change induced by the laser treatment used to realize sensor's membrane.