- Email: [email protected]
Fabrication of a thin double-layer thermistor based on DVB-modified polymer-derived SiCN ceramics
Journal of Alloys and Compounds 732 (2018) 491e497
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Fabrication of a thin double-layer thermistor based on DVB-modified polymer-derived SiCN ceramics Baisheng Ma a, 1, Yejie Cao a, 1, Yan Gao b, Yiguang Wang a, * a b
Science and Technology on Thermostructural Composite Materials Laboratory, Northwestern Polytechnical University, Xi'an, Shaanxi, 710072 China State Key Laboratory of Traction Power, School of Mechanics and Engineering, Southwest Jiaotong University, Chengdu, Sichuan, 610031 China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 September 2017 Received in revised form 25 October 2017 Accepted 26 October 2017 Available online 27 October 2017
Polymer-derived ceramics (PDCs) are considered to be promising candidate materials for hightemperature sensor applications. Even though the PDCs have been studied for decades, the reliable PDC-based thin film sensors have not been produced, because ceramic films are too fragile to be handled without a substrate, but the huge shrinkage during polymer-to-ceramic conversion makes it difficult to fabricate PDC-based sensing elements on conventional substrates such as Al2O3. In this study, a thin polymer-derived ceramic double-layer thermistor was developed, using a layer with a lower resistivity as the sensing element and a layer with a higher resistivity as the substrate. The difficulty in PDC-based film sensing elements fabrication on a substrate was consequently resolved, because both the sensing elements and the substrate sustained a similar shrinkage during pyrolysis. The proposed double-layer thermistor was fabricated through step-growth photopolymerization and co-pyrolysis of two similar polymer precursors: commercially available polysilazane for the high-resistivity substrate and DVBmodified polysilazane for the low-resistivity sensing part. The DVB-modified polysilazane showed a similar shrinkage during pyrolysis, whereas the resistivity of the pyrolytic product was five orders of magnitude lower. The obtained thermistor exhibited a negative temperature coefficient in the temperature range from the room temperature to 650 C, and the resistance of the thermistor decreased smoothly as the temperature increased. The resistance-to-voltage conversion was conducted with a voltage divider equipped with a fixed resistor and the output voltage was observed to increase smoothly along with temperature. The current sensor configuration based on PDCs was reported for the first time, and is much closer to applicable sensors. © 2017 Elsevier B.V. All rights reserved.
Keywords: Polymer-derived ceramics Precursor modification Photopolymerization Negative temperature coefficient Thermistor
1. Introduction Polymer-derived silicon-based ceramics (PDCs), such as SiCN and SiBCN, are considered for high-temperature sensor applications, due to the corresponding good functional properties at high temperatures, such as semi-conducting behavior [1,2], and piezoresistivity [3], as well as the corresponding excellent chemical and mechanical properties, such as high thermal stability [4], good oxidation/corrosion resistance [5,6], and creep resistance in extreme environments [7]. Furthermore, the unique direct polymer-to-ceramic transition process allows for attractive patterning options, such as photopolymerization and soft
* Corresponding author. E-mail address: [email protected] (Y. Wang). 1 These authors contribute to the work equally. https://doi.org/10.1016/j.jallcom.2017.10.242 0925-8388/© 2017 Elsevier B.V. All rights reserved.
lithography, making PDCs a promising candidate material for hightemperature micro-electromechanical systems (MEMS) [8e10]. Temperature sensors are one promising application of PDCs (as a thermistor), built upon the corresponding semi-conducting behavior. Usually, a thermistor is required to be made quite small for high-response and high-sensitivity peculiarity to be achieved. Ceramics, although good candidate materials, are significantly fragile compared to other materials. Consequently, ceramic microsensor components are usually mounted to substrates to make sure that they are safe to handle. Though PDCs have been studied for decades, reliable PDC-based thin film sensors have not been produced, due to difficulty in PDC films fabrication on conventional substrates. This occurs due to the huge shrinkage of the PDCs (up to ~40%) during the polymer-to-ceramic transition process and the significant thermal stress that might occur between the substrate and the PDC film. One way to solve this problem is to eliminate the shrinkage, in
492
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
order to obtain a near-net-shape polymer-to-ceramic transition. This can, for instance, be accomplished through the addition of fillers to the polymer precursor, as demonstrated by Greil [11]. Jung et al. [12] successfully fabricated PDC-based resistance-temperature detector (RTD) arrays on quartz wafers through the addition of active fillers to the polysilazane. Still, with consideration to the resultant thermal stress from the various coefficients of thermal expansion, a safe bonding between the PDC component and the substrate could not be ensured through this approach, since the adhesion is not sufficient. Especially, this occurs when the device is operated at high temperatures. Another way is to totally alter the configuration of the conventional film ceramic sensors by changing the mounting substrate. As it was demonstrated earlier [13], one advantage of PDCs utilization, is that the properties of PDCs could vary over a wide range through the synthesis or modification of preceramic precursors with various compositions and structures. For instance, the electrical conductivity of PDCs prepared from various polymer precursors under various pyrolysis conditions could vary by up to 15 orders of magnitude. Taking advantage of this, PDCs with different resistivity values, whereas with similar shrinkage can be prepared. The PDCs with a low resistivity can be used as sensor elements, while the PDCs with a high resistivity as substrates. Through step-growth photopolymerization, the sensor element and the substrate can be cross-linked and solidified together, producing a whole body. Subsequently to pyrolysis, a microsensor component consisting of two different layers can consequently be obtained. These layers are connected together by chemical bonds and the corresponding similar coefficients of thermal expansion ensure the stability at high temperature. The two key elements of this strategy are: (1) the thermopyrolysis process, which requires the two precursors to sustain a similar shrinkage during the polymer-to-ceramic transition and shrink simultaneously, and (2) the resistivity, which requires the two derived ceramics to present a significant difference in resistivity. In previous studies, it was demonstrated that the free carbon played a quite important role in determining the electrical properties of PDCs [14,15]. The electrical conductivity could be adjusted through the content, distribution, and graphitization level alterations of the free carbon in the PDCs. To simplify the research, the conductivity of silicon carbonitride (SiCN) was selected to be increased, through the addition of a carbon source precursor to a commercially available polysilazane (PSN). The modified ceramics were utilized as sensing elements, whereas the unmodified ceramics were utilized as substrates for this strategy to be demonstrated. Divinylbenzene (DVB), as a widely utilized carbon source precursor [16,17], was used in this study. The effect of the addition of DVB on the thermal behavior of the modified polysilazane, which was of great importance to the double-layer structure synthesis, was characterized through Fourier transform infrared (FT-IR) spectroscopy, thermogravimetric analysis (TGA), as well as shrinkage analysis. The chemical composition and free carbon structure of the derived ceramics, which were crucial to the resistance-temperature dependence, were analyzed through elemental analysis and Raman spectroscopy, respectively. Two ceramic layers with similar shrinkage during pyrolysis, whereas with a significant difference in resistivity of up to 5 orders of magnitude, were obtained and could be fully pyrolyzed into a whole body without crack formation. Finally, in this study, a thin double-layer thermistor prototype was successfully fabricated with this step-growth photopolymerization strategy and tested from the room temperature (RT) to 650 C.
2. Experimental procedure 2.1. Materials and fabrication methods In this study, a commercially available polysilazane (PSN, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China) was selected as the SiCN precursor, DVB (Sigma-Aldrich, United States) was utilized as the carbon source to improve the conductivity of SiCN, and Phenylbis (2,4,6-trimethylbenzoyl)phosphine oxide (BAPO, Sigma-Aldrich, United States) was used as the photoinitiator. The effect of DVB amount on the shrinkage of modified precursor was studied in Ref. [17]. The results demonstrated that the shrinkage during pyrolysis increased as the DVB content increased. 20 wt% of DVB addition produced a similar shrinkage compared to the unmodified precursor, while 40 wt% of DVB addition or higher, resulted in significant difference in shrinkage. Therefore, 20 wt% of DVB was selected in this study. Firstly, 20 wt% of DVB was added to the PSN and magnetically stirred for ~2 h at room temperature at 400 rpm, to form a homogeneous mixture (labeled as PSN-DVB). Consequently, 4 wt% of BAPO was added to the PSN-DVB and the pure PSN, respectively, followed by magnetic stirring at 50 C, until it completely dissolved. The liquid mixtures were slightly yellowish and transparent prior to crosslinking. Following, the mixtures were placed in a vacuum oven for degassing. Two transparent glass plates covered with Teflon film were utilized for the photopolymerization of the mixtures. In this case, the Teflon film was used to permit an easy removal of the crosslinked polymer structures, because the adhesion between the polymer and the Teflon was quite weaker compared to glass. In brief, the degased mixtures were dropped onto one glass plate and the other plate was slowly lowered down. A gap of approximately 1 mm between the plates was controlled through metal strips insertion between the two glass plates and the mixtures were confined between the plates. The entire setup was subsequently exposed to a shuttered UV floodlight (IntelliRay 400, Uvitron International, Inc., United States) and the material was cross-linked for 1 h. Following cross-linking, the pure PSN preceramic plates were yellowish and transparent, while the PSN-DVB plates became slightly white and less transparent. To prepare the layered thermistor prototype, a pure PSN film of ~0.8 mm in thickness was firstly cross-linked, and consequently another layer of PSN-DVB of ~0.2 mm in thickness was cross-linked onto the surface of the pure PSN film. The thickness was controlled through the selection of metal strips with corresponding thicknesses to be inserted between the two glass plates. The cross-linked precursors were subsequently cut into individual samples with a surface area of roughly 20 mm 100 mm, covered with graphite papers on both sides and placed into an alumina boat for the subsequent pyrolysis. The pyrolysis was conducted under nitrogen flow at 1000 C for 4 h in a quartz tube furnace (GSL-1100X, MTI KJ GROUP, China), with a heating rate of 1 C/min. 2.2. Characterization methods The linear shrinkage of the samples subsequently to pyrolysis was calculated from the change in dimensions and the density was measured in water through the Archimedes' principle. The thermogravimetric (TG) curves of the cross-linked polymers were recorded over the temperature range from the RT to 1400 C with a Mettler-Toledo thermogravimetric analyzer (DSC 3þ), through the application of a heating rate of 5 C/min under nitrogen protection.
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
The ceramic yield was calculated from the TG curves. The Fouriertransform infrared (FT-IR) spectra of the raw materials and the cross-linked polymers were recorded with a Nicolet iS50 FT-IR Spectrometer (Thermo Fisher Scientific, United States). X-ray powder diffraction (XRD) was performed on the pyrolyzed ceramics with a Bruker D8 advance diffractometer (Germany), operated with Cu Ka radiation and the ceramics were discovered to be amorphous. The composition of the derived ceramics was determined through elemental analysis (vario EL cube, Elementar, Germany). Approximately 10 mg of a sample were used for the measurements. The mass percentage of hydrogen was below 0.5 wt % in all prepared ceramics and it was ignored. The silicon mass percentage was obtained from the difference to 100% of the sum of the mass percentages of carbon, nitrogen, and oxygen, under the assumption that no other elements were present. The free carbon in the ceramics was characterized with a confocal Raman microscope (inVia, Renishaw, UK), equipped with a 532 nm Si solid laser excitation source and a sensitive Peltier-cooled couple-charged device detector. Scanning electron microscopy (SEM, S-4700, Hitachi, Japan) was employed to observe the cross section of the prepared double-layer thermistor.
493
Fig. 2. Circuit configuration of the voltage divider for thermistor output voltage measurements.
2.3. Resistance tests and thermistor configuration To study the temperature dependence of the resistance, the ceramics were cut into samples of 7 mm 3.5 mm in dimensions and polished to a 1 mm finish. Comb-shaped Ag electrodes with a surface area of 2 mm 3.5 mm were consequently pasted and fired on the samples at 300 C in air. A schematic illustration of the setup is presented in Fig. 1. The electrical resistance of the samples was measured with a Keithley digital multimeter (Model 2002, United States) over the temperature range from the RT to 650 C under a protective Ar atmosphere utilizing a two-point probe configuration and a heating rate of 2 C/min. A K-type thermocouple connected to the multimeter was placed directly above the samples to monitor the temperature. The sample configuration and testing processes were the same for the pure PSN and the PSN-DVB-derived ceramics, for electrical resistance comparisons. The double-layer thermistor prototype with the same electrode configuration was prepared. The electrical resistance of the thermistor was converted into voltage with a voltage divider, equipped with a fixed resistor with a configuration as presented in Fig. 2. The input voltage was provided through a Keithley source meter unit (SourceMeter, Model 2450, United States) and the output voltage was recorded with the Model 2002 multimeter, along with temperature. The high temperature stability of the thermistor was studied through steady heating at 600 C for 160 h, for the resistance change of the thermistor to be observed. 3. Results and discussion 3.1. Material characterization Fig. 3 presents the FT-IR spectra, as recorded for the as-received
Fig. 1. Schematic illustration of the construction of the combed electrode on the samples.
Fig. 3. FT-IR spectra of (a) as-received PSN, (b) PSN after UV cross-linking, (c) the DVBmodified PSN after UV cross-linking, and (d) as-received DVB.
PSN and DVB, as well as for the photopolymerized PSN and PSNDVB mixture. For the as-received PSN, peaks corresponding to the SieNHeSi groups (at 3380 cm1 for NeH, at 1170 cm1 for Si2NeH), the vinylsilyl groups (H2C]CHSi, at 3048 cm1 for CeH, at 1593 cm1 for C]C), the SieCH3 groups (at 1254 and 1403 cm1 for SiCeH, and at 2958 and 2900 cm1 for the methyl vibrations), and the SieH bond (at 2128 cm1) were observed. The broad peaks between 640 and 1000 cm1 resulted from both the SieN and SieC bonds. After cross-linking, the peaks corresponding to the vinylsilyl groups almost disappeared and the other signals became weaker. This indicated that certain chemical reactions, such as vinyl polymerization, hydrosilylation, dehydrocoupling, and transamination, occurred during the photopolymerization process [18]. For the cross-linked PSN-DVB mixture, a similar FT-IR spectrum was observed, whereas certain signals from DVB were still visible, as presented in Fig. 3. This indicated that the cross-linking process was possibly accomplished through the vinyl polymerization between DVB and PSN and/or the self-polymerization of DVB. Fig. 4 presents the TG curves obtained for the cross-linked PSN and PSN-DVB. Both precursors sustained a similar pyrolysis process. Firstly, a low weight loss occurred over the interval from approximately 120 to 400 C, as a result of a further cross-linking of the precursors. Beyond 400 C, the polymer-to-ceramic transition process began, resulting in a comparatively high weight loss. The transition ended at approximately 800 C, beyond which, the weight was maintained almost constant up to the final temperature of 1400 C. It was apparent that the addition of DVB did not change the thermal pyrolysis process, whereas instead it resulted in a lower ceramic yield. The weight loss of the two precursors began and ended at almost the same temperatures, indicating that the
494
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
Fig. 4. TGA curves obtained for UV cross-linked precursors.
shrinkages during pyrolysis occurred simultaneously. The ceramic yields at 1000 C were measured from the TGA data to be 82 and 75% for the PSN and PSN-DVB, respectively, as shown in Table 1. Table 1 also presents the linear shrinkage rate and density comparisons of the samples. Both types of samples were observed to be nearly fully dense subsequently to pyrolysis, which are similar to the densities reported previously [19]. Even though the PSN-DVB precursor resulted in a lower ceramic yield and a lower density compared to the pure PSN precursor, the linear shrinkage rates of the ceramics derived from these two different types of precursors were similar which determines the success of the fabrication of the layered structure prepared through step-growth photopolymerization. Fig. 5 presents the temperature dependence of the resistance of the samples. Both types of ceramics had a negative temperature coefficient (NTC), i.e., the resistance decreased with increasing temperature. The resistance of the PSN-derived ceramics was almost five orders of magnitude higher than that of the PSN-DVBderived ceramics at RT, as well as three orders of magnitude higher at 650 C. The temperature dependence of the resistance of the PSN-DVB-derived ceramics was highly similar to that of the PSN-derived ceramics, demonstrating that addition of DVB did not significantly affect the temperature dependence of the resistance. To understand the difference in electrical conductivity of the materials with and without DVB modification, the elemental composition and the free carbon in the materials were characterized. The elemental composition of the samples is compared in Table 2. A fair amount of oxygen was observed in the samples, which is likely due to the BAPO and contamination by the air, during the photo cross-linking process. The empirical formula on a molar basis (SiCxNyOz) was determined through scaling with respect to silicon. The free carbon content was calculated with the approach proposed by Sorarù et al. [20], based on the fact that all the oxygen and nitrogen atoms were bonded to silicon [14,21]. The addition of DVB apparently increased the absolute and free carbon contents, compared to the pure PSN-derived SiCN. The free carbon
Fig. 5. Resistance of (a) PSN-derived and (b) PSN-DVB-derived ceramic samples at different temperatures.
Table 2 Elemental composition of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. Samples
Composition (wt%)
Empirical formula
Si
C
N
O
PSN PSN-DVB
51.03 46.54
22.71 30.63
19.31 16.96
6.95 5.87
Free carbon (wt%)
SiC1.04N0.76O0.24 SiC1.54N0.73O0.22
15.94 23.86
content increased from 15.9 to 23.9 wt% following the 20 wt% of DVB addition to the precursor. The free carbon in the samples was further analyzed through Raman spectroscopy. Fig. 6 displays the Raman spectra obtained for the ceramics pyrolyzed from PSN and PSN-DVB. The characteristic carbon Raman vibrations could be observed in all spectra: the D band corresponding to the disordered graphitic lattice centered at 1330 cm1, the G band attributed to the in-plane stretching motion of the symmetric sp2 carbon bond centered at 1610 cm1, the 2D band representing the second order of the D band centered at 2660 cm1, and the D þ G combination band centered at 2920 cm1 [22,23]. All Raman spectra showed a similar pattern, indicating that the DVB modification did not change the morphology of the free carbon, even though the concentration of free carbon apparently increased. The Raman spectra were modeled using a Lorentzian function for the D-peak and a BreiteWignereFano function (BWF)
Table 1 Linear shrinkage rate, ceramic yield, and density of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. Samples
Linear shrinkage rate (%)
Ceramic yield (%)
Density (g/cm3)
PSN PSN-DVB
25.42 ± 0.09 28.87 ± 1.57
81.98 ± 1.12 75.42 ± 1.97
2.01 ± 0.02 1.94 ± 0.05
Fig. 6. Raman spectra of (a) PSN-derived and (b) PSN-DVB-derived ceramics.
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
495
Table 3 Curve-fitting parameters from Raman spectra of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. FWHM (cm1)
Samples
Peak position (cm1) D peak
G peak
D peak
G peak
PSN PSN-DVB
1329 1328
1611 1606
128.54 139.36
66.65 76.87
I(D)/I(G)
La (nm)
1.98 1.76
1.79 1.68
for the G-peaks [24]. The peak position, the full-width-at-halfmaximum (FWHM) of the D peak and the G peak, as well as the IðDÞ=IðGÞ intensity ratio are listed in Table 3. The lateral size of the carbon clusters La was consequently calculated with the FerrarieRobertson equation [24]:
IðDÞ=IðGÞ ¼ C 0 ðlÞL2a
(1)
where, l is the wavelength of the excitation source and C 0 ðlÞ is a wavelength-dependent constant [25,26] and was calculated to be ~0.0062 for the wavelength of 5320 Å used in this experiment according to Ref. [24,26]. The lateral size comparisons of the carbon clusters are also presented in Table 3. The peak positions obtained for the two samples were almost identical, indicating that the free carbon phases in the two samples were similar in morphology. The addition of DVB to PSN caused an increase of the FWHM and a decrease of the IðDÞ=IðGÞ ratio (also the carbon cluster size) in the final ceramics. In previous studies [14,21], it was demonstrated that the free carbon in the polymer-derived SiCN ceramics underwent a graphitizing transition from amorphous carbon to nanocrystalline graphite during pyrolysis and that the carbon graphitizing transition was accompanied by a decrease of the FWHM of the Raman
Fig. 8. Temperature dependence of the resistance of the thermistor. The logarithm of the resistance versus the reciprocal of temperature is shown on the insert, revealing a nonlinear relationship.
peaks and an increase of the carbon cluster size [21,24]. The Raman results showed that the addition of DVB to the PSN precursor inhibited the graphitization process in its derived ceramic. Therefore, the difference in resistance was mainly caused by the higher free carbon content in the PSN-DVB-derived ceramics, even though the graphitization degree of the free carbon in the PSN-DVBderived ceramics was lower. Based on the difference in electrical resistance values between the PSN and the PSN-DVB derived ceramic films, a thermistor through step-growth photopolymerization was fabricated, as described above. 3.2. Thermistor characterization Fig. 7 presents the SEM micrograph of the cross section of the thermistor. To allow us to find the boundary between the two layers, the thermistor was fabricated in such a way that the PSNDVB precursor did not fully cover the PSN precursor during the step-growth photopolymerization process, in order for the boundary between the two layers to be easily determined at the edge of the thermistor. The SEM micrograph clearly showed that the two different precursors were fully co-pyrolyzed and formed a whole body without gaps at the boundary. No apparent difference was observed with respect to the micro-features between the two layers, although certain non-interconnected voids were observed under high magnification (Fig. 7b). The thickness of the PSN-DVBderived ceramic layer was approximately 100 mm. Fig. 8 presents the temperature dependence of the electrical resistance of the thermistor. The resistance of the thermistor gradually decreased as the temperature increased over the temperature range from RT to 650 C. The sensitivity coefficient aT of the NTC thermistor can be defined as:
aT ¼
1 vr B ¼ 2 r vT T
(2)
where, r is the resistivity of thermistor, T is the absolute temperature and B is a material constant. The value of B can be calculated through [27]:
ln B¼ Fig. 7. (a) SEM micrograph of the cross section at thermistor edge; (b) highmagnification image of the cross section. The dashed arrow line indicates the boundary between the two layers.
1 T1
R1 R2
T12
(3)
496
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
Fig. 9. Output voltage as a function of temperature of the thermistor for the first four runs. The overlapping curves indicated a good repeatability.
where, R1 and R2 are the resistance values at temperatures T1 and T2, respectively. The inset in Fig. 8 presents the logarithm of the resistance of the thermistor plotted versus the reciprocal of temperature. Unlike conventional metal-oxide thermistors, where B is a constant and the resistance can be represented through R ¼ AexpðB=TÞ, which means that the slope in the lnR - 1/T plot is constant, the lnR - 1/T plot was nonlinear for the PSN-DVB thermistor. As shown in the inset of Fig. 8, the value of B increased along with the temperature. The nonlinear behavior was also widely observed in the polymer derived ceramics, of which the conductivity followed a 3D hopping mechanism [1,2,14]. However, the average of B could be used to indicate the sensitivity. In practice, the value of B is specified between two standard temperatures of 25 C and 100 C [27]. The B25/100 value of the thermistor was calculated to be 960 K through Eq. (3), resulting in the sensitivity coefficient of the thermistor of 1.08% at 25 C and 0.11% at 650 C. The resistance to voltage conversion was conducted with a voltage divider equipped with a fixed resistor, with a configuration as shown in Fig. 2. The temperature of the thermistor was simultaneously recorded along with the output voltage. The resistance of the fixed resistor in this study was 300 kU, corresponding to the resistance of the thermistor at approximately 330 C, about the middle of the temperature range, to obtain a high voltage sensitivity. The input voltage was 10 V. The variation of the output voltage as a function of the temperature is presented in Fig. 9. The voltage was found to increase nonlinearly but smoothly along with temperature. The repeatability of the measurements was studied by recording the output voltage over different runs. The output voltage-response curves, as a function of the temperature recorded for different runs, were observed to overlap each other, indicating good repeatability. Fig. 9 displays the curves obtained for the first four runs. The high temperature stability of the thermistor was also studied through heating at 600 C for 160 h. The change in resistance of the thermistor was only 1.37%. Several PDC-based temperature sensors were reported in recent years. R. Zhao et al. [28] and Y. Yu et al. [29] reported the PDC-based temperature sensors fabricated by the conventional powder consolidation route. The sensors were in the form of bulk components of several millimeters in dimensions, which is not favorable when considering the response time and the sensors fabricated in this way were porous (at least 10% of porosity, usually can be 30% according to the described processes). This fabrication method was
not applicable to produce micro-sized sensors. Jung et al. [12] reported the PDC-based RTD arrays on quartz wafers through the addition of fillers to the precursors, to eliminate the shrinkage during the polymer-to-ceramic transitions. The bonding between the RTDs and the quartz wafers might not be reliable as no chemical bonds existed between them. The distribution of the solid fillers in the precursors was hard to be uniform, which was disadvantageous to the repeatability from sensor to sensor. D. Seo et al. [30] reported the bare PDC-based thin film RTDs without substrates. The bare thin film RTDs are considered to be fragile in real applications. Compared to previous reports, the thermistor reported in this study was in the form of dense thin film, firmly bonded to the substrates. The distribution of the liquid DVB could be quite uniform compared to solid fillers. This type of temperature sensors, based on PDCs, was reported for the first time and is quite applicable. However, the sensitivity coefficient of the thermistor reported in this work was slightly lower compared to the unmodified PDC-based sensors [28,29]. The decrease of the sensitivity was caused by the increase of the conductivity, whereas a similar result was reported for SiCNO-GO composite sensors [29]. Therefore, the discovery of methods to modify the precursor to improve the resistance of the substrate ceramics in this two-layer strategy was highly desirable. In this paper, the feasibility of the step-growth photopolymerization strategy to fabricate thin film temperature sensors from polymer-derived ceramics was demonstrated. The thickness of the sensing elements could be further reduced through improved precursor spraying methods, such as spin coating, and more work will be conducted in the future to improve this strategy. 4. Conclusions In this paper, a new configuration of PDC-based thin film temperature sensors was proposed, where the PDCs with a low resistivity were utilized as sensing elements, while the PDCs with high resistivity as substrates. The two key points of this strategy were: (1) the thermo-pyrolysis process, which requires the two precursors to undergo a similar shrinkage during the polymer-toceramic transition and shrink simultaneously, and (2) the resistivity, which requires the two derived ceramics to present a significant difference in resistivity. These could be accomplished through the synthesis of suitable polymer precursors or the proper modification of precursors. As a demonstration, 20 wt% of DVB was utilized to modify the PSN precursor in this study. The electrical resistance of the modified PSN derived ceramics was quite lower compared to the unmodified ceramics, whereas the shrinkage of the two ceramics during pyrolysis was similar. This made it possible for a thin ceramic film to be fabricated on top of another film without crack formation during pyrolysis. A thin thermistor was consequently successfully fabricated using a PSN derived ceramic as a substrate and a thin DVB-modified PSN derived ceramic as a sensing element, demonstrating that the proposed method was feasible. The thickness of the sensing element was approximately 100 mm and a lower thickness could probably be achieved through the improved precursor spraying methods. The thermistor showed a negative temperature coefficient up to 650 C, with a sensitivity coefficient of 1.08% at 25 C and 0.11% at 650 C. Furthermore, a resistance-to-voltage conversion was conducted and the output voltage was observed to increase smoothly along with temperature, demonstrating the high application potential of the prepared thermistor. Acknowledgments This work was supported by the Chinese Natural Science Foundation (Grant #51372202, #51602264, and #51532003) and
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
the project “111” (B08040). References [1] H. Ryu, Q. Wang, R. Raj, Ultrahigh-temperature semiconductors made from polymer-derived ceramics, J. Am. Ceram. Soc. 93 (2010) 1668e1676. [2] Y. Wang, T. Jiang, L. Zhang, L. An, Electron transport in polymer-derived amorphous silicon oxycarbonitride ceramics, J. Am. Ceram. Soc. 92 (2009) 1603e1606. [3] L. Zhang, Y. Wang, Y. Wei, W. Xu, D. Fang, L. Zhai, K.-C. Lin, L. An, A silicon carbonitride ceramic with anomalously high piezoresistivity, J. Am. Ceram. Soc. 91 (2008) 1346e1349. [4] R. Riedel, A. Kienzle, W. Dressler, L. Ruwisch, J. Bill, F. Aldinger, A silicoboron carbonotride ceramics stable to 2000 degrees C, Nature 382 (1996) 796e798. [5] Y. Wang, Y. Fan, L. Zhang, W. Zhang, L. An, Polymer-derived SiAlCN ceramics resist to oxidation at 1400 C, Scr. Mater. 55 (2006) 295e297. [6] Y. Wang, W. Fei, Y. Fan, L. Zhang, W. Zhang, L. An, A silicoaluminum carbonitride ceramic resist oxidation/corrosion in water vapour, J. Mater. Res. 21 (2006) 1625e1628. [7] R. Riedel, L.M. Ruwisch, L. An, R. Raj, Amorphous silicoboron carbonitride ceramics with anomalously high resistance to creep, J. Am. Ceram. Soc. 81 (1998) 3341e3344. [8] L.-A. Liew, Y. Liu, R. Luo, T. Cross, L. An, V.M. Bright, M.L. Dunn, J.W. Daily, R. Raj, Fabrication of SiCN MEMS by photopolymerization of pre-ceramic polymer, Sens. Actuators A Phys. 95 (2002) 120e134. [9] Y. Liu, L.-A. Liew, R. Luo, L. An, M.L. Dunn, V.M. Bright, J.W. Daily, R. Raj, Application of microforging to SiCN MEMS fabrication, Sens. Actuators A Phys. 95 (2002) 143e151. [10] D.-H. Lee, K.-H. Park, L.-Y. Hong, D.-P. Kim, SiCN ceramic patterns fabricated by soft lithography techniques, Sens. Actuators A Phys. 135 (2007) 895e901. [11] P. Greil, Active-filler-controlled pyrolysis of preceramic polymers, J. Am. Ceram. Soc. 78 (2005) 835e848. [12] S. Jung, D. Seo, S.J. Lombardo, Z.C. Feng, J.K. Chen, Y. Zhang, Fabrication using filler controlled pyrolysis and characterization of polysilazane PDC RTD arrays on quartz wafers, Sens. Actuators A Phys. 175 (2012) 53e59. [13] P. Colombo, G. Mera, R. Riedel, G.D. Sorarù, Polymer-derived ceramics: 40 years of research and innovation in advanced ceramics, J. Am. Ceram. Soc. 93 (2010) 1805e1837. [14] B. Ma, Y. Wang, K. Wang, X. Li, J. Liu, L. An, Frequency-dependent conductive behavior of polymer-derived amorphous silicon carbonitride, Acta Mater. 89 (2015) 215e224. [15] Y. Chen, F. Yang, L. An, On electric conduction of amorphous silicon
[16]
[17] [18]
[19]
[20]
[21]
[22]
[23]
[24] [25] [26]
[27] [28]
[29]
[30]
497
carbonitride derived from a polymeric precursor, Appl. Phys. Lett. 102 (2013) 231902. G. Liu, J. Kaspar, L.M. Reinold, M. Graczyk-Zajac, R. Riedel, Electrochemical performance of DVB-modified SiOC and SiCN polymer-derived negative electrodes for lithium-ion batteries, Electrochim. Acta 106 (2013) 101e108. K. Lu, D. Erb, M. Liu, Thermal stability and electrical conductivity of carbonenriched silicon oxycarbide, J. Mater. Chem. C 4 (2016) 1829e1837. N.S. Choong Kwet Yive, R.J.P. Corriu, D. Leclercq, P.H. Mutin, A. Vioux, Silicon carbonitride from polymeric precursors: thermal cross-linking and pyrolysis of oligosilazane model compounds, Chem. Mater. 4 (1992) 141e146. N. Janakiraman, F. Aldinger, Fabrication and characterization of fully dense SieCeN ceramics from a poly(ureamethylvinyl)silazane precursor, J. Eur. Ceram. Soc. 29 (2009) 163e173. G.D. Sorarù, S. Modena, E. Guadagnino, P. Colombo, J. Egan, C. Pantano, Chemical durability of silicon oxycarbide glasses, J. Am. Ceram. Soc. 85 (2002) 1529e1536. Y. Chen, X. Yang, Y. Cao, Z. Gan, L. An, Quantitative study on structural evolutions and associated energetics in polysilazane-derived amorphous silicon carbonitride ceramics, Acta Mater. 72 (2014) 22e31. M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cancado, A. Jorio, R. Saito, Studying disorder in graphite-based systems by Raman spectroscopy, Phys. Chem. Chem. Phys. 9 (2007) 1276e1290. A.C. Ferrari, Raman spectroscopy of graphene and graphite: disorder, electronephonon coupling, doping and nonadiabatic effects, Solid State Commun. 143 (2007) 47e57. A.C. Ferrari, J. Robertson, Interpretation of Raman spectra of disordered and amorphous carbon, Phys. Rev. B 61 (2000) 14095. Y. Wang, D.C. Alsmeyer, R.L. McCreery, Raman spectroscopy of carbon materials: structural basis of observed spectra, Chem. Mater. 2 (1990) 557e563. M.J. Matthews, M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, M. Endo, Origin of dispersive effects of the Raman D band in carbon materials, Phys. Rev. B 59 (1999) 6585e6588. A. Feteira, Negative temperature coefficient resistance (NTCR) ceramic thermistors: an industrial perspective, J. Am. Ceram. Soc. 92 (2009) 967e983. R. Zhao, G. Shao, Y. Cao, L. An, C. Xu, Temperature sensor made of polymerderived ceramics for high-temperature applications, Sens. Actuators A Phys. 219 (2014) 58e64. Y. Yu, Q. Huang, S. Rhodes, J. Fang, L. An, SiCNOeGO composites with the negative temperature coefficient of resistance for high-temperature sensor applications, J. Am. Ceram. Soc. 100 (2017) 592e601. D. Seo, S. Jung, S.J. Lombardo, Z.C. Feng, J.K. Chen, Y. Zhang, Fabrication and electrical properties of polymer-derived ceramic (PDC) thin films for hightemperature heat flux sensors, Sens. Actuators A Phys. 165 (2011) 250e255.
Contents lists available at ScienceDirect
Journal of Alloys and Compounds journal homepage: http://www.elsevier.com/locate/jalcom
Fabrication of a thin double-layer thermistor based on DVB-modified polymer-derived SiCN ceramics Baisheng Ma a, 1, Yejie Cao a, 1, Yan Gao b, Yiguang Wang a, * a b
Science and Technology on Thermostructural Composite Materials Laboratory, Northwestern Polytechnical University, Xi'an, Shaanxi, 710072 China State Key Laboratory of Traction Power, School of Mechanics and Engineering, Southwest Jiaotong University, Chengdu, Sichuan, 610031 China
a r t i c l e i n f o
a b s t r a c t
Article history: Received 15 September 2017 Received in revised form 25 October 2017 Accepted 26 October 2017 Available online 27 October 2017
Polymer-derived ceramics (PDCs) are considered to be promising candidate materials for hightemperature sensor applications. Even though the PDCs have been studied for decades, the reliable PDC-based thin film sensors have not been produced, because ceramic films are too fragile to be handled without a substrate, but the huge shrinkage during polymer-to-ceramic conversion makes it difficult to fabricate PDC-based sensing elements on conventional substrates such as Al2O3. In this study, a thin polymer-derived ceramic double-layer thermistor was developed, using a layer with a lower resistivity as the sensing element and a layer with a higher resistivity as the substrate. The difficulty in PDC-based film sensing elements fabrication on a substrate was consequently resolved, because both the sensing elements and the substrate sustained a similar shrinkage during pyrolysis. The proposed double-layer thermistor was fabricated through step-growth photopolymerization and co-pyrolysis of two similar polymer precursors: commercially available polysilazane for the high-resistivity substrate and DVBmodified polysilazane for the low-resistivity sensing part. The DVB-modified polysilazane showed a similar shrinkage during pyrolysis, whereas the resistivity of the pyrolytic product was five orders of magnitude lower. The obtained thermistor exhibited a negative temperature coefficient in the temperature range from the room temperature to 650 C, and the resistance of the thermistor decreased smoothly as the temperature increased. The resistance-to-voltage conversion was conducted with a voltage divider equipped with a fixed resistor and the output voltage was observed to increase smoothly along with temperature. The current sensor configuration based on PDCs was reported for the first time, and is much closer to applicable sensors. © 2017 Elsevier B.V. All rights reserved.
Keywords: Polymer-derived ceramics Precursor modification Photopolymerization Negative temperature coefficient Thermistor
1. Introduction Polymer-derived silicon-based ceramics (PDCs), such as SiCN and SiBCN, are considered for high-temperature sensor applications, due to the corresponding good functional properties at high temperatures, such as semi-conducting behavior [1,2], and piezoresistivity [3], as well as the corresponding excellent chemical and mechanical properties, such as high thermal stability [4], good oxidation/corrosion resistance [5,6], and creep resistance in extreme environments [7]. Furthermore, the unique direct polymer-to-ceramic transition process allows for attractive patterning options, such as photopolymerization and soft
* Corresponding author. E-mail address: [email protected] (Y. Wang). 1 These authors contribute to the work equally. https://doi.org/10.1016/j.jallcom.2017.10.242 0925-8388/© 2017 Elsevier B.V. All rights reserved.
lithography, making PDCs a promising candidate material for hightemperature micro-electromechanical systems (MEMS) [8e10]. Temperature sensors are one promising application of PDCs (as a thermistor), built upon the corresponding semi-conducting behavior. Usually, a thermistor is required to be made quite small for high-response and high-sensitivity peculiarity to be achieved. Ceramics, although good candidate materials, are significantly fragile compared to other materials. Consequently, ceramic microsensor components are usually mounted to substrates to make sure that they are safe to handle. Though PDCs have been studied for decades, reliable PDC-based thin film sensors have not been produced, due to difficulty in PDC films fabrication on conventional substrates. This occurs due to the huge shrinkage of the PDCs (up to ~40%) during the polymer-to-ceramic transition process and the significant thermal stress that might occur between the substrate and the PDC film. One way to solve this problem is to eliminate the shrinkage, in
492
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
order to obtain a near-net-shape polymer-to-ceramic transition. This can, for instance, be accomplished through the addition of fillers to the polymer precursor, as demonstrated by Greil [11]. Jung et al. [12] successfully fabricated PDC-based resistance-temperature detector (RTD) arrays on quartz wafers through the addition of active fillers to the polysilazane. Still, with consideration to the resultant thermal stress from the various coefficients of thermal expansion, a safe bonding between the PDC component and the substrate could not be ensured through this approach, since the adhesion is not sufficient. Especially, this occurs when the device is operated at high temperatures. Another way is to totally alter the configuration of the conventional film ceramic sensors by changing the mounting substrate. As it was demonstrated earlier [13], one advantage of PDCs utilization, is that the properties of PDCs could vary over a wide range through the synthesis or modification of preceramic precursors with various compositions and structures. For instance, the electrical conductivity of PDCs prepared from various polymer precursors under various pyrolysis conditions could vary by up to 15 orders of magnitude. Taking advantage of this, PDCs with different resistivity values, whereas with similar shrinkage can be prepared. The PDCs with a low resistivity can be used as sensor elements, while the PDCs with a high resistivity as substrates. Through step-growth photopolymerization, the sensor element and the substrate can be cross-linked and solidified together, producing a whole body. Subsequently to pyrolysis, a microsensor component consisting of two different layers can consequently be obtained. These layers are connected together by chemical bonds and the corresponding similar coefficients of thermal expansion ensure the stability at high temperature. The two key elements of this strategy are: (1) the thermopyrolysis process, which requires the two precursors to sustain a similar shrinkage during the polymer-to-ceramic transition and shrink simultaneously, and (2) the resistivity, which requires the two derived ceramics to present a significant difference in resistivity. In previous studies, it was demonstrated that the free carbon played a quite important role in determining the electrical properties of PDCs [14,15]. The electrical conductivity could be adjusted through the content, distribution, and graphitization level alterations of the free carbon in the PDCs. To simplify the research, the conductivity of silicon carbonitride (SiCN) was selected to be increased, through the addition of a carbon source precursor to a commercially available polysilazane (PSN). The modified ceramics were utilized as sensing elements, whereas the unmodified ceramics were utilized as substrates for this strategy to be demonstrated. Divinylbenzene (DVB), as a widely utilized carbon source precursor [16,17], was used in this study. The effect of the addition of DVB on the thermal behavior of the modified polysilazane, which was of great importance to the double-layer structure synthesis, was characterized through Fourier transform infrared (FT-IR) spectroscopy, thermogravimetric analysis (TGA), as well as shrinkage analysis. The chemical composition and free carbon structure of the derived ceramics, which were crucial to the resistance-temperature dependence, were analyzed through elemental analysis and Raman spectroscopy, respectively. Two ceramic layers with similar shrinkage during pyrolysis, whereas with a significant difference in resistivity of up to 5 orders of magnitude, were obtained and could be fully pyrolyzed into a whole body without crack formation. Finally, in this study, a thin double-layer thermistor prototype was successfully fabricated with this step-growth photopolymerization strategy and tested from the room temperature (RT) to 650 C.
2. Experimental procedure 2.1. Materials and fabrication methods In this study, a commercially available polysilazane (PSN, Institute of Chemistry, Chinese Academy of Sciences, Beijing, China) was selected as the SiCN precursor, DVB (Sigma-Aldrich, United States) was utilized as the carbon source to improve the conductivity of SiCN, and Phenylbis (2,4,6-trimethylbenzoyl)phosphine oxide (BAPO, Sigma-Aldrich, United States) was used as the photoinitiator. The effect of DVB amount on the shrinkage of modified precursor was studied in Ref. [17]. The results demonstrated that the shrinkage during pyrolysis increased as the DVB content increased. 20 wt% of DVB addition produced a similar shrinkage compared to the unmodified precursor, while 40 wt% of DVB addition or higher, resulted in significant difference in shrinkage. Therefore, 20 wt% of DVB was selected in this study. Firstly, 20 wt% of DVB was added to the PSN and magnetically stirred for ~2 h at room temperature at 400 rpm, to form a homogeneous mixture (labeled as PSN-DVB). Consequently, 4 wt% of BAPO was added to the PSN-DVB and the pure PSN, respectively, followed by magnetic stirring at 50 C, until it completely dissolved. The liquid mixtures were slightly yellowish and transparent prior to crosslinking. Following, the mixtures were placed in a vacuum oven for degassing. Two transparent glass plates covered with Teflon film were utilized for the photopolymerization of the mixtures. In this case, the Teflon film was used to permit an easy removal of the crosslinked polymer structures, because the adhesion between the polymer and the Teflon was quite weaker compared to glass. In brief, the degased mixtures were dropped onto one glass plate and the other plate was slowly lowered down. A gap of approximately 1 mm between the plates was controlled through metal strips insertion between the two glass plates and the mixtures were confined between the plates. The entire setup was subsequently exposed to a shuttered UV floodlight (IntelliRay 400, Uvitron International, Inc., United States) and the material was cross-linked for 1 h. Following cross-linking, the pure PSN preceramic plates were yellowish and transparent, while the PSN-DVB plates became slightly white and less transparent. To prepare the layered thermistor prototype, a pure PSN film of ~0.8 mm in thickness was firstly cross-linked, and consequently another layer of PSN-DVB of ~0.2 mm in thickness was cross-linked onto the surface of the pure PSN film. The thickness was controlled through the selection of metal strips with corresponding thicknesses to be inserted between the two glass plates. The cross-linked precursors were subsequently cut into individual samples with a surface area of roughly 20 mm 100 mm, covered with graphite papers on both sides and placed into an alumina boat for the subsequent pyrolysis. The pyrolysis was conducted under nitrogen flow at 1000 C for 4 h in a quartz tube furnace (GSL-1100X, MTI KJ GROUP, China), with a heating rate of 1 C/min. 2.2. Characterization methods The linear shrinkage of the samples subsequently to pyrolysis was calculated from the change in dimensions and the density was measured in water through the Archimedes' principle. The thermogravimetric (TG) curves of the cross-linked polymers were recorded over the temperature range from the RT to 1400 C with a Mettler-Toledo thermogravimetric analyzer (DSC 3þ), through the application of a heating rate of 5 C/min under nitrogen protection.
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
The ceramic yield was calculated from the TG curves. The Fouriertransform infrared (FT-IR) spectra of the raw materials and the cross-linked polymers were recorded with a Nicolet iS50 FT-IR Spectrometer (Thermo Fisher Scientific, United States). X-ray powder diffraction (XRD) was performed on the pyrolyzed ceramics with a Bruker D8 advance diffractometer (Germany), operated with Cu Ka radiation and the ceramics were discovered to be amorphous. The composition of the derived ceramics was determined through elemental analysis (vario EL cube, Elementar, Germany). Approximately 10 mg of a sample were used for the measurements. The mass percentage of hydrogen was below 0.5 wt % in all prepared ceramics and it was ignored. The silicon mass percentage was obtained from the difference to 100% of the sum of the mass percentages of carbon, nitrogen, and oxygen, under the assumption that no other elements were present. The free carbon in the ceramics was characterized with a confocal Raman microscope (inVia, Renishaw, UK), equipped with a 532 nm Si solid laser excitation source and a sensitive Peltier-cooled couple-charged device detector. Scanning electron microscopy (SEM, S-4700, Hitachi, Japan) was employed to observe the cross section of the prepared double-layer thermistor.
493
Fig. 2. Circuit configuration of the voltage divider for thermistor output voltage measurements.
2.3. Resistance tests and thermistor configuration To study the temperature dependence of the resistance, the ceramics were cut into samples of 7 mm 3.5 mm in dimensions and polished to a 1 mm finish. Comb-shaped Ag electrodes with a surface area of 2 mm 3.5 mm were consequently pasted and fired on the samples at 300 C in air. A schematic illustration of the setup is presented in Fig. 1. The electrical resistance of the samples was measured with a Keithley digital multimeter (Model 2002, United States) over the temperature range from the RT to 650 C under a protective Ar atmosphere utilizing a two-point probe configuration and a heating rate of 2 C/min. A K-type thermocouple connected to the multimeter was placed directly above the samples to monitor the temperature. The sample configuration and testing processes were the same for the pure PSN and the PSN-DVB-derived ceramics, for electrical resistance comparisons. The double-layer thermistor prototype with the same electrode configuration was prepared. The electrical resistance of the thermistor was converted into voltage with a voltage divider, equipped with a fixed resistor with a configuration as presented in Fig. 2. The input voltage was provided through a Keithley source meter unit (SourceMeter, Model 2450, United States) and the output voltage was recorded with the Model 2002 multimeter, along with temperature. The high temperature stability of the thermistor was studied through steady heating at 600 C for 160 h, for the resistance change of the thermistor to be observed. 3. Results and discussion 3.1. Material characterization Fig. 3 presents the FT-IR spectra, as recorded for the as-received
Fig. 1. Schematic illustration of the construction of the combed electrode on the samples.
Fig. 3. FT-IR spectra of (a) as-received PSN, (b) PSN after UV cross-linking, (c) the DVBmodified PSN after UV cross-linking, and (d) as-received DVB.
PSN and DVB, as well as for the photopolymerized PSN and PSNDVB mixture. For the as-received PSN, peaks corresponding to the SieNHeSi groups (at 3380 cm1 for NeH, at 1170 cm1 for Si2NeH), the vinylsilyl groups (H2C]CHSi, at 3048 cm1 for CeH, at 1593 cm1 for C]C), the SieCH3 groups (at 1254 and 1403 cm1 for SiCeH, and at 2958 and 2900 cm1 for the methyl vibrations), and the SieH bond (at 2128 cm1) were observed. The broad peaks between 640 and 1000 cm1 resulted from both the SieN and SieC bonds. After cross-linking, the peaks corresponding to the vinylsilyl groups almost disappeared and the other signals became weaker. This indicated that certain chemical reactions, such as vinyl polymerization, hydrosilylation, dehydrocoupling, and transamination, occurred during the photopolymerization process [18]. For the cross-linked PSN-DVB mixture, a similar FT-IR spectrum was observed, whereas certain signals from DVB were still visible, as presented in Fig. 3. This indicated that the cross-linking process was possibly accomplished through the vinyl polymerization between DVB and PSN and/or the self-polymerization of DVB. Fig. 4 presents the TG curves obtained for the cross-linked PSN and PSN-DVB. Both precursors sustained a similar pyrolysis process. Firstly, a low weight loss occurred over the interval from approximately 120 to 400 C, as a result of a further cross-linking of the precursors. Beyond 400 C, the polymer-to-ceramic transition process began, resulting in a comparatively high weight loss. The transition ended at approximately 800 C, beyond which, the weight was maintained almost constant up to the final temperature of 1400 C. It was apparent that the addition of DVB did not change the thermal pyrolysis process, whereas instead it resulted in a lower ceramic yield. The weight loss of the two precursors began and ended at almost the same temperatures, indicating that the
494
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
Fig. 4. TGA curves obtained for UV cross-linked precursors.
shrinkages during pyrolysis occurred simultaneously. The ceramic yields at 1000 C were measured from the TGA data to be 82 and 75% for the PSN and PSN-DVB, respectively, as shown in Table 1. Table 1 also presents the linear shrinkage rate and density comparisons of the samples. Both types of samples were observed to be nearly fully dense subsequently to pyrolysis, which are similar to the densities reported previously [19]. Even though the PSN-DVB precursor resulted in a lower ceramic yield and a lower density compared to the pure PSN precursor, the linear shrinkage rates of the ceramics derived from these two different types of precursors were similar which determines the success of the fabrication of the layered structure prepared through step-growth photopolymerization. Fig. 5 presents the temperature dependence of the resistance of the samples. Both types of ceramics had a negative temperature coefficient (NTC), i.e., the resistance decreased with increasing temperature. The resistance of the PSN-derived ceramics was almost five orders of magnitude higher than that of the PSN-DVBderived ceramics at RT, as well as three orders of magnitude higher at 650 C. The temperature dependence of the resistance of the PSN-DVB-derived ceramics was highly similar to that of the PSN-derived ceramics, demonstrating that addition of DVB did not significantly affect the temperature dependence of the resistance. To understand the difference in electrical conductivity of the materials with and without DVB modification, the elemental composition and the free carbon in the materials were characterized. The elemental composition of the samples is compared in Table 2. A fair amount of oxygen was observed in the samples, which is likely due to the BAPO and contamination by the air, during the photo cross-linking process. The empirical formula on a molar basis (SiCxNyOz) was determined through scaling with respect to silicon. The free carbon content was calculated with the approach proposed by Sorarù et al. [20], based on the fact that all the oxygen and nitrogen atoms were bonded to silicon [14,21]. The addition of DVB apparently increased the absolute and free carbon contents, compared to the pure PSN-derived SiCN. The free carbon
Fig. 5. Resistance of (a) PSN-derived and (b) PSN-DVB-derived ceramic samples at different temperatures.
Table 2 Elemental composition of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. Samples
Composition (wt%)
Empirical formula
Si
C
N
O
PSN PSN-DVB
51.03 46.54
22.71 30.63
19.31 16.96
6.95 5.87
Free carbon (wt%)
SiC1.04N0.76O0.24 SiC1.54N0.73O0.22
15.94 23.86
content increased from 15.9 to 23.9 wt% following the 20 wt% of DVB addition to the precursor. The free carbon in the samples was further analyzed through Raman spectroscopy. Fig. 6 displays the Raman spectra obtained for the ceramics pyrolyzed from PSN and PSN-DVB. The characteristic carbon Raman vibrations could be observed in all spectra: the D band corresponding to the disordered graphitic lattice centered at 1330 cm1, the G band attributed to the in-plane stretching motion of the symmetric sp2 carbon bond centered at 1610 cm1, the 2D band representing the second order of the D band centered at 2660 cm1, and the D þ G combination band centered at 2920 cm1 [22,23]. All Raman spectra showed a similar pattern, indicating that the DVB modification did not change the morphology of the free carbon, even though the concentration of free carbon apparently increased. The Raman spectra were modeled using a Lorentzian function for the D-peak and a BreiteWignereFano function (BWF)
Table 1 Linear shrinkage rate, ceramic yield, and density of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. Samples
Linear shrinkage rate (%)
Ceramic yield (%)
Density (g/cm3)
PSN PSN-DVB
25.42 ± 0.09 28.87 ± 1.57
81.98 ± 1.12 75.42 ± 1.97
2.01 ± 0.02 1.94 ± 0.05
Fig. 6. Raman spectra of (a) PSN-derived and (b) PSN-DVB-derived ceramics.
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
495
Table 3 Curve-fitting parameters from Raman spectra of ceramics pyrolyzed from pure PSN and PSN-DVB precursors. FWHM (cm1)
Samples
Peak position (cm1) D peak
G peak
D peak
G peak
PSN PSN-DVB
1329 1328
1611 1606
128.54 139.36
66.65 76.87
I(D)/I(G)
La (nm)
1.98 1.76
1.79 1.68
for the G-peaks [24]. The peak position, the full-width-at-halfmaximum (FWHM) of the D peak and the G peak, as well as the IðDÞ=IðGÞ intensity ratio are listed in Table 3. The lateral size of the carbon clusters La was consequently calculated with the FerrarieRobertson equation [24]:
IðDÞ=IðGÞ ¼ C 0 ðlÞL2a
(1)
where, l is the wavelength of the excitation source and C 0 ðlÞ is a wavelength-dependent constant [25,26] and was calculated to be ~0.0062 for the wavelength of 5320 Å used in this experiment according to Ref. [24,26]. The lateral size comparisons of the carbon clusters are also presented in Table 3. The peak positions obtained for the two samples were almost identical, indicating that the free carbon phases in the two samples were similar in morphology. The addition of DVB to PSN caused an increase of the FWHM and a decrease of the IðDÞ=IðGÞ ratio (also the carbon cluster size) in the final ceramics. In previous studies [14,21], it was demonstrated that the free carbon in the polymer-derived SiCN ceramics underwent a graphitizing transition from amorphous carbon to nanocrystalline graphite during pyrolysis and that the carbon graphitizing transition was accompanied by a decrease of the FWHM of the Raman
Fig. 8. Temperature dependence of the resistance of the thermistor. The logarithm of the resistance versus the reciprocal of temperature is shown on the insert, revealing a nonlinear relationship.
peaks and an increase of the carbon cluster size [21,24]. The Raman results showed that the addition of DVB to the PSN precursor inhibited the graphitization process in its derived ceramic. Therefore, the difference in resistance was mainly caused by the higher free carbon content in the PSN-DVB-derived ceramics, even though the graphitization degree of the free carbon in the PSN-DVBderived ceramics was lower. Based on the difference in electrical resistance values between the PSN and the PSN-DVB derived ceramic films, a thermistor through step-growth photopolymerization was fabricated, as described above. 3.2. Thermistor characterization Fig. 7 presents the SEM micrograph of the cross section of the thermistor. To allow us to find the boundary between the two layers, the thermistor was fabricated in such a way that the PSNDVB precursor did not fully cover the PSN precursor during the step-growth photopolymerization process, in order for the boundary between the two layers to be easily determined at the edge of the thermistor. The SEM micrograph clearly showed that the two different precursors were fully co-pyrolyzed and formed a whole body without gaps at the boundary. No apparent difference was observed with respect to the micro-features between the two layers, although certain non-interconnected voids were observed under high magnification (Fig. 7b). The thickness of the PSN-DVBderived ceramic layer was approximately 100 mm. Fig. 8 presents the temperature dependence of the electrical resistance of the thermistor. The resistance of the thermistor gradually decreased as the temperature increased over the temperature range from RT to 650 C. The sensitivity coefficient aT of the NTC thermistor can be defined as:
aT ¼
1 vr B ¼ 2 r vT T
(2)
where, r is the resistivity of thermistor, T is the absolute temperature and B is a material constant. The value of B can be calculated through [27]:
ln B¼ Fig. 7. (a) SEM micrograph of the cross section at thermistor edge; (b) highmagnification image of the cross section. The dashed arrow line indicates the boundary between the two layers.
1 T1
R1 R2
T12
(3)
496
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
Fig. 9. Output voltage as a function of temperature of the thermistor for the first four runs. The overlapping curves indicated a good repeatability.
where, R1 and R2 are the resistance values at temperatures T1 and T2, respectively. The inset in Fig. 8 presents the logarithm of the resistance of the thermistor plotted versus the reciprocal of temperature. Unlike conventional metal-oxide thermistors, where B is a constant and the resistance can be represented through R ¼ AexpðB=TÞ, which means that the slope in the lnR - 1/T plot is constant, the lnR - 1/T plot was nonlinear for the PSN-DVB thermistor. As shown in the inset of Fig. 8, the value of B increased along with the temperature. The nonlinear behavior was also widely observed in the polymer derived ceramics, of which the conductivity followed a 3D hopping mechanism [1,2,14]. However, the average of B could be used to indicate the sensitivity. In practice, the value of B is specified between two standard temperatures of 25 C and 100 C [27]. The B25/100 value of the thermistor was calculated to be 960 K through Eq. (3), resulting in the sensitivity coefficient of the thermistor of 1.08% at 25 C and 0.11% at 650 C. The resistance to voltage conversion was conducted with a voltage divider equipped with a fixed resistor, with a configuration as shown in Fig. 2. The temperature of the thermistor was simultaneously recorded along with the output voltage. The resistance of the fixed resistor in this study was 300 kU, corresponding to the resistance of the thermistor at approximately 330 C, about the middle of the temperature range, to obtain a high voltage sensitivity. The input voltage was 10 V. The variation of the output voltage as a function of the temperature is presented in Fig. 9. The voltage was found to increase nonlinearly but smoothly along with temperature. The repeatability of the measurements was studied by recording the output voltage over different runs. The output voltage-response curves, as a function of the temperature recorded for different runs, were observed to overlap each other, indicating good repeatability. Fig. 9 displays the curves obtained for the first four runs. The high temperature stability of the thermistor was also studied through heating at 600 C for 160 h. The change in resistance of the thermistor was only 1.37%. Several PDC-based temperature sensors were reported in recent years. R. Zhao et al. [28] and Y. Yu et al. [29] reported the PDC-based temperature sensors fabricated by the conventional powder consolidation route. The sensors were in the form of bulk components of several millimeters in dimensions, which is not favorable when considering the response time and the sensors fabricated in this way were porous (at least 10% of porosity, usually can be 30% according to the described processes). This fabrication method was
not applicable to produce micro-sized sensors. Jung et al. [12] reported the PDC-based RTD arrays on quartz wafers through the addition of fillers to the precursors, to eliminate the shrinkage during the polymer-to-ceramic transitions. The bonding between the RTDs and the quartz wafers might not be reliable as no chemical bonds existed between them. The distribution of the solid fillers in the precursors was hard to be uniform, which was disadvantageous to the repeatability from sensor to sensor. D. Seo et al. [30] reported the bare PDC-based thin film RTDs without substrates. The bare thin film RTDs are considered to be fragile in real applications. Compared to previous reports, the thermistor reported in this study was in the form of dense thin film, firmly bonded to the substrates. The distribution of the liquid DVB could be quite uniform compared to solid fillers. This type of temperature sensors, based on PDCs, was reported for the first time and is quite applicable. However, the sensitivity coefficient of the thermistor reported in this work was slightly lower compared to the unmodified PDC-based sensors [28,29]. The decrease of the sensitivity was caused by the increase of the conductivity, whereas a similar result was reported for SiCNO-GO composite sensors [29]. Therefore, the discovery of methods to modify the precursor to improve the resistance of the substrate ceramics in this two-layer strategy was highly desirable. In this paper, the feasibility of the step-growth photopolymerization strategy to fabricate thin film temperature sensors from polymer-derived ceramics was demonstrated. The thickness of the sensing elements could be further reduced through improved precursor spraying methods, such as spin coating, and more work will be conducted in the future to improve this strategy. 4. Conclusions In this paper, a new configuration of PDC-based thin film temperature sensors was proposed, where the PDCs with a low resistivity were utilized as sensing elements, while the PDCs with high resistivity as substrates. The two key points of this strategy were: (1) the thermo-pyrolysis process, which requires the two precursors to undergo a similar shrinkage during the polymer-toceramic transition and shrink simultaneously, and (2) the resistivity, which requires the two derived ceramics to present a significant difference in resistivity. These could be accomplished through the synthesis of suitable polymer precursors or the proper modification of precursors. As a demonstration, 20 wt% of DVB was utilized to modify the PSN precursor in this study. The electrical resistance of the modified PSN derived ceramics was quite lower compared to the unmodified ceramics, whereas the shrinkage of the two ceramics during pyrolysis was similar. This made it possible for a thin ceramic film to be fabricated on top of another film without crack formation during pyrolysis. A thin thermistor was consequently successfully fabricated using a PSN derived ceramic as a substrate and a thin DVB-modified PSN derived ceramic as a sensing element, demonstrating that the proposed method was feasible. The thickness of the sensing element was approximately 100 mm and a lower thickness could probably be achieved through the improved precursor spraying methods. The thermistor showed a negative temperature coefficient up to 650 C, with a sensitivity coefficient of 1.08% at 25 C and 0.11% at 650 C. Furthermore, a resistance-to-voltage conversion was conducted and the output voltage was observed to increase smoothly along with temperature, demonstrating the high application potential of the prepared thermistor. Acknowledgments This work was supported by the Chinese Natural Science Foundation (Grant #51372202, #51602264, and #51532003) and
B. Ma et al. / Journal of Alloys and Compounds 732 (2018) 491e497
the project “111” (B08040). References [1] H. Ryu, Q. Wang, R. Raj, Ultrahigh-temperature semiconductors made from polymer-derived ceramics, J. Am. Ceram. Soc. 93 (2010) 1668e1676. [2] Y. Wang, T. Jiang, L. Zhang, L. An, Electron transport in polymer-derived amorphous silicon oxycarbonitride ceramics, J. Am. Ceram. Soc. 92 (2009) 1603e1606. [3] L. Zhang, Y. Wang, Y. Wei, W. Xu, D. Fang, L. Zhai, K.-C. Lin, L. An, A silicon carbonitride ceramic with anomalously high piezoresistivity, J. Am. Ceram. Soc. 91 (2008) 1346e1349. [4] R. Riedel, A. Kienzle, W. Dressler, L. Ruwisch, J. Bill, F. Aldinger, A silicoboron carbonotride ceramics stable to 2000 degrees C, Nature 382 (1996) 796e798. [5] Y. Wang, Y. Fan, L. Zhang, W. Zhang, L. An, Polymer-derived SiAlCN ceramics resist to oxidation at 1400 C, Scr. Mater. 55 (2006) 295e297. [6] Y. Wang, W. Fei, Y. Fan, L. Zhang, W. Zhang, L. An, A silicoaluminum carbonitride ceramic resist oxidation/corrosion in water vapour, J. Mater. Res. 21 (2006) 1625e1628. [7] R. Riedel, L.M. Ruwisch, L. An, R. Raj, Amorphous silicoboron carbonitride ceramics with anomalously high resistance to creep, J. Am. Ceram. Soc. 81 (1998) 3341e3344. [8] L.-A. Liew, Y. Liu, R. Luo, T. Cross, L. An, V.M. Bright, M.L. Dunn, J.W. Daily, R. Raj, Fabrication of SiCN MEMS by photopolymerization of pre-ceramic polymer, Sens. Actuators A Phys. 95 (2002) 120e134. [9] Y. Liu, L.-A. Liew, R. Luo, L. An, M.L. Dunn, V.M. Bright, J.W. Daily, R. Raj, Application of microforging to SiCN MEMS fabrication, Sens. Actuators A Phys. 95 (2002) 143e151. [10] D.-H. Lee, K.-H. Park, L.-Y. Hong, D.-P. Kim, SiCN ceramic patterns fabricated by soft lithography techniques, Sens. Actuators A Phys. 135 (2007) 895e901. [11] P. Greil, Active-filler-controlled pyrolysis of preceramic polymers, J. Am. Ceram. Soc. 78 (2005) 835e848. [12] S. Jung, D. Seo, S.J. Lombardo, Z.C. Feng, J.K. Chen, Y. Zhang, Fabrication using filler controlled pyrolysis and characterization of polysilazane PDC RTD arrays on quartz wafers, Sens. Actuators A Phys. 175 (2012) 53e59. [13] P. Colombo, G. Mera, R. Riedel, G.D. Sorarù, Polymer-derived ceramics: 40 years of research and innovation in advanced ceramics, J. Am. Ceram. Soc. 93 (2010) 1805e1837. [14] B. Ma, Y. Wang, K. Wang, X. Li, J. Liu, L. An, Frequency-dependent conductive behavior of polymer-derived amorphous silicon carbonitride, Acta Mater. 89 (2015) 215e224. [15] Y. Chen, F. Yang, L. An, On electric conduction of amorphous silicon
[16]
[17] [18]
[19]
[20]
[21]
[22]
[23]
[24] [25] [26]
[27] [28]
[29]
[30]
497
carbonitride derived from a polymeric precursor, Appl. Phys. Lett. 102 (2013) 231902. G. Liu, J. Kaspar, L.M. Reinold, M. Graczyk-Zajac, R. Riedel, Electrochemical performance of DVB-modified SiOC and SiCN polymer-derived negative electrodes for lithium-ion batteries, Electrochim. Acta 106 (2013) 101e108. K. Lu, D. Erb, M. Liu, Thermal stability and electrical conductivity of carbonenriched silicon oxycarbide, J. Mater. Chem. C 4 (2016) 1829e1837. N.S. Choong Kwet Yive, R.J.P. Corriu, D. Leclercq, P.H. Mutin, A. Vioux, Silicon carbonitride from polymeric precursors: thermal cross-linking and pyrolysis of oligosilazane model compounds, Chem. Mater. 4 (1992) 141e146. N. Janakiraman, F. Aldinger, Fabrication and characterization of fully dense SieCeN ceramics from a poly(ureamethylvinyl)silazane precursor, J. Eur. Ceram. Soc. 29 (2009) 163e173. G.D. Sorarù, S. Modena, E. Guadagnino, P. Colombo, J. Egan, C. Pantano, Chemical durability of silicon oxycarbide glasses, J. Am. Ceram. Soc. 85 (2002) 1529e1536. Y. Chen, X. Yang, Y. Cao, Z. Gan, L. An, Quantitative study on structural evolutions and associated energetics in polysilazane-derived amorphous silicon carbonitride ceramics, Acta Mater. 72 (2014) 22e31. M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, L.G. Cancado, A. Jorio, R. Saito, Studying disorder in graphite-based systems by Raman spectroscopy, Phys. Chem. Chem. Phys. 9 (2007) 1276e1290. A.C. Ferrari, Raman spectroscopy of graphene and graphite: disorder, electronephonon coupling, doping and nonadiabatic effects, Solid State Commun. 143 (2007) 47e57. A.C. Ferrari, J. Robertson, Interpretation of Raman spectra of disordered and amorphous carbon, Phys. Rev. B 61 (2000) 14095. Y. Wang, D.C. Alsmeyer, R.L. McCreery, Raman spectroscopy of carbon materials: structural basis of observed spectra, Chem. Mater. 2 (1990) 557e563. M.J. Matthews, M.A. Pimenta, G. Dresselhaus, M.S. Dresselhaus, M. Endo, Origin of dispersive effects of the Raman D band in carbon materials, Phys. Rev. B 59 (1999) 6585e6588. A. Feteira, Negative temperature coefficient resistance (NTCR) ceramic thermistors: an industrial perspective, J. Am. Ceram. Soc. 92 (2009) 967e983. R. Zhao, G. Shao, Y. Cao, L. An, C. Xu, Temperature sensor made of polymerderived ceramics for high-temperature applications, Sens. Actuators A Phys. 219 (2014) 58e64. Y. Yu, Q. Huang, S. Rhodes, J. Fang, L. An, SiCNOeGO composites with the negative temperature coefficient of resistance for high-temperature sensor applications, J. Am. Ceram. Soc. 100 (2017) 592e601. D. Seo, S. Jung, S.J. Lombardo, Z.C. Feng, J.K. Chen, Y. Zhang, Fabrication and electrical properties of polymer-derived ceramic (PDC) thin films for hightemperature heat flux sensors, Sens. Actuators A Phys. 165 (2011) 250e255.