Non-hermetic encapsulation for implantable electronic devices based on epoxy Fabian Boeser, Student Member IEEE, Juan S. Ordonez, Member IEEE, Martin Schuettler, Member IEEE, Thomas Stieglitz, Senior Member IEEE, Dennis T.T. Plachta, Member IEEE
Abstract—Hermetic and non-hermetic implant packaging are the two strategies to protect electronic systems from the humid conditions inside the human body. Within the scope of this work twelve different material combinations for a nonhermetic, high-reliable epoxy based encapsulation technique were characterized. Three EPO-TEK (ET) epoxies and one low budget epoxy were chosen for studies with respect to their processability, water vapor transmission rate (WVTR) and adhesion to two different ceramic-based substrates as well as to one standard FR4-substrate. Setups were built to analyze the mentioned properties for at least 30 days using an aging test in a moist environment. As secondary test subjects, commercially available USB flash drives (UFD) were successfully encapsulated inside the epoxies, soaked in phosphate buffered saline (PBS, pH=7.4), stored in an incubator (37°C) and tested for 256 days without failure. By means of epoxy WVTR (0.0278 g/day/m²) and degrease of adhesion (24.59 %) during 30 days in PBS, the combination of the standard FR4-substrate and the epoxy ET 301-2 was found to feature the best encapsulation properties. If a ceramic-based electronic system has to be used, the most promising combination consists of the alumina substrate and the epoxy ET 302-3M (WVTR: 0.0588 g/day/m²; adhesion drop: 49.58 %).
I. INTRODUCTION Since the late 1950s, electronic devices have been successfully used to record intracorporeal signals of the human body. Up till now, the field of application extended from determination of body temperature or blood pressure to selective neuronal stimulation [1]. The diversity of application is based on two improvements: the miniaturization of integrated circuits (ICs), enabling an increased complexity of implantable devices [2] and the everexpanding amount of biocompatible materials [3]. To guarantee long-term stability and functionality, electronic devices have to be protected from the harsh environment inside the body. The extracellular space provides a warm, humid, salty, and protein-rich compartment, which dramatically accelerates the failure of electronic components. These malfunctions were induced due to corrosion processes, dissolution of inorganic passivation layers or electronic shortcuts [4, 5]. The protection of implantable medical applications by encapsulation is one of the most challenging fields in neuroprosthetics [6]. The claimed biocompatibility of the used materials additionally aggravates the encapF. Boeser, J.S. Ordonez and T. Stieglitz are with the Laboratory for Biomedical Microtechnology at the Department of Microsystems Engineering, IMTEK, University of Freiburg, 79110 Freiburg, Germany, (phone: +49(0)761 2037428; fax: +49(0)761 2037472; e-mail: [email protected]). M. Schuettler is with the CorTec GmbH, 79110 Freiburg, Germany. Dennis T. T. Plachta is with the Department of Neurosurgery, University Medical Center Freiburg, 79106 Freiburg, Germany
978-1-4244-9270-1/15/$31.00 ©2015 EU
sulation challenge [7]. Basically, the protection of electronic devices can be implemented with a hermetic housing or a non-hermetic polymer encapsulation [5]. Hermetically sealed housings are made of materials featuring a low water permeability rate. They preserve a dry atmosphere inside the hermetic package and guarantee the device functionality over decades. One elaborated example is the cardiac pacemaker, featuring a metal housing made of titanium. However, ceramics and glasses can be used as well [8]. Water ingress into the dry interior of hermetic encapsulations is based on diffusion processes through the housing material, along emerged cracks inside the sealing or along the electronic feedthroughs [9]. For certain conditions, non-hermetic encapsulations can provide a sufficient alternative to protect electronics components. A suitable polymer-based casting material has to wet the whole surfaces of the implant components to adhere in an adequate way to prevent delamination. Wetting the surfaces prevents water to condense from the vaporous to the liquid phase into existing voids inside the encapsulate [2]. To keep liquid water out of the substrate/adherent interface prevents the mentioned, water induced failure mechanisms, which ensures the implants long-term functionality. Commercially available implantable devices illustrate the feasibility of non-hermetic encapsulated devices made of medical silicone rubber [10]. Nevertheless, non-hermetic encapsulation technique is only suitable for devices which are fabricated out of discrete electronic components [3]. In Addition to the well investigated silicone rubber, epoxies can be considered for short time applications including ICs [11]. The up to three orders of magnitude lower water permeability and the low viscosity are the main advantages of epoxies over silicone rubber [2]. This paper describes the processability of epoxies as encapsulation materials, and characterizes the most important parameters: adhesion and WVTR, experimentally. The promising substrate/adherent combination can be used for in vivo studies. II. MATERIALS & METHODS The used epoxy materials are the elaborated ET 301, ET 301-2, ET 302-3M and the low budget “Epoxidharz300” (EH300) (properties see table 1). The selected substrates are a low temperature co-fired ceramic (LTCC, LTCC41020, ESL ElectroScience, King of Prussia, PA, USA), a 0,635 mm thick alumina (Al2O3 96 %, Rubalith 708S, CeramTec, Plochingen, Germany) and a glass-reinforced epoxy (FR4, R1755M FR4 incl. solder frame, Andus Electronics, Berlin, Germany) including a planar solder resist.
809
TABLE 1: MAIN PROPERTIES OF SELECTED EPOXIES
Manufacturer Mixing ratio by weight Pot Life (h) Cure schedule Viscosity (mPa s) Shore D Hardness Standard
ET301
ET301-2
ET302-3M
EH300
Epoxy Technology
Epoxy Technology
Epoxy Technology
BehnkeEpoxid
100:25
100:35
100:45
100:55
1-2 1 h @ 65 °C 24 h @ 23 °C 100-200
8 3 h @ 80 °C 48 h @ 23 °C 225-425
1 3 h @ 65 °C 24 h @ 23 °C 800-1600
1
85 ISO-10993 USP Class VI ASTM E595
80 ISO-10993 USP Class VI ASTM E595
80
A. Water vapor transmission rate (WVTR) To characterize the epoxies’ WVTR according to the standard ASTM E96, we created a setup based on the hermetic ceramic-metal package of thick film processing from the early 1970s [12]. The setup assembly consists of an alumina substrate (Al2O3 96 %, Rubalith 708S, CeramTec, Plochingen, Germany) and a connection nipple (series 4243G, 15 mm x 1/2") forming the measurement cavity. In total, 15 single measuring packages were assembled. The screen-printed Al2O3 substrate (Fig. 1-a) features two metal layers separated by a dielectric layer. The first metal layer provides the electronic feedthroughs out of the cavity, while the second acts as a solder frame. The printed circuit board (PCB) enables the readout of the relative humidity φ (RH%) inside the package by means of a combined humidity and temperature sensor (SHT25, Sensirion, Staefa, Schweiz). The volume is defined throughout the connection nipple, which was soldered onto the metallic frame. To close the cavity’s opening, a fabricated epoxy membrane was placed between two identical o-rings made of viton (15.00 mm x 2.50 mm, FPM75). For clamping, a screwcap (G1/2 SW24) was screwed onto the connection nipple (Fig. 1-b) with the defined torsional moment of 12 Nm, preventing the dis-placement of the o-rings. Before closing, all parts were dried in a two-step process. The parts were placed in an incubator for ten days at 75 °C. After transferring the parts into a glovebox (SG1200/750TS, Vigor Tech. Co. Ltd., Houston, USA) providing a controlled dry helium atmosphere (cH2O = 575 ppm), the assemblies were allowed to stand for another five days at room temperature. The completed assembly was immersed in PBS (P3813 Sigma-Aldrich, St. Louis, USA) upside down. Surrounding water could access the measuring chamber via three locally separated ways, forced by the concentration gradient along the inner and outer media. The critical water path for the WVTR characterization is across the epoxy membrane. The two remaining pathways along the thread and across the o-ring seal act as disturbance factors (Fig. 1-c). To validate the impact of these two disturbances, reference measurements were conducted. The first pathway was measured by blocking the primary water path across the epoxy membrane. A 25 µm strong MP35N foil (MP35N®–LTI, Hamilton Precision
24 h @ 23°C 750-950
USP Class VI ASTM E595
Metals, Lancaster, USA) was fixed on the outer surface of the membrane (REF1, closing water path 1). To further investigate the leakage paths, a metallic plate was soldered onto the opening of the screw cap to quantify the water ingress along the thread exclusively (REF2, closing water path 1 and 3). Sensor readout was fully automated. Therefore, the bidirectional serial data (SDA) and serial clock (SCL) lines of the sensirion evaluation kit EK-H5 (Sensirion) were connected to a multiplexer (MUX). To provide the sensors power supply, the supply voltage (VDD) and the ground (VSS) ports were connected to the evaluation kit all in parallel. The MUX was controlled by a DAQ-6008 unit via a LabView program (both National Instruments, Austin, TX, USA). After writing one set of humidity/temperature data, the MUX was triggered to switch to the next channel. The measurement intervals were defined within the EK-H5 software (Sensirion).
Figure 1: Assembly of a single chamber to measure the water ingress across the epoxy membrane. a) Ceramic PCB for SHT25 readout 1) Al2O3substrate, 2) Pt/Au tracks, 3) Glass-ceramic isolation layer, 4) Pt/Au solder frame, 5) SHT25 sensor & 100 nF buffering capacity; b) “Hermetic housing” 1) Connection nipple with an external thread, 2, 4) O-ring seal, 3) Epoxy membrane, 5) Screw cap; c) Cross section showing the water paths 1) Desired water path across the epoxy membrane, 2) Disturbance path through thread, 3) Disturbance path through o-ring seal.
According to the standard ASTM E96 the WVTR can be calculated from the weight change m over time t, by means of the cavity’s opening surface A according to (1).
810
The linear region of the received data had to be fitted within a defined range. Based on the linear slope dy/dx the mass change can be calculated having (2), where ps is the saturation pressure, RD is the specific gas constant and T the absolute temperature.
III. RESULTS
A. WVTR The described measuring units with an inner volume V of V = 44,179E-6 m3 and an opening area A of A = 1,767E-4 m2 were successfully fabricated. For mass WVTR = m / (t * A) (1) change calculations according to (1), the following values m = (dy/dx * ps) / (RD * T) (2) were used: ps (37 °C) = 6276 Pa; RD = 461.5 J/kg/K; T = 310 K. The results are summarized in Table 2. In order B. Adhesion to get the epoxies real WVTR, the mean value of REF1 was The substrate/adherent adhesion was measured with a subtracted from the original data. shear test based on the military standard MIL-STD-883G (method 2019.7). We used a bond tester (dage4000, Nordson, TABLE 2: WVTR OF TESTED EPOXIES Westlake, Ohio, USA) with the loaded cartridge DS100kg WVTR in capable of performing shear tests up to 1000 N. Each Δ mass in WVTR in Sample (S) dy/dx g/day/m2 g/day * g/day/m2 adhesion measurement was performed on a total amount of (incl. REF1) 14 shear structures for all the aging steps. Tests were made in n.a.** n.a.** n.a.** n.a.** the initial state (0 days, dry) and after 10-, 20-, 30 days of S1 ET 301 incubation in PBS at 37 °C. Each structure had an ideally S2 ET 301 1.100 2.132E-5 0.1201 0.0752 base area of 3 mm x 3 mm and a height of 6 mm. Shear speed S3 ET 301 n.a.** n.a.** n.a.** n.a.** was 150 µm/s at a shear height of 200 µm. C. Lifetime test The functionality of the non-hermetic encapsulation was tested under body-like conditions. For this purpose commercially available USB flash drives (UFD) were casted into the mentioned epoxies (in total twelve, three of each epoxy), which represent the implanted electronics. As a first step a negative-form was fabricated (Fig. 2-a) out of a twocomponent cast-silicone (AV8010, fiberglasdiscount, Fahrenzhausen, Germany). Up next, the connectors of the UFDs were molded into a protective cap, which could be removed easily. The cap prevented the connectors’ closure during the epoxy casting, ensuring their functionality. Additionally, the adjusted cap geometry stabilized the circuit during casting (Fig. 2-b-c). Before casting the UFDs were cleaned with ethanol, rinsed with deionized (DI) water and were dried for 48 hours at 85°C. After vulcanization performed at 37 °C the UFDs were demoulded (Fig 2-d).
Figure 2: Fabrication steps for the non-hermetic encapsulation of commercially available USB flash drives.
The encapsulated UFDs were clamped onto a containers lid, immersed in PBS and incubated at 37 °C, to test the longterm functionality. Throughout the measurement the UFDs were connected to a personal computer (PC) allowing the data transfer to a certain file on each UFD. The writing took place three to five times a week. For safety reasons each file was stored locally on the PC after each writing event concerning UFD failure.
S1 ET301-2
0.545
1.056E-5
0.0598
0.0149
S2 ET301-2
0.712
1.380E-5
0.0781
0.0332
S3 ET301-2
0.731
1.417 E-5
0.0802
0.0353
S1 ET302-3M
0.959
1.859 E-5
0.1052
0.0603
S2 ET302-3M
0.945
1.831 E-5
0.1036
0.0587
S3 ET302-3M
0.933
1.808 E-5
0.1023
0.0574
S1-S3 EH300
n.a***
n.a.***
n.a.***
n.a.***
S1 REF1
0.400
7.752 E-6
0.0439
S2 REF1
0.419
8.120 E-6
0.0460
S1 REF2
0.046
8.915 E-7
0.0050
S2 REF2
0.004
7.752 E-8
0,0004
* ** ***
calculated with respect to the cavity volume V broken sample data not matching to standard ASTM E96
B. Adhesion Adhesion in form of shear stress is illustrated over the different substrates including the four aging steps (Fig. 3). To calculate the shear stress out of the measured shear forces, the mean base area of A = 9.394 ± 0.013 mm2 was used, based on 25 random chosen shear structures. The adhesion characteristics between the ceramic substrates and the tested epoxies are nearly identical. After ten days stored in PBS, both ceramic combinations showed up to 80 % lower adhesion values, compared to the initial state (21.05 MPa). If detectable, the shear stress for the 20and 30 days measurements were even lower. Compared to the ceramics the practical adhesion between the FR4 and the epoxies was 38 % stronger (0 days, 29.06 MPa). After 30 days stored in PBS the mean shear stress reduced by 31 %.
811
Figure 3: Adhesion behavior during the aging test inside PBS incubated at 37 °C. Each substrate/adherent combination was measured in the dry state (0 days) and after 10-, 20- and 30 days incubated in PBS. Shear stress was calculated out of the measured shear force and the shear structure’s mean base area. Data show mean and standard variation.
C. Lifetime test UFD’s with the wanted non-hermetic encapsulation were successfully fabricated. The casting process was implemented without any air bubbles with exception of the UFD’s bottom side (Fig. 4). Due to the closed, horizontal surface of the casting process, voids were disabled to rise. They remained inside the encapsulation and may influencing the UFD’s lifetime during the experiment. Up to now, the experiment has been running for 256 days. In total, 272 writing processes were made on each UFD. None of the encapsulated UFD’s failed so far.
In general, we found the adhesion between the epoxies and the standard FR4 substrate to be stronger than between epoxy/ceramic during the aging test. This difference might originate from the superior mechanical interlocking between the epoxies and the FR4 substrate due to the higher surface topology of the FR4 substrate. In addition, both polymer materials can form covalent bonds across the interface increasing the adhesion strength. Regarding the adhesion between the EPO-TEK epoxies and the FR4 substrate, the ET301 showed the best adhesion behavior. This merit might be a result of the low viscosity and the high shear hardness. Low viscosity values increase the effective surface between the adherent and the substrate. High material hardness improves the resistance against acting shear forces. The lifetime test with the casted UFDs highlights the processability and feasibility of the non-hermetic encapsulation based on epoxies. Remaining embedded air bubbles of the casting process might be avoided by placing the negative form in a vertically aligned position. A reasonable and universally applicable method to protect electronic devices from the harsh conditions inside the human body was successfully realized. In future experiments, we will validate the reliability of the proposed method in in-vivo experiments.
REFERENCES [1] [2]
[3] [4] [5] [6] Figure 4: a) Negative casting-form (1) with removed, vulcanized UFD (3) including the proection cap (2); b) top and bottom view of a demoulded UFD. At the bottom side (upper UFD) voids are visible at the substrate/adherent interface may influncing the electronics lifetime.
IV. DISCUSSION AND CONCLUSIONS
[7] [8] [9]
We successfully measured the epoxies WVTR with our described assembly. Closing individual water paths reduced the water ingress into the measuring cavity. The water path with the strongest leakage was observed along the o-ring seal. Up to 50 % of the detected water reached the inner cavity by passing through the epoxy/o-ring interface or via diffusion processes throughout the viton material. As an alternative to the viton material, the sealing could be realized with an indium filament providing superior gas permeability.
[10] [11]
[12]
812
A. Imnann, D. Hodgins, Implantable sensor systems for medical applications. Cambridge, UK, Philadelphia: Woodhead Pub Ltd, 2013. A. Vanhoestenberghe, N. Donaldson, “Corrosion of silicon integrated circuits and lifetime predictions in implantable electronic devices,” (eng), J Neural Eng, vol. 10, no. 3, p. 031002, 2013. E. Greenbaum, D. Zhou, Techniques and engineering approaches. New York, NY: Springer Science+Business Media LLC, 2009. J. W. Osenbach, “Water-Induced Corrosion of Materials Used for Semiconductor Passivation,” J. Electrochem. Soc, vol. 140, no. 12, p. 3667, 1993. P. E. K. Donaldson, “The Encapsulation of Microelectronic Devices for Long-Term Surgical Implantation,” IEEE Trans. Biomed. Eng, vol. BME-23, no. 4, pp. 281–285, 1976. M. F. Nichols, “The challenges for hermetic encapsulation of implanted devices--a review,” (eng), Crit Rev Biomed Eng, vol. 22, no. 1, pp. 39–67, 1994. Y. Kanda, R. Aoshima, A. Takada, “Blood compatibility of components and materials in silicon integrated circuits,” Electron. Lett, vol. 17, no. 16, p. 558, 1981. R. Traeger, “Nonhermeticity of Polymeric Lid Sealants,” IEEE Trans. Parts, Hybrids, Packag, vol. 13, no. 2, pp. 147–152, 1977. R. Thomas, “Moisture, Myths, and Microcircuits,” IEEE Trans. Parts, Hybrids, Packag, vol. 12, no. 3, pp. 167–171, 1976. G. S. Brindley, “The first 500 patients with sacral anterior root stimulator implants: general description,” (eng), Paraplegia, vol. 32, no. 12, pp. 795–805, 1994. P. Bellot, M. W. Welker, G. Soriano, M. van Schaewen, B. Appenrodt, R. Wiest, F. Nevens, P. Zapater, J. Such, “Automated low flow pump system for the treatment of refractory ascites: a multi-center safety and efficacy study,” (eng), J. Hepatol, vol. 58, no. 5, pp. 922–927, 2013. J. R. Corkhill, “Packaging Hybrid Microcircuits,” in Handbook of Thick Film Technology, pp. 333–379, 1976.
Abstract—Hermetic and non-hermetic implant packaging are the two strategies to protect electronic systems from the humid conditions inside the human body. Within the scope of this work twelve different material combinations for a nonhermetic, high-reliable epoxy based encapsulation technique were characterized. Three EPO-TEK (ET) epoxies and one low budget epoxy were chosen for studies with respect to their processability, water vapor transmission rate (WVTR) and adhesion to two different ceramic-based substrates as well as to one standard FR4-substrate. Setups were built to analyze the mentioned properties for at least 30 days using an aging test in a moist environment. As secondary test subjects, commercially available USB flash drives (UFD) were successfully encapsulated inside the epoxies, soaked in phosphate buffered saline (PBS, pH=7.4), stored in an incubator (37°C) and tested for 256 days without failure. By means of epoxy WVTR (0.0278 g/day/m²) and degrease of adhesion (24.59 %) during 30 days in PBS, the combination of the standard FR4-substrate and the epoxy ET 301-2 was found to feature the best encapsulation properties. If a ceramic-based electronic system has to be used, the most promising combination consists of the alumina substrate and the epoxy ET 302-3M (WVTR: 0.0588 g/day/m²; adhesion drop: 49.58 %).
I. INTRODUCTION Since the late 1950s, electronic devices have been successfully used to record intracorporeal signals of the human body. Up till now, the field of application extended from determination of body temperature or blood pressure to selective neuronal stimulation [1]. The diversity of application is based on two improvements: the miniaturization of integrated circuits (ICs), enabling an increased complexity of implantable devices [2] and the everexpanding amount of biocompatible materials [3]. To guarantee long-term stability and functionality, electronic devices have to be protected from the harsh environment inside the body. The extracellular space provides a warm, humid, salty, and protein-rich compartment, which dramatically accelerates the failure of electronic components. These malfunctions were induced due to corrosion processes, dissolution of inorganic passivation layers or electronic shortcuts [4, 5]. The protection of implantable medical applications by encapsulation is one of the most challenging fields in neuroprosthetics [6]. The claimed biocompatibility of the used materials additionally aggravates the encapF. Boeser, J.S. Ordonez and T. Stieglitz are with the Laboratory for Biomedical Microtechnology at the Department of Microsystems Engineering, IMTEK, University of Freiburg, 79110 Freiburg, Germany, (phone: +49(0)761 2037428; fax: +49(0)761 2037472; e-mail: [email protected]). M. Schuettler is with the CorTec GmbH, 79110 Freiburg, Germany. Dennis T. T. Plachta is with the Department of Neurosurgery, University Medical Center Freiburg, 79106 Freiburg, Germany
978-1-4244-9270-1/15/$31.00 ©2015 EU
sulation challenge [7]. Basically, the protection of electronic devices can be implemented with a hermetic housing or a non-hermetic polymer encapsulation [5]. Hermetically sealed housings are made of materials featuring a low water permeability rate. They preserve a dry atmosphere inside the hermetic package and guarantee the device functionality over decades. One elaborated example is the cardiac pacemaker, featuring a metal housing made of titanium. However, ceramics and glasses can be used as well [8]. Water ingress into the dry interior of hermetic encapsulations is based on diffusion processes through the housing material, along emerged cracks inside the sealing or along the electronic feedthroughs [9]. For certain conditions, non-hermetic encapsulations can provide a sufficient alternative to protect electronics components. A suitable polymer-based casting material has to wet the whole surfaces of the implant components to adhere in an adequate way to prevent delamination. Wetting the surfaces prevents water to condense from the vaporous to the liquid phase into existing voids inside the encapsulate [2]. To keep liquid water out of the substrate/adherent interface prevents the mentioned, water induced failure mechanisms, which ensures the implants long-term functionality. Commercially available implantable devices illustrate the feasibility of non-hermetic encapsulated devices made of medical silicone rubber [10]. Nevertheless, non-hermetic encapsulation technique is only suitable for devices which are fabricated out of discrete electronic components [3]. In Addition to the well investigated silicone rubber, epoxies can be considered for short time applications including ICs [11]. The up to three orders of magnitude lower water permeability and the low viscosity are the main advantages of epoxies over silicone rubber [2]. This paper describes the processability of epoxies as encapsulation materials, and characterizes the most important parameters: adhesion and WVTR, experimentally. The promising substrate/adherent combination can be used for in vivo studies. II. MATERIALS & METHODS The used epoxy materials are the elaborated ET 301, ET 301-2, ET 302-3M and the low budget “Epoxidharz300” (EH300) (properties see table 1). The selected substrates are a low temperature co-fired ceramic (LTCC, LTCC41020, ESL ElectroScience, King of Prussia, PA, USA), a 0,635 mm thick alumina (Al2O3 96 %, Rubalith 708S, CeramTec, Plochingen, Germany) and a glass-reinforced epoxy (FR4, R1755M FR4 incl. solder frame, Andus Electronics, Berlin, Germany) including a planar solder resist.
809
TABLE 1: MAIN PROPERTIES OF SELECTED EPOXIES
Manufacturer Mixing ratio by weight Pot Life (h) Cure schedule Viscosity (mPa s) Shore D Hardness Standard
ET301
ET301-2
ET302-3M
EH300
Epoxy Technology
Epoxy Technology
Epoxy Technology
BehnkeEpoxid
100:25
100:35
100:45
100:55
1-2 1 h @ 65 °C 24 h @ 23 °C 100-200
8 3 h @ 80 °C 48 h @ 23 °C 225-425
1 3 h @ 65 °C 24 h @ 23 °C 800-1600
1
85 ISO-10993 USP Class VI ASTM E595
80 ISO-10993 USP Class VI ASTM E595
80
A. Water vapor transmission rate (WVTR) To characterize the epoxies’ WVTR according to the standard ASTM E96, we created a setup based on the hermetic ceramic-metal package of thick film processing from the early 1970s [12]. The setup assembly consists of an alumina substrate (Al2O3 96 %, Rubalith 708S, CeramTec, Plochingen, Germany) and a connection nipple (series 4243G, 15 mm x 1/2") forming the measurement cavity. In total, 15 single measuring packages were assembled. The screen-printed Al2O3 substrate (Fig. 1-a) features two metal layers separated by a dielectric layer. The first metal layer provides the electronic feedthroughs out of the cavity, while the second acts as a solder frame. The printed circuit board (PCB) enables the readout of the relative humidity φ (RH%) inside the package by means of a combined humidity and temperature sensor (SHT25, Sensirion, Staefa, Schweiz). The volume is defined throughout the connection nipple, which was soldered onto the metallic frame. To close the cavity’s opening, a fabricated epoxy membrane was placed between two identical o-rings made of viton (15.00 mm x 2.50 mm, FPM75). For clamping, a screwcap (G1/2 SW24) was screwed onto the connection nipple (Fig. 1-b) with the defined torsional moment of 12 Nm, preventing the dis-placement of the o-rings. Before closing, all parts were dried in a two-step process. The parts were placed in an incubator for ten days at 75 °C. After transferring the parts into a glovebox (SG1200/750TS, Vigor Tech. Co. Ltd., Houston, USA) providing a controlled dry helium atmosphere (cH2O = 575 ppm), the assemblies were allowed to stand for another five days at room temperature. The completed assembly was immersed in PBS (P3813 Sigma-Aldrich, St. Louis, USA) upside down. Surrounding water could access the measuring chamber via three locally separated ways, forced by the concentration gradient along the inner and outer media. The critical water path for the WVTR characterization is across the epoxy membrane. The two remaining pathways along the thread and across the o-ring seal act as disturbance factors (Fig. 1-c). To validate the impact of these two disturbances, reference measurements were conducted. The first pathway was measured by blocking the primary water path across the epoxy membrane. A 25 µm strong MP35N foil (MP35N®–LTI, Hamilton Precision
24 h @ 23°C 750-950
USP Class VI ASTM E595
Metals, Lancaster, USA) was fixed on the outer surface of the membrane (REF1, closing water path 1). To further investigate the leakage paths, a metallic plate was soldered onto the opening of the screw cap to quantify the water ingress along the thread exclusively (REF2, closing water path 1 and 3). Sensor readout was fully automated. Therefore, the bidirectional serial data (SDA) and serial clock (SCL) lines of the sensirion evaluation kit EK-H5 (Sensirion) were connected to a multiplexer (MUX). To provide the sensors power supply, the supply voltage (VDD) and the ground (VSS) ports were connected to the evaluation kit all in parallel. The MUX was controlled by a DAQ-6008 unit via a LabView program (both National Instruments, Austin, TX, USA). After writing one set of humidity/temperature data, the MUX was triggered to switch to the next channel. The measurement intervals were defined within the EK-H5 software (Sensirion).
Figure 1: Assembly of a single chamber to measure the water ingress across the epoxy membrane. a) Ceramic PCB for SHT25 readout 1) Al2O3substrate, 2) Pt/Au tracks, 3) Glass-ceramic isolation layer, 4) Pt/Au solder frame, 5) SHT25 sensor & 100 nF buffering capacity; b) “Hermetic housing” 1) Connection nipple with an external thread, 2, 4) O-ring seal, 3) Epoxy membrane, 5) Screw cap; c) Cross section showing the water paths 1) Desired water path across the epoxy membrane, 2) Disturbance path through thread, 3) Disturbance path through o-ring seal.
According to the standard ASTM E96 the WVTR can be calculated from the weight change m over time t, by means of the cavity’s opening surface A according to (1).
810
The linear region of the received data had to be fitted within a defined range. Based on the linear slope dy/dx the mass change can be calculated having (2), where ps is the saturation pressure, RD is the specific gas constant and T the absolute temperature.
III. RESULTS
A. WVTR The described measuring units with an inner volume V of V = 44,179E-6 m3 and an opening area A of A = 1,767E-4 m2 were successfully fabricated. For mass WVTR = m / (t * A) (1) change calculations according to (1), the following values m = (dy/dx * ps) / (RD * T) (2) were used: ps (37 °C) = 6276 Pa; RD = 461.5 J/kg/K; T = 310 K. The results are summarized in Table 2. In order B. Adhesion to get the epoxies real WVTR, the mean value of REF1 was The substrate/adherent adhesion was measured with a subtracted from the original data. shear test based on the military standard MIL-STD-883G (method 2019.7). We used a bond tester (dage4000, Nordson, TABLE 2: WVTR OF TESTED EPOXIES Westlake, Ohio, USA) with the loaded cartridge DS100kg WVTR in capable of performing shear tests up to 1000 N. Each Δ mass in WVTR in Sample (S) dy/dx g/day/m2 g/day * g/day/m2 adhesion measurement was performed on a total amount of (incl. REF1) 14 shear structures for all the aging steps. Tests were made in n.a.** n.a.** n.a.** n.a.** the initial state (0 days, dry) and after 10-, 20-, 30 days of S1 ET 301 incubation in PBS at 37 °C. Each structure had an ideally S2 ET 301 1.100 2.132E-5 0.1201 0.0752 base area of 3 mm x 3 mm and a height of 6 mm. Shear speed S3 ET 301 n.a.** n.a.** n.a.** n.a.** was 150 µm/s at a shear height of 200 µm. C. Lifetime test The functionality of the non-hermetic encapsulation was tested under body-like conditions. For this purpose commercially available USB flash drives (UFD) were casted into the mentioned epoxies (in total twelve, three of each epoxy), which represent the implanted electronics. As a first step a negative-form was fabricated (Fig. 2-a) out of a twocomponent cast-silicone (AV8010, fiberglasdiscount, Fahrenzhausen, Germany). Up next, the connectors of the UFDs were molded into a protective cap, which could be removed easily. The cap prevented the connectors’ closure during the epoxy casting, ensuring their functionality. Additionally, the adjusted cap geometry stabilized the circuit during casting (Fig. 2-b-c). Before casting the UFDs were cleaned with ethanol, rinsed with deionized (DI) water and were dried for 48 hours at 85°C. After vulcanization performed at 37 °C the UFDs were demoulded (Fig 2-d).
Figure 2: Fabrication steps for the non-hermetic encapsulation of commercially available USB flash drives.
The encapsulated UFDs were clamped onto a containers lid, immersed in PBS and incubated at 37 °C, to test the longterm functionality. Throughout the measurement the UFDs were connected to a personal computer (PC) allowing the data transfer to a certain file on each UFD. The writing took place three to five times a week. For safety reasons each file was stored locally on the PC after each writing event concerning UFD failure.
S1 ET301-2
0.545
1.056E-5
0.0598
0.0149
S2 ET301-2
0.712
1.380E-5
0.0781
0.0332
S3 ET301-2
0.731
1.417 E-5
0.0802
0.0353
S1 ET302-3M
0.959
1.859 E-5
0.1052
0.0603
S2 ET302-3M
0.945
1.831 E-5
0.1036
0.0587
S3 ET302-3M
0.933
1.808 E-5
0.1023
0.0574
S1-S3 EH300
n.a***
n.a.***
n.a.***
n.a.***
S1 REF1
0.400
7.752 E-6
0.0439
S2 REF1
0.419
8.120 E-6
0.0460
S1 REF2
0.046
8.915 E-7
0.0050
S2 REF2
0.004
7.752 E-8
0,0004
* ** ***
calculated with respect to the cavity volume V broken sample data not matching to standard ASTM E96
B. Adhesion Adhesion in form of shear stress is illustrated over the different substrates including the four aging steps (Fig. 3). To calculate the shear stress out of the measured shear forces, the mean base area of A = 9.394 ± 0.013 mm2 was used, based on 25 random chosen shear structures. The adhesion characteristics between the ceramic substrates and the tested epoxies are nearly identical. After ten days stored in PBS, both ceramic combinations showed up to 80 % lower adhesion values, compared to the initial state (21.05 MPa). If detectable, the shear stress for the 20and 30 days measurements were even lower. Compared to the ceramics the practical adhesion between the FR4 and the epoxies was 38 % stronger (0 days, 29.06 MPa). After 30 days stored in PBS the mean shear stress reduced by 31 %.
811
Figure 3: Adhesion behavior during the aging test inside PBS incubated at 37 °C. Each substrate/adherent combination was measured in the dry state (0 days) and after 10-, 20- and 30 days incubated in PBS. Shear stress was calculated out of the measured shear force and the shear structure’s mean base area. Data show mean and standard variation.
C. Lifetime test UFD’s with the wanted non-hermetic encapsulation were successfully fabricated. The casting process was implemented without any air bubbles with exception of the UFD’s bottom side (Fig. 4). Due to the closed, horizontal surface of the casting process, voids were disabled to rise. They remained inside the encapsulation and may influencing the UFD’s lifetime during the experiment. Up to now, the experiment has been running for 256 days. In total, 272 writing processes were made on each UFD. None of the encapsulated UFD’s failed so far.
In general, we found the adhesion between the epoxies and the standard FR4 substrate to be stronger than between epoxy/ceramic during the aging test. This difference might originate from the superior mechanical interlocking between the epoxies and the FR4 substrate due to the higher surface topology of the FR4 substrate. In addition, both polymer materials can form covalent bonds across the interface increasing the adhesion strength. Regarding the adhesion between the EPO-TEK epoxies and the FR4 substrate, the ET301 showed the best adhesion behavior. This merit might be a result of the low viscosity and the high shear hardness. Low viscosity values increase the effective surface between the adherent and the substrate. High material hardness improves the resistance against acting shear forces. The lifetime test with the casted UFDs highlights the processability and feasibility of the non-hermetic encapsulation based on epoxies. Remaining embedded air bubbles of the casting process might be avoided by placing the negative form in a vertically aligned position. A reasonable and universally applicable method to protect electronic devices from the harsh conditions inside the human body was successfully realized. In future experiments, we will validate the reliability of the proposed method in in-vivo experiments.
REFERENCES [1] [2]
[3] [4] [5] [6] Figure 4: a) Negative casting-form (1) with removed, vulcanized UFD (3) including the proection cap (2); b) top and bottom view of a demoulded UFD. At the bottom side (upper UFD) voids are visible at the substrate/adherent interface may influncing the electronics lifetime.
IV. DISCUSSION AND CONCLUSIONS
[7] [8] [9]
We successfully measured the epoxies WVTR with our described assembly. Closing individual water paths reduced the water ingress into the measuring cavity. The water path with the strongest leakage was observed along the o-ring seal. Up to 50 % of the detected water reached the inner cavity by passing through the epoxy/o-ring interface or via diffusion processes throughout the viton material. As an alternative to the viton material, the sealing could be realized with an indium filament providing superior gas permeability.
[10] [11]
[12]
812
A. Imnann, D. Hodgins, Implantable sensor systems for medical applications. Cambridge, UK, Philadelphia: Woodhead Pub Ltd, 2013. A. Vanhoestenberghe, N. Donaldson, “Corrosion of silicon integrated circuits and lifetime predictions in implantable electronic devices,” (eng), J Neural Eng, vol. 10, no. 3, p. 031002, 2013. E. Greenbaum, D. Zhou, Techniques and engineering approaches. New York, NY: Springer Science+Business Media LLC, 2009. J. W. Osenbach, “Water-Induced Corrosion of Materials Used for Semiconductor Passivation,” J. Electrochem. Soc, vol. 140, no. 12, p. 3667, 1993. P. E. K. Donaldson, “The Encapsulation of Microelectronic Devices for Long-Term Surgical Implantation,” IEEE Trans. Biomed. Eng, vol. BME-23, no. 4, pp. 281–285, 1976. M. F. Nichols, “The challenges for hermetic encapsulation of implanted devices--a review,” (eng), Crit Rev Biomed Eng, vol. 22, no. 1, pp. 39–67, 1994. Y. Kanda, R. Aoshima, A. Takada, “Blood compatibility of components and materials in silicon integrated circuits,” Electron. Lett, vol. 17, no. 16, p. 558, 1981. R. Traeger, “Nonhermeticity of Polymeric Lid Sealants,” IEEE Trans. Parts, Hybrids, Packag, vol. 13, no. 2, pp. 147–152, 1977. R. Thomas, “Moisture, Myths, and Microcircuits,” IEEE Trans. Parts, Hybrids, Packag, vol. 12, no. 3, pp. 167–171, 1976. G. S. Brindley, “The first 500 patients with sacral anterior root stimulator implants: general description,” (eng), Paraplegia, vol. 32, no. 12, pp. 795–805, 1994. P. Bellot, M. W. Welker, G. Soriano, M. van Schaewen, B. Appenrodt, R. Wiest, F. Nevens, P. Zapater, J. Such, “Automated low flow pump system for the treatment of refractory ascites: a multi-center safety and efficacy study,” (eng), J. Hepatol, vol. 58, no. 5, pp. 922–927, 2013. J. R. Corkhill, “Packaging Hybrid Microcircuits,” in Handbook of Thick Film Technology, pp. 333–379, 1976.
