Biomed Microdevices (2015) 17:1 DOI 10.1007/s10544-014-9904-y
Effect of bias voltage and temperature on lifetime of wireless neural interfaces with Al2O3 and parylene bilayer encapsulation Xianzong Xie & Loren Rieth & Ryan Caldwell & Sandeep Negi & Rajmohan Bhandari & Rohit Sharma & Prashant Tathireddy & Florian Solzbacher
# Springer Science+Business Media New York 2015
Abstract The lifetime of neural interfaces is a critical challenge for chronic implantations, as therapeutic devices (e.g., neural prosthetics) will require decades of lifetime. We evaluated the lifetime of wireless Utah electrode array (UEA) based neural interfaces with a bilayer encapsulation scheme utilizing a combination of alumina deposited by Atomic Layer Deposition (ALD) and parylene C. Wireless integrated neural interfaces (INIs), equipped with recording version 9 (INI-R9) ASIC chips, were used to monitor the encapsulation performance through radiofrequency (RF) power and telemetry. The wireless devices were encapsulated with 52 nm of ALD Al2O3 and 6 μm of parylene C, and tested by soaking in phosphate buffered solution (PBS) at 57 °C for 4× accelerated lifetime testing. The INIs were also powered continuously through 2.765 MHz inductive power and forward telemetry link at unregulated 5 V. The bilayer encapsulated INIs were fully functional for ∼35 days (140 days at 37 °C equivalent) with consistent power-up frequencies (∼910 MHz), stable RF signal (∼−75 dBm), and 100 % command reception rate. This is ∼10 times of equivalent lifetime of INIs with parylene-only encapsulation (13 days) under same power condition at 37 °C. The bilayer coated INIs without continuous powering lasted over 1860 equivalent days (still working) at 37 °C. Those results suggest that bias stress is a significant factor to accelerate the failure of the encapsulated devices. The INIs failed completely within 5 days of the initial frequency shift of RF signal at 57 °C, which implied that the RF frequency shift is an early indicator of encapsulation/device failure. X. Xie (*) : L. Rieth : R. Caldwell : S. Negi : R. Bhandari : R. Sharma : P. Tathireddy : F. Solzbacher University of Utah, Salt Lake City, UT, USA e-mail: [email protected]
Keywords Neural interface . Utah electrode array . Atomic layer deposited (ALD) Al2O3 . Parylene . Bilayer encapsulation . Continuous bias voltage . Accelerated aging . Long-term reliability
1 Introduction Implantable neural interfaces have drawn tremendous interests from both researchers and clinicians for usage in neuroprosthetics to diagnose and treat neural disorders (Hochberg et al. 2006; Donoghue 2002; Santhanam et al. 2006; Wessberg et al. 2000; Normann 2007; Velliste et al. 2008). One representative example of this technology is the Utah electrode array (UEA), which is well developed and has FDA-clearance through an Investigation Device Exemption (IDE) and is able to record single unit signals (Normann 2007; Rousche and Normann 1998; Branner et al. 1999; Ledbetter et al. 2013). Wired UEAs continued to be deployed, which use gold wire bundles and percutaneous connectors, to transfer recording and stimulation signals. However, later findings show that wire bundles and percutaneous connectors are more likely to introduce foreign body response (Biran et al. 2007) and cause infections (Scott 2006), especially for chronic implantation. Moreover, wire connections and connector systems have been found to be one of the least reliable parts of neural interfaces due to mechanical stress, handling forces, etc. (Bulusu et al. 2013) and become the bottleneck of chronic applications. To eliminate wired connections, significant efforts have been devoted to develop wireless neural interfaces integrated with active electronics, capable of inductive powering and RF telemetry systems (Yin et al. 2006; Yin and Ghovanloo 2009; Wise et al. 2004; Harrison et al. 2009; Chestek et al. 2009; Kim et al. 2009a). Fully integrated neural
1
Page 2 of 8
interfaces (INIs) with capabilities of stimulating/recording from 100 channels have been developed at the University of Utah (Harrison et al. 2009; Kim et al. 2009a; Xie et al. 2014a). Neural interfaces are required to perform recording/ stimulation for years in vivo for chronic implantations in order to justify the risk associated with implantation. Encapsulation of neural interfaces is critical for their long-term functionalities. Water and ions permeate and accumulate at the interface between the encapsulation and the device, leading to open circuits, short circuits, corrosion of different components, and leakage current (DerMarderosian and Electrochemical migration of metals 1978; Webster 2009). The consequences are catastrophic and often lead to complete device failure. The significant bias voltage associated with integrated electronics further challenges the encapsulation by activating electrochemical degradation modes and the potential for accelerating ion transport. Two major approaches have been widely adopted to encapsulate implantable devices: hermetic enclosures and conformal encapsulation. Hermetic enclosures such as those used by implantable pulse generators (IPGs) are the most commonly used option to seal implantable devices (Dawes 1992). Electromagnetic power and data transmission and device miniaturization are challenges for traditional hermetic packaging. Conformal encapsulation has been widely developed and used, especially for small implants with electromagnetic wireless techniques. Potential encapsulation materials for neural interfaces include polyimide (Lago et al. 2005), parylene (Loeb et al. 1977; Hsu et al. 2009), silicone (Wu et al. 2000), amorphous silicon carbide (Hsu et al. 2007; Cogan et al. 2003), silicon nitride (Cogan et al. 2003), and diamond-like carbon (DLC) (Roy and Lee 2007). Several of these materials can be applied through thin film deposition techniques, with differing degrees of conformality and layer thicknesses. The high requirements make it almost impossible to find one material that meets all encapsulation requirements. Atomic layer deposited (ALD) Al2O3 and parylene C (A+ P) bilayer was found to be an excellent combination for encapsulating UEA based wireless neural interfaces, based on soaking results from interdigitated electrodes (IDEs) (Xie et al. 2013; Xie et al. 2012; Minnikanti et al. 2014). ALD alumina is well-known to be an excellent moisture barrier with water vapor transmission rates (WVTR) at the order of ∼10−10 g · mm/m2 · day with careful optimization (Ghosh et al. 2005; Langereis et al. 2006; Ferrari et al. 2007; Carcia et al. 2006), and is extremely conformal, allowing it to passivate complex surfaces/structures. However, alumina alone is not a suitable encapsulation since it dissolves in water (Potts et al. 2011), most likely due to the incorporation of hydrogen in the form of OH groups in the film as defects (Bulusu et al. 2013; Carcia et al. 2010). Parylene C has been widely used as a coating material for biomedical implantable devices (Hsu et al. 2009; Seymour et al. 2009; Hassler et al. 2010; Xie et al. 2011; Guenther
Biomed Microdevices (2015) 17:1
et al. 2013) due to attractive properties including chemically inertness, low dielectric constant (εr =3.15) (Fortin and Lu 2004), high resistivity (∼1015 Ω·cm), relative low water vapor transmission rate (WVTR) 0.2 g·mm/m2 ·day (Licari 2003), and good ion barrier properties (Szwarc 1976). This bilayer encapsulation approach takes full advantage of the highly effective moisture barrier property of ALD Al2O3, and parylene C to prevent the contact between alumina and liquid water, therefore drastically slowing kinetics of alumina dissolution. The parylene C also acts as an ion barrier, and ions can aid in the nucleation of a liquid water film at the substrate-film interface, which could accelerate failure of the ALD film. The bilayer encapsulation has demonstrated equivalent lifetime of over 6 years based IDE test structures with pA range leakage current without continuous voltage bias (Xie et al. 2013). Test structures are helpful in early stage studies for parameters and process optimization, but do not represent the full complexity of our device’s structure. Factors in neural interfaces like complex geometries (gold coils and SMD capacitors), materials and surfaces with different coefficient of thermal expansion and adhesion, and additional processing steps (oxygen plasma etching, BOE etching) are not fully represented in IDE test structures but significantly affect the performance of the bilayer encapsulation. Fully integrated wireless neural interfaces encapsulated with parylene C have been tested at room temperature (100 equivalent days at 37 °C) without continuous powering in phosphate buffered solution (PBS) (Sharma et al. 2012). Both the lower temperature (compared with body temperature) and lack of powering during the soak testing decrease the accuracy of simulating the aging process in vivo. Ideally, the soak testing should be performed at body temperature with periodic powering similar to that used in in vivo experiments. In reality, given the long lifetime of this A+P bilayer encapsulation based on IDE test structures, we performed soak testing at accelerated temperature (57 °C) with continuous powering, while parylene-only coated samples were tested at 37 °C. Additionally, to investigate the effect of bias voltage on devices lifetime, bilayer encapsulated UEA based neural interfaces were soaked at 57 °C without continuous powering. Analysis comparing these three different soaking conditions was used to estimate the improvements in lifetime of bilayer encapsulated neural interfaces, and the impact of temperature and continuous powering on lifetime.
2 Experimental details Figure 1 is a photograph of a Utah electrode array (UEA) based fully integrated wireless neural interface with an INI chip, SMD capacitors and gold wire-wound coil. The UEA has 100 channels, with electrode length of 1.5 mm and pitch of 400 μm. The fabrication details of the UEA are reported
Biomed Microdevices (2015) 17:1
Page 3 of 8 1
Powering coil INI-R9
UEA
SMD Caps
1 mm Fig. 1 Utah electrode array based fully integrated wireless neural interfaces, with flip-chip bond INI-R9 ASIC, gold coil for inductive powering, and SMD capacitors
elsewhere (Campbell et al. 1991; Bhandari et al. 2010). A newly designed 100-channel wireless neural recording IC, designated as INI-R9 (Integrated Neural Interface Recording version-9), was fabricated utilizing the 0.35 μm BiCMOS process (X-fab Semiconductor) and integrated with the array to provide signal processing and wireless capabilities. The details of the system integration were described elsewhere (Harrison et al. 2009; Kim et al. 2009a). The INI-R9 was flip-chip bonded to the backside of a 10×10 UEA using Au/ Sn reflow soldering. A 20-nF smoothing capacitor for DC power supply and a 56-pF resonating capacitor were used on the backside of UEA. A 2-mil Au (1 % Pd) wire-wound flat spiral coil of 5.5 mm in diameter was wire-bonded to resonate with SMD capacitor at 2.765 MHz and wirelessly power up the neural interfaces and receive telemetry data (Kim et al. 2009b). The fully integrated neural interfaces were then encapsulated with 52 nm of ALD Al2O3 and 6 μm of parylene C. Al2O3 was deposited by plasma-assisted (PA) ALD on integrated neural interfaces at a substrate temperature of 120 °C, which is within the thermal budget for the system at this point in the process. Details of the deposition process have been previously reported (Xie et al. 2013; Xie et al. 2012). A-174 (Momentive Performance Materials), an organosilane, was then used as adhesion promoter between the Al2O3 and parylene C layer. A 6-μm-thick parylene C layer was then deposited by chemical vapor deposition (CVD) using the Gorham process (Fortin and Lu 2004) on top of Al2O3. The electrode tips for neural recording were deinsulated by a threestep method. Laser ablation was used to remove the parylene C from the tips, followed by a 2 min O2 reactive ion etching (RIE) process to remove carbon and parylene residue from laser ablation process. The last process is an 8 min etch in buffered oxide etch (BOE), which was used to remove the 52-nm alumina film (Xie et al. 2014b). The neural interface was powered up through an inductive link from a custom-made transmit coil which uses a class E
power amplifier to create a 2.765 MHz waveform up to 80 Vrms at a 10 V supply. Figure 2 is a schematic drawing of the experimental setup for the soak testing, which was placed in an oven for temperature controlled measurements. Through a National Instrument (NI) board, the transmit coil was connected to a PC equipped with control software for controlling the devices through ASK commands, and receiving the telemetered data on the 900-MHz ISM band. The transmitted RF signal was detected with two different receiver systems. A spectrum analyzer with a monopole antenna was used to detect and show the RF spectra. Another custom-built wireless receiver board was used to display the frequency and signal strength of the RF signal and decode the real-time information in MATLAB on PC through a universal serial bus (USB) interface. The unregulated voltage from inductive link was monitored over the telemetry link, and the bit error rate was also reported by the receiver electronics using error-detection algorithms. The fully integrated neural interface was put into a 6-ml vial filled with 1× PBS. The vial was placed on top of the transmit coil for inductive powering, in an oven where the temperature can be control with accuracy of ±0.1 °C, as shown in Fig. 2.
3 Results and discussion 3.1 RF functionality The fully integrated and encapsulated neural interfaces were tested in PBS to check their wireless power/ command reception and FSK telemetry capabilities. Figure 3 shows the RF spectra transmitted from a fully integrated INI-R9 in PBS, obtained through the monopole antenna connected to the spectrum analyzer. Upon the reception of inductive power, the RF signal appeared on the spectrum analyzer at 909.75 MHz with signal strength of −79 dBm at the default channel 8 (Fig. 3 (a)), which demonstrates the wireless capability of the system. The telemetry frequency can be changed through commands issued by the MATLAB graphical user interface (GUI), resulting in an 8-MHz difference between adjacent channels as shown in Fig. 3. Figure 3 (b) presents the shifted telemetry frequency (917.6 MHz at channel 9). The RF signal strengths changed very little (79 vs 78 dBm) between those two channels. Responsiveness to telemetry frequency change commands in PBS indicates the telemetry is fully functional. 3 INI devices were inductively powered with unregulated 5 V continuously at 57 °C. For a fully functional INI device, the command reception rate is >90 % and the typical bit error rate for fully functional INI devices is
Effect of bias voltage and temperature on lifetime of wireless neural interfaces with Al2O3 and parylene bilayer encapsulation Xianzong Xie & Loren Rieth & Ryan Caldwell & Sandeep Negi & Rajmohan Bhandari & Rohit Sharma & Prashant Tathireddy & Florian Solzbacher
# Springer Science+Business Media New York 2015
Abstract The lifetime of neural interfaces is a critical challenge for chronic implantations, as therapeutic devices (e.g., neural prosthetics) will require decades of lifetime. We evaluated the lifetime of wireless Utah electrode array (UEA) based neural interfaces with a bilayer encapsulation scheme utilizing a combination of alumina deposited by Atomic Layer Deposition (ALD) and parylene C. Wireless integrated neural interfaces (INIs), equipped with recording version 9 (INI-R9) ASIC chips, were used to monitor the encapsulation performance through radiofrequency (RF) power and telemetry. The wireless devices were encapsulated with 52 nm of ALD Al2O3 and 6 μm of parylene C, and tested by soaking in phosphate buffered solution (PBS) at 57 °C for 4× accelerated lifetime testing. The INIs were also powered continuously through 2.765 MHz inductive power and forward telemetry link at unregulated 5 V. The bilayer encapsulated INIs were fully functional for ∼35 days (140 days at 37 °C equivalent) with consistent power-up frequencies (∼910 MHz), stable RF signal (∼−75 dBm), and 100 % command reception rate. This is ∼10 times of equivalent lifetime of INIs with parylene-only encapsulation (13 days) under same power condition at 37 °C. The bilayer coated INIs without continuous powering lasted over 1860 equivalent days (still working) at 37 °C. Those results suggest that bias stress is a significant factor to accelerate the failure of the encapsulated devices. The INIs failed completely within 5 days of the initial frequency shift of RF signal at 57 °C, which implied that the RF frequency shift is an early indicator of encapsulation/device failure. X. Xie (*) : L. Rieth : R. Caldwell : S. Negi : R. Bhandari : R. Sharma : P. Tathireddy : F. Solzbacher University of Utah, Salt Lake City, UT, USA e-mail: [email protected]
Keywords Neural interface . Utah electrode array . Atomic layer deposited (ALD) Al2O3 . Parylene . Bilayer encapsulation . Continuous bias voltage . Accelerated aging . Long-term reliability
1 Introduction Implantable neural interfaces have drawn tremendous interests from both researchers and clinicians for usage in neuroprosthetics to diagnose and treat neural disorders (Hochberg et al. 2006; Donoghue 2002; Santhanam et al. 2006; Wessberg et al. 2000; Normann 2007; Velliste et al. 2008). One representative example of this technology is the Utah electrode array (UEA), which is well developed and has FDA-clearance through an Investigation Device Exemption (IDE) and is able to record single unit signals (Normann 2007; Rousche and Normann 1998; Branner et al. 1999; Ledbetter et al. 2013). Wired UEAs continued to be deployed, which use gold wire bundles and percutaneous connectors, to transfer recording and stimulation signals. However, later findings show that wire bundles and percutaneous connectors are more likely to introduce foreign body response (Biran et al. 2007) and cause infections (Scott 2006), especially for chronic implantation. Moreover, wire connections and connector systems have been found to be one of the least reliable parts of neural interfaces due to mechanical stress, handling forces, etc. (Bulusu et al. 2013) and become the bottleneck of chronic applications. To eliminate wired connections, significant efforts have been devoted to develop wireless neural interfaces integrated with active electronics, capable of inductive powering and RF telemetry systems (Yin et al. 2006; Yin and Ghovanloo 2009; Wise et al. 2004; Harrison et al. 2009; Chestek et al. 2009; Kim et al. 2009a). Fully integrated neural
1
Page 2 of 8
interfaces (INIs) with capabilities of stimulating/recording from 100 channels have been developed at the University of Utah (Harrison et al. 2009; Kim et al. 2009a; Xie et al. 2014a). Neural interfaces are required to perform recording/ stimulation for years in vivo for chronic implantations in order to justify the risk associated with implantation. Encapsulation of neural interfaces is critical for their long-term functionalities. Water and ions permeate and accumulate at the interface between the encapsulation and the device, leading to open circuits, short circuits, corrosion of different components, and leakage current (DerMarderosian and Electrochemical migration of metals 1978; Webster 2009). The consequences are catastrophic and often lead to complete device failure. The significant bias voltage associated with integrated electronics further challenges the encapsulation by activating electrochemical degradation modes and the potential for accelerating ion transport. Two major approaches have been widely adopted to encapsulate implantable devices: hermetic enclosures and conformal encapsulation. Hermetic enclosures such as those used by implantable pulse generators (IPGs) are the most commonly used option to seal implantable devices (Dawes 1992). Electromagnetic power and data transmission and device miniaturization are challenges for traditional hermetic packaging. Conformal encapsulation has been widely developed and used, especially for small implants with electromagnetic wireless techniques. Potential encapsulation materials for neural interfaces include polyimide (Lago et al. 2005), parylene (Loeb et al. 1977; Hsu et al. 2009), silicone (Wu et al. 2000), amorphous silicon carbide (Hsu et al. 2007; Cogan et al. 2003), silicon nitride (Cogan et al. 2003), and diamond-like carbon (DLC) (Roy and Lee 2007). Several of these materials can be applied through thin film deposition techniques, with differing degrees of conformality and layer thicknesses. The high requirements make it almost impossible to find one material that meets all encapsulation requirements. Atomic layer deposited (ALD) Al2O3 and parylene C (A+ P) bilayer was found to be an excellent combination for encapsulating UEA based wireless neural interfaces, based on soaking results from interdigitated electrodes (IDEs) (Xie et al. 2013; Xie et al. 2012; Minnikanti et al. 2014). ALD alumina is well-known to be an excellent moisture barrier with water vapor transmission rates (WVTR) at the order of ∼10−10 g · mm/m2 · day with careful optimization (Ghosh et al. 2005; Langereis et al. 2006; Ferrari et al. 2007; Carcia et al. 2006), and is extremely conformal, allowing it to passivate complex surfaces/structures. However, alumina alone is not a suitable encapsulation since it dissolves in water (Potts et al. 2011), most likely due to the incorporation of hydrogen in the form of OH groups in the film as defects (Bulusu et al. 2013; Carcia et al. 2010). Parylene C has been widely used as a coating material for biomedical implantable devices (Hsu et al. 2009; Seymour et al. 2009; Hassler et al. 2010; Xie et al. 2011; Guenther
Biomed Microdevices (2015) 17:1
et al. 2013) due to attractive properties including chemically inertness, low dielectric constant (εr =3.15) (Fortin and Lu 2004), high resistivity (∼1015 Ω·cm), relative low water vapor transmission rate (WVTR) 0.2 g·mm/m2 ·day (Licari 2003), and good ion barrier properties (Szwarc 1976). This bilayer encapsulation approach takes full advantage of the highly effective moisture barrier property of ALD Al2O3, and parylene C to prevent the contact between alumina and liquid water, therefore drastically slowing kinetics of alumina dissolution. The parylene C also acts as an ion barrier, and ions can aid in the nucleation of a liquid water film at the substrate-film interface, which could accelerate failure of the ALD film. The bilayer encapsulation has demonstrated equivalent lifetime of over 6 years based IDE test structures with pA range leakage current without continuous voltage bias (Xie et al. 2013). Test structures are helpful in early stage studies for parameters and process optimization, but do not represent the full complexity of our device’s structure. Factors in neural interfaces like complex geometries (gold coils and SMD capacitors), materials and surfaces with different coefficient of thermal expansion and adhesion, and additional processing steps (oxygen plasma etching, BOE etching) are not fully represented in IDE test structures but significantly affect the performance of the bilayer encapsulation. Fully integrated wireless neural interfaces encapsulated with parylene C have been tested at room temperature (100 equivalent days at 37 °C) without continuous powering in phosphate buffered solution (PBS) (Sharma et al. 2012). Both the lower temperature (compared with body temperature) and lack of powering during the soak testing decrease the accuracy of simulating the aging process in vivo. Ideally, the soak testing should be performed at body temperature with periodic powering similar to that used in in vivo experiments. In reality, given the long lifetime of this A+P bilayer encapsulation based on IDE test structures, we performed soak testing at accelerated temperature (57 °C) with continuous powering, while parylene-only coated samples were tested at 37 °C. Additionally, to investigate the effect of bias voltage on devices lifetime, bilayer encapsulated UEA based neural interfaces were soaked at 57 °C without continuous powering. Analysis comparing these three different soaking conditions was used to estimate the improvements in lifetime of bilayer encapsulated neural interfaces, and the impact of temperature and continuous powering on lifetime.
2 Experimental details Figure 1 is a photograph of a Utah electrode array (UEA) based fully integrated wireless neural interface with an INI chip, SMD capacitors and gold wire-wound coil. The UEA has 100 channels, with electrode length of 1.5 mm and pitch of 400 μm. The fabrication details of the UEA are reported
Biomed Microdevices (2015) 17:1
Page 3 of 8 1
Powering coil INI-R9
UEA
SMD Caps
1 mm Fig. 1 Utah electrode array based fully integrated wireless neural interfaces, with flip-chip bond INI-R9 ASIC, gold coil for inductive powering, and SMD capacitors
elsewhere (Campbell et al. 1991; Bhandari et al. 2010). A newly designed 100-channel wireless neural recording IC, designated as INI-R9 (Integrated Neural Interface Recording version-9), was fabricated utilizing the 0.35 μm BiCMOS process (X-fab Semiconductor) and integrated with the array to provide signal processing and wireless capabilities. The details of the system integration were described elsewhere (Harrison et al. 2009; Kim et al. 2009a). The INI-R9 was flip-chip bonded to the backside of a 10×10 UEA using Au/ Sn reflow soldering. A 20-nF smoothing capacitor for DC power supply and a 56-pF resonating capacitor were used on the backside of UEA. A 2-mil Au (1 % Pd) wire-wound flat spiral coil of 5.5 mm in diameter was wire-bonded to resonate with SMD capacitor at 2.765 MHz and wirelessly power up the neural interfaces and receive telemetry data (Kim et al. 2009b). The fully integrated neural interfaces were then encapsulated with 52 nm of ALD Al2O3 and 6 μm of parylene C. Al2O3 was deposited by plasma-assisted (PA) ALD on integrated neural interfaces at a substrate temperature of 120 °C, which is within the thermal budget for the system at this point in the process. Details of the deposition process have been previously reported (Xie et al. 2013; Xie et al. 2012). A-174 (Momentive Performance Materials), an organosilane, was then used as adhesion promoter between the Al2O3 and parylene C layer. A 6-μm-thick parylene C layer was then deposited by chemical vapor deposition (CVD) using the Gorham process (Fortin and Lu 2004) on top of Al2O3. The electrode tips for neural recording were deinsulated by a threestep method. Laser ablation was used to remove the parylene C from the tips, followed by a 2 min O2 reactive ion etching (RIE) process to remove carbon and parylene residue from laser ablation process. The last process is an 8 min etch in buffered oxide etch (BOE), which was used to remove the 52-nm alumina film (Xie et al. 2014b). The neural interface was powered up through an inductive link from a custom-made transmit coil which uses a class E
power amplifier to create a 2.765 MHz waveform up to 80 Vrms at a 10 V supply. Figure 2 is a schematic drawing of the experimental setup for the soak testing, which was placed in an oven for temperature controlled measurements. Through a National Instrument (NI) board, the transmit coil was connected to a PC equipped with control software for controlling the devices through ASK commands, and receiving the telemetered data on the 900-MHz ISM band. The transmitted RF signal was detected with two different receiver systems. A spectrum analyzer with a monopole antenna was used to detect and show the RF spectra. Another custom-built wireless receiver board was used to display the frequency and signal strength of the RF signal and decode the real-time information in MATLAB on PC through a universal serial bus (USB) interface. The unregulated voltage from inductive link was monitored over the telemetry link, and the bit error rate was also reported by the receiver electronics using error-detection algorithms. The fully integrated neural interface was put into a 6-ml vial filled with 1× PBS. The vial was placed on top of the transmit coil for inductive powering, in an oven where the temperature can be control with accuracy of ±0.1 °C, as shown in Fig. 2.
3 Results and discussion 3.1 RF functionality The fully integrated and encapsulated neural interfaces were tested in PBS to check their wireless power/ command reception and FSK telemetry capabilities. Figure 3 shows the RF spectra transmitted from a fully integrated INI-R9 in PBS, obtained through the monopole antenna connected to the spectrum analyzer. Upon the reception of inductive power, the RF signal appeared on the spectrum analyzer at 909.75 MHz with signal strength of −79 dBm at the default channel 8 (Fig. 3 (a)), which demonstrates the wireless capability of the system. The telemetry frequency can be changed through commands issued by the MATLAB graphical user interface (GUI), resulting in an 8-MHz difference between adjacent channels as shown in Fig. 3. Figure 3 (b) presents the shifted telemetry frequency (917.6 MHz at channel 9). The RF signal strengths changed very little (79 vs 78 dBm) between those two channels. Responsiveness to telemetry frequency change commands in PBS indicates the telemetry is fully functional. 3 INI devices were inductively powered with unregulated 5 V continuously at 57 °C. For a fully functional INI device, the command reception rate is >90 % and the typical bit error rate for fully functional INI devices is
