Polyimide Encapsulation of Spider-Inspired Crack-Based Sensors for Durability Improvement

In mechanical sensory systems, encapsulation is one of the crucial issues to take care of when it comes to protection of the systems from external damage. Recently, a new type of a mechanical strain sensor inspired by spider’s slit organ has been reported, which has incredibly high sensitivity, flexibility, wearability, and multifunctional sensing abilities. In spite of many of these advantages, the sensor is still vulnerable in harsh environments of liquids and/or high temperature, because it has heat-vulnerable polyethylene terephthalate (PET) substrate without any encapsulation layer. Here, we present a mechanical crack-based strain sensor with heat, water and saline solution resistance by alternating the substrate from polyester film to polyimide film and encapsulating the sensor with polyimide. We have demonstrated the ability of the encapsulated crack-based sensor against heat, water, saline solution damage through experiments. Our sensor exhibited reproducibility and durability with high sensitivity to strain (gauge factor above 10,000 at strain of two percent). These results show a new potential of the crack-based sensory system to be used as a wearable voice/motion/pulse sensing device and a high-temperature strain sensor.


Introduction
Sensor encapsulation is a crucial issue that needs to be taken care of in order to increase its applicability to various research fields.For example, bio-implantable devices, artificial skin, and wearable sensors need to endure wet and biochemical environments, and encapsulation of the device would guarantee the device performance [1][2][3][4][5][6][7][8][9][10][11][12].Recently, a new type of mechanosensor inspired by a spider's sensory organ system has been reported, which includes nanoscale cracks on a metallic layer over a polymer substrate [13].This spider-inspired sensor dramatically changes its electric resistance with the applied strain which widens the crack width.It also shows incredibly high sensitivity to strain with its extremely high gauge factor of over 2000 and vibration which is estimated to be measurable up to 10 nm amplitude [13].The gauge factor (GF)-one of the most important indicators of the sensitivity of strain sensors-is defined as GF = (∆R/R 0 )/ε.Where ∆R is the resistance change with deformation, R 0 is the resistance before deformation, and ε is the applied deformation.Moreover, it has diverse advantages including flexibility, wearability, and multifunctional sensing abilities [13][14][15][16][17].However, the durability of the sensor in harsh environments of liquids and/or high temperature is Appl.Sci.2018, 8, 367 2 of 8 still a challenging issue.Sensor systems are usually exposed to various surrounding environments including different surrounding liquid, humidity and temperature, which lead to damage of the sensory system and shortening of the sensor's life.To solve this matter, we have demonstrated the extended capability of the crack sensory system by replacing the substrate from polyethylene terephthalate (PET) with polyimide (PI) and encapsulating the sensor with PI.Because the polyimide is widely used in flexible electronics or sensor systems due to its high durability and flexibility [2,18,19], there are various advantages in applying our crack-based sensor system.Furthermore, Jonathan Viventi et al. has developed a device encapsulated with polyimide, demonstrating new opportunities for the sensors to be applied in clinical fields [20,21].The encapsulated crack sensor can be used in diverse applications such as a wearable voice/pulse/motion sensor, strain gauge sensor at high temperature, vibration monitoring sensor for identifying internal damage to bridges, ultra-sensitive pressure/force sensor for surgical devices, and skin attachable health monitoring sensor.

The Fabrication of an Encapsulated Thin Metal Layered Crack Sensor
The basic features of the encapsulated crack sensor are demonstrated in Figure 1a.The sensor was 5 mm in width and 50 mm in length, and the gauge length was 30 mm.It was composed of five layers: 7.5 µm thick PI film (3022-5 Kapton thin film, Chemplex, Palm City, FL, USA) or 6 µm thick PET film (3026 Mylar thin film, Chemplex, Palm City, FL, USA) as the bottom substrate layer, 5 nm thick molybdenum trioxide (MoO 3 ) as an adhesion layer, 50 nm thick chromium (Cr) as a crack inducing layer, 30 nm thick gold (Au) for an electrical conductor, and lastly 650 nm thick cured polyimide film (Poly(pyromellitic dianhydride-co-4,4 -oxydianiline), amic acid solution, Sigma-Aldrich, St. Louis, MO, USA ) on the top layer for the encapsulation of the total system.The PI precursor was spin-coated on the cracked metal films and it was cured on a hot plate in the N 2 atmosphere (80 • C for 30 min and 290 • C for 30 min).MoO 3 is known as an adhesion layer between metal and polymer [17], and we simply tested it with "scotch tape test" [22].It was confirmed that no defect occurred when the scotch tape was attached and detached on the metal surface (Figure S1 †).The thin film layers (MoO 3 , Cr, and Au) on the polymer substrate are deposited by the thermal evaporation system (Thermal Evaporation System, DD High Tech.Co., Gimpo-si, Gyeonggi-do, Korea).The thickness of deposited metals and polymers are measured by AFM (Atomic Force Microscopy) measurement.
temperature is still a challenging issue.Sensor systems are usually exposed to various surrounding environments including different surrounding liquid, humidity and temperature, which lead to damage of the sensory system and shortening of the sensor's life.To solve this matter, we have demonstrated the extended capability of the crack sensory system by replacing the substrate from polyethylene terephthalate (PET) with polyimide (PI) and encapsulating the sensor with PI.Because the polyimide is widely used in flexible electronics or sensor systems due to its high durability and flexibility [2,18,19], there are various advantages in applying our crack-based sensor system.Furthermore, Jonathan Viventi et al. has developed a device encapsulated with polyimide, demonstrating new opportunities for the sensors to be applied in clinical fields [20,21].The encapsulated crack sensor can be used in diverse applications such as a wearable voice/pulse/motion sensor, strain gauge sensor at high temperature, vibration monitoring sensor for identifying internal damage to bridges, ultra-sensitive pressure/force sensor for surgical devices, and skin attachable health monitoring sensor.

The Fabrication of an Encapsulated Thin Metal Layered Crack Sensor
The basic features of the encapsulated crack sensor are demonstrated in Figure 1a.The sensor was 5 mm in width and 50 mm in length, and the gauge length was 30 mm.It was composed of five layers: 7.5 μm thick PI film (3022-5 Kapton thin film, Chemplex, Palm City, FL, USA) or 6 μm thick PET film (3026 Mylar thin film, Chemplex, Palm City, FL, USA) as the bottom substrate layer, 5 nm thick molybdenum trioxide (MoO3) as an adhesion layer, 50 nm thick chromium (Cr) as a crack inducing layer, 30 nm thick gold (Au) for an electrical conductor, and lastly 650 nm thick cured polyimide film (Poly(pyromellitic dianhydride-co-4,4′-oxydianiline), amic acid solution, Sigma-Aldrich, St. Louis, MO, USA ) on the top layer for the encapsulation of the total system.The PI precursor was spin-coated on the cracked metal films and it was cured on a hot plate in the N2 atmosphere (80 °C for 30 min and 290 °C for 30 min).MoO3 is known as an adhesion layer between metal and polymer [17], and we simply tested it with "scotch tape test" [22].It was confirmed that no defect occurred when the scotch tape was attached and detached on the metal surface (Figure S1 †).The thin film layers (MoO3, Cr, and Au) on the polymer substrate are deposited by the thermal evaporation system (Thermal Evaporation System, DD High Tech.Co., Gimpo-si, Gyeonggi-do, Korea).The thickness of deposited metals and polymers are measured by AFM (Atomic Force Microscopy) measurement.

Thermal, Water, and Saline Solution Resistance Test
To verify the performance of the device to detect vibration and strain in harsh environments such as liquid environment or a steam turbine blade, we tested the sensors in three different conditions (at high temperature, in water, and in saline solution).In these conditions, we also compared the sensors with three different types of substrates: sensors with a PET substrate without encapsulation, a PI substrate without encapsulation, and an encapsulated PI substrate.Each sensor was heated on a hot plate for 10 min at 100, 200 and 300 • C and checked for each gauge factor by 50 cycles of straining from 0% to 2%.For water and saline solution resistance tests, the sensors were dipped into water or saline solution for a certain period of time (1, 10, 30 h for water resistance test and 12, 24, 36 h for saline solution resistance test) and checked for the gauge factor by 50 cycles of straining from 0% to 2% after being dried out in the ambient condition.

Measuring Resistance Variation
Figure S2 shows the experimental setup for measuring resistance variation by straining of the sensor.The sensors were stretched from 0% to 2% strains up to 15,000 times repeatedly with the sweeping speed of 20 mm/min to measure the durability by using a material testing machine (3342 UTM, Instron Co., Norwood, MA, USA).The variation of resistance of the metal layered crack sensors was measured simultaneously by Lab View-based data acquisition system (PXI-4071, National Instruments Inc., Austin, TX, USA).Standard deviation of the measurement when measuring 2 wire resistance with 5 1/2 digit was 0.01 Ω.

Results and Discussion
In general, metal layers on the crack lips can be delaminated by repetitive stretching, heating, and in water or chemical environments [23][24][25][26][27][28][29][30][31][32].From repeated exposure to these environments, a sensor can have stress concentration on the edge due to different Poisson's ratio or thermal expansion mismatch [23,25,33].A water or chemical environment causes degradation of the adhesion force between the metal film and the polymer substrate [32].To measure the durability of our crack sensors, we have performed three kinds of experimental tests: thermal resistance test, water and saline solution resistance test, and marathon test.
In order to protect the crack sensor from the harsh environment and improve the durability, we encapsulated the crack-based sensor with PI.The schematic illustration of PI encapsulated crack-based sensor is presented in Figure 1a.The crack on the thin films was generated by stretching them at two-percent strain before encapsulation (Figure 1b left).Additionally, the polyimide precursor solution was spin-coated on the thin cracked metal film.The cross-sectional SEM image (Figure 1b right) shows that the solution does not penetrate into the gap of a crack because the gap was too narrow (10-15 nm). Figure 1c illustrates the cracks generated in the transverse direction to the extension force, and the crack gap increases with applied strain as shown in Figure 1d.The gauge factor of the encapsulated sensor reached above 10,000 at a strain of two percent with high reproducibility and durability.
The results in Figure 2 show that the thermal, water and saline solution durability of the crack sensor have been greatly improved with encapsulation by cured PI.In the thermal resistance test (Figure 2a), the PET substrate of the crack sensor maintained its gauge factor for about 90% at 200 • C because the deformation of the polymer surface reduced the stress concentration at the metal-polymer interface by going above glass-transition temperature (T g = 85 • C) [34].However, it starts to break down when it is heated up to 300 • C, exceeding the melting temperature of the PET (T m = 290 • C) [35].In the case of the PI substrate crack sensor without encapsulation, the relative GF rapidly decreased by heating, although the PI substrate was supposed to endure at the temperature above 300 • C.This was due to the delamination caused by the different thermal expansion coefficient and weak adhesive force between the metal layer and the polymer substrate [36].However, for the encapsulated crack sensor, the gauge factor remained at 77% even after heating at 300 • C. The top encapsulating layer was responsible for enduring the heat treatment, which prevents the delamination of the metal layer from the substrate.
Appl.Sci.2018, 8, x 4 of 8 encapsulated crack sensor, the gauge factor remained at 77% even after heating at 300 °C.The top encapsulating layer was responsible for enduring the heat treatment, which prevents the delamination of the metal layer from the substrate.The graphs in Figure 2b clearly show that sensors without encapsulation are very vulnerable in liquid environment.The metal layer of the PET substrate crack sensor delaminated as soon as it is immersed in water, and the PI substrate crack sensor also cannot resist liquid.However, the PI encapsulated crack sensor could endure 30 h in water and 24 h in 0.9% saline water with the gauge factor of two-percent strain even though the relative gauge factor decreased to 50%.However, we could see that straining two percent in a state where the adhesion force was degraded due to wet conditions accelerated degradation of its gauge factor.Figure S3 shows that when the strain range was from zero to one percent, the sensor maintains its GF of 90% or more for 130 h, and Figure S4 shows that there was no base resistance change for 190 h without straining.Because the actual operating range of the sensor would not be higher than one percent, it could endure longer in a wet environment.
Figure 2c, PI is also verified as a very good passivation layer compared with PDMS (PDMS: polydimethylsiloxane) by its water permeability.Compared to a sensor encapsulated with a PDMS of 30 μm thickness, it showed superior performance in saline solution.The major cause of GF decreases in liquid environment is delamination.In Figure S5, the PI or PDMS encapsulated crack sensor immersed in saline solution shows that the metal layer is separated from the substrate, which indicates an increase of the base line.However, the delamination issue in these environments can be enhanced by a thicker encapsulation layer.Figure 3c shows that if the thickness of the encapsulation layer is increased to 10 μm, it can last more than 100 h in saline solution, but it must be appropriately adjusted because it is in a trade-off relationship with the sensitivity of the sensor (Figure 3a,b).These results show a large expandability of the crack sensor in a humid environment, such as a wearable device.The graphs in Figure 2b clearly show that sensors without encapsulation are very vulnerable in liquid environment.The metal layer of the PET substrate crack sensor delaminated as soon as it is immersed in water, and the PI substrate crack sensor also cannot resist liquid.However, the PI encapsulated crack sensor could endure 30 h in water and 24 h in 0.9% saline water with the gauge factor of two-percent strain even though the relative gauge factor decreased to 50%.However, we could see that straining two percent in a state where the adhesion force was degraded due to wet conditions accelerated degradation of its gauge factor.Figure S3 shows that when the strain range was from zero to one percent, the sensor maintains its GF of 90% or more for 130 h, and Figure S4 shows that there was no base resistance change for 190 h without straining.Because the actual operating range of the sensor would not be higher than one percent, it could endure longer in a wet environment.
Figure 2c, PI is also verified as a very good passivation layer compared with PDMS (PDMS: polydimethylsiloxane) by its water permeability.Compared to a sensor encapsulated with a PDMS of 30 µm thickness, it showed superior performance in saline solution.The major cause of GF decreases in liquid environment is delamination.In Figure S5, the PI or PDMS encapsulated crack sensor immersed in saline solution shows that the metal layer is separated from the substrate, which indicates an increase of the base line.However, the delamination issue in these environments can be enhanced by a thicker encapsulation layer.Figure 3c shows that if the thickness of the encapsulation layer is increased to 10 µm, it can last more than 100 h in saline solution, but it must be appropriately adjusted because it is in a trade-off relationship with the sensitivity of the sensor (Figure 3a,b).These results show a large expandability of the crack sensor in a humid environment, such as a wearable device.
Appl.Sci.2018, 8, x 4 of 8 encapsulated crack sensor, the gauge factor remained at 77% even after heating at 300 °C.The top encapsulating layer was responsible for enduring the heat treatment, which prevents the delamination of the metal layer from the substrate.The graphs in Figure 2b clearly show that sensors without encapsulation are very vulnerable in liquid environment.The metal layer of the PET substrate crack sensor delaminated as soon as it is immersed in water, and the PI substrate crack sensor also cannot resist liquid.However, the PI encapsulated crack sensor could endure 30 h in water and 24 h in 0.9% saline water with the gauge factor of two-percent strain even though the relative gauge factor decreased to 50%.However, we could see that straining two percent in a state where the adhesion force was degraded due to wet conditions accelerated degradation of its gauge factor.Figure S3 shows that when the strain range was from zero to one percent, the sensor maintains its GF of 90% or more for 130 h, and Figure S4 shows that there was no base resistance change for 190 h without straining.Because the actual operating range of the sensor would not be higher than one percent, it could endure longer in a wet environment.
Figure 2c, PI is also verified as a very good passivation layer compared with PDMS (PDMS: polydimethylsiloxane) by its water permeability.Compared to a sensor encapsulated with a PDMS of 30 μm thickness, it showed superior performance in saline solution.The major cause of GF decreases in liquid environment is delamination.In Figure S5, the PI or PDMS encapsulated crack sensor immersed in saline solution shows that the metal layer is separated from the substrate, which indicates an increase of the base line.However, the delamination issue in these environments can be enhanced by a thicker encapsulation layer.Figure 3c shows that if the thickness of the encapsulation layer is increased to 10 μm, it can last more than 100 h in saline solution, but it must be appropriately adjusted because it is in a trade-off relationship with the sensitivity of the sensor (Figure 3a,b).These results show a large expandability of the crack sensor in a humid environment, such as a wearable device.Figure 4 shows durability (marathon test) performance of the sensors up to 15,000 cycles.First, the crack sensor on PET substrate without the encapsulating layer lost its gauge factor rapidly by 20% from the first 1000 to 15,000 cycles of repetitive stretching (Figure 4a).Second, the GF of the crack sensor on PI substrate without the encapsulating layer increased until 3000 cycles and started to decrease gradually until 7000 cycles, and it became 30% of maximum GF at 15,000 cycles (Figure 4b).The reason why the sensitivity increased first and then decreased was that deepening of crack depth by penetration and incomplete recovery of crack gap by residual stress occurred simultaneously.When a crack penetrates into the substrate, sensitivity increases by deepening the crack depth [16].However, repetitive straining accumulates residual stress on the crack vertex, disturbing closing of the crack gap, causing the gauge factor to decrease [37].Third, the crack sensor on PI substrate with the encapsulating layer had lower sensitivity than that of the crack sensor on PI substrate without encapsulation, but sensitivity drop with repeated tensile load improved slightly.It maintains its GF above 65% until 11,000 cycles (Figure 4c).These results show that the encapsulating layer contributes to delay of the progress of the crack.Figure 4 shows durability (marathon test) performance of the sensors up to 15,000 cycles.First, the crack sensor on PET substrate without the encapsulating layer lost its gauge factor rapidly by 20% from the first 1000 to 15,000 cycles of repetitive stretching (Figure 4a).Second, the GF of the crack sensor on PI substrate without the encapsulating layer increased until 3000 cycles and started to decrease gradually until 7000 cycles, and it became 30% of maximum GF at 15,000 cycles (Figure 4b).The reason why the sensitivity increased first and then decreased was that deepening of crack depth by penetration and incomplete recovery of crack gap by residual stress occurred simultaneously.When a crack penetrates into the substrate, sensitivity increases by deepening the crack depth [16].However, repetitive straining accumulates residual stress on the crack vertex, disturbing closing of the crack gap, causing the gauge factor to decrease [37].Third, the crack sensor on PI substrate with the encapsulating layer had lower sensitivity than that of the crack sensor on PI substrate without encapsulation, but sensitivity drop with repeated tensile load improved slightly.It maintains its GF above 65% until 11,000 cycles (Figure 4c).These results show that the encapsulating layer contributes to delay of the progress of the crack.The resistance of the metal films on the polymer substrate dramatically increases due to the crack gap opening by stretching it up to two percent.To use it as a sensor, the reverse process should return the initial crack gap and its initial resistance as well.Figure 5a-c show that a 10-cycle set of strains from zero to two percent at a sweeping speed of 20 mm/min with one second of the interval was nearly identical.The linearly fitted gauge factor of crack sensor on PET substrate with the strain range from zero to two percent by regression analysis was ~2800, that of the crack sensor on PI substrate was 28,000, and that of the PI encapsulated crack sensor was 12,500.Interestingly, it shows that by only changing substrates from PET to PI, sensitivity increased up to 10 times.We assume that this phenomenon was caused by crack penetration into the PI substrate because the sensitivity of the sensor increases as the depth of crack increases [16].Figure S6 † shows the change in resistance of three samples when applied strain from zero to two percent.The result demonstrates that the change in resistance to the applied strain of the PI encapsulated crack sensor was more linear than the other sensors.However, the PI encapsulation process induced reduced gauge factor of ~45%. Figure 5d-f each shows a graph of the averaged normalized resistance of three different samples along with an error bar (n = 5).Black square marks are for the loading to the strain of two percent, and red circle marks are for unloading.The response time of the sensor was about five milliseconds (Figure S7 †).It shows that these three different types of sensors had quite good reproducibility without hysteresis.The resistance of the metal films on the polymer substrate dramatically increases due to the crack gap opening by stretching it up to two percent.To use it as a sensor, the reverse process should return the initial crack gap and its initial resistance as well.Figure 5a-c show that a 10-cycle set of strains from zero to two percent at a sweeping speed of 20 mm/min with one second of the interval was nearly identical.The linearly fitted gauge factor of crack sensor on PET substrate with the strain range from zero to two percent by regression analysis was ~2800, that of the crack sensor on PI substrate was 28,000, and that of the PI encapsulated crack sensor was 12,500.Interestingly, it shows that by only changing substrates from PET to PI, sensitivity increased up to 10 times.We assume that this phenomenon was caused by crack penetration into the PI substrate because the sensitivity of the sensor increases as the depth of crack increases [16].Figure S6 † shows the change in resistance of three samples when applied strain from zero to two percent.The result demonstrates that the change in resistance to the applied strain of the PI encapsulated crack sensor was more linear than the other sensors.However, the PI encapsulation process induced reduced gauge factor of ~45%. Figure 5d-f each shows a graph of the averaged normalized resistance of three different samples along with an error bar (n = 5).Black square marks are for the loading to the strain of two percent, and red circle marks are for unloading.The response time of the sensor was about five milliseconds (Figure S7 †).It shows that these three different types of sensors had quite good reproducibility without hysteresis.

Figure 1 .
Figure 1.Schematic illustrations and scanning electron microscope (SEM) images of polyimide (PI) encapsulated crack sensor.(a) The schematic illustration and optical microscope image of the PI encapsulated crack sensor.(b) SEM image of the crack sensor before encapsulation (left), and cross-

Figure 1 .
Figure 1.Schematic illustrations and scanning electron microscope (SEM) images of polyimide (PI) encapsulated crack sensor.(a) The schematic illustration and optical microscope image of the PI encapsulated crack sensor.(b) SEM image of the crack sensor before encapsulation (left), and cross-sectional SEM image after encapsulation (right).(c) The schematic illustration of generated cracks on metal layer before stretching.(d) The schematic illustration of widened cracks after stretching.

Figure 2 .
Figure 2. The performance comparison of the crack sensors in different environments.(a) Performances of the sensors after thermal treatment; (b) Performances of the sensors after dipping in water; (c) Performances of the sensors dipping in saline solution (0.9% NaCl).PET: polyethylene terephthalate; PDMS: polydimethylsiloxane.

Figure 3 .
Figure 3.Comparison of performance changes according to PI encapsulation thickness.(a) Resistance response in two-percent strain for various encapsulation thicknesses.(b) Gauge factor comparison by

Figure 2 .
Figure 2. The performance comparison of the crack sensors in different environments.(a) Performances of the sensors after thermal treatment; (b) Performances of the sensors after dipping in water; (c) Performances of the sensors dipping in saline solution (0.9% NaCl).PET: polyethylene terephthalate; PDMS: polydimethylsiloxane.

Figure 2 .
Figure 2. The performance comparison of the crack sensors in different environments.(a) Performances of the sensors after thermal treatment; (b) Performances of the sensors after dipping in water; (c) Performances of the sensors dipping in saline solution (0.9% NaCl).PET: polyethylene terephthalate; PDMS: polydimethylsiloxane.

Figure 3 .
Figure 3.Comparison of performance changes according to PI encapsulation thickness.(a) Resistance response in two-percent strain for various encapsulation thicknesses.(b) Gauge factor comparison by

Figure 3 .
Figure 3.Comparison of performance changes according to PI encapsulation thickness.(a) Resistance response in two-percent strain for various encapsulation thicknesses.(b) Gauge factor comparison by encapsulation thickness.(c) Performance comparison by encapsulation thickness in saline solution (0.9% NaCl).

Figure 4 .
Figure 4. Marathon test of the crack sensors up to 15,000 cycles.(a) Durability test result of the crack sensor on PET substrate.(b) Durability test result of the crack sensor on PI substrate.(c) Durability test result of the PI encapsulated crack sensor on PI substrate.

Figure 4 .
Figure 4. Marathon test of the crack sensors up to 15,000 cycles.(a) Durability test result of the crack sensor on PET substrate.(b) Durability test result of the crack sensor on PI substrate.(c) Durability test result of the PI encapsulated crack sensor on PI substrate.

Figure 5 .
Figure 5.The normalized resistance variations of the crack sensors with strain.(a) The normalized resistance variance versus two-percent strain in 10 times cyclic test of the crack sensor on PET substrate.(b) The crack sensor on PI substrate without the encapsulating layer.(c) The crack sensor on PI substrate with the encapsulating layer.(d) Reproducible and hysteresis test obtained from crack sensors on PET substrate.(e) Crack sensors on PI substrate without the encapsulating layer.(f) The crack sensors on PI substrate with the encapsulating layer.