Highly Conformable Chip-in-Foil Implants for Neural Applications

doi:10.21203/rs.3.rs-2373599/v1

Download PDF

Article

Highly Conformable Chip-in-Foil Implants for Neural Applications

https://doi.org/10.21203/rs.3.rs-2373599/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 09 May, 2023

Read the published version in Microsystems & Nanoengineering →

You are reading this latest preprint version

Demands on neural interfaces in terms of functionality, high spatial resolution, and longevity have recently increased. These requirements can be met with sophisticated silicon-based integrated circuits. Embedding miniaturized dice in flexible polymer substrates significantly improves the adaptation to the mechanical environment in the body and thus the systems’ structural biocompatibility as well as the ability to cover larger areas of the brain. This work addresses main challenges in developing a hybrid chip-in-foil neural implant. Assessments were related to: first, the mechanical compliance to the recipient tissue that allows a long-term application, and second, the suitable design that allows the implant’s scaling and modular adaptation of chip arrangement. Finite element model studies were performed to identify design rules regarding die geometry, interconnect routing, and positions for contact pads on dice. Providing edge fillets in the die base shape was an effective measure to improve die-substrate integrity and increase the area available for contact pads. Furthermore, the routing of interconnects in the immediate vicinity of die corners should be avoided, as the substrate in these areas is prone to mechanical stress concentration. Contact pads on dice should be placed with a clearance from the die rim to avoid delamination when the implant is conformed to a curvilinear body.

A microfabrication process was developed to transfer, align and electrically interconnect multiple dice into conformable polyimide-based substrates. The process enabled arbitrary die shape and size and independent target positions on the conformable substrate from the die position on the fabrication wafer.

Physical sciences/Engineering/Electrical and electronic engineering

Physical sciences/Nanoscience and technology/Nanoscale devices/Electronic devices

Chip-in-Foil

Conformable Substrate

Microfabrication

Hybrid Implants

Bioelectronic Devices

Polyimide

Implantable Flexible System

Trends in neurotechnology today are moving towards mechanically compliant, miniaturized implants with high integration density and targeted lifetimes of several decades. A large number of neural implants such as brain pacemakers ¹, cochlear implants ^2,3, or brain-computer interfaces ⁴ are subject of research or being used in rehabilitation medicine to restore lost body functions, promote health and improve the life quality of patients by providing a link between the human nervous and a technological system ^5–7. Better understanding and cross-curricular work in the fields of neuroscience, biotechnology, material sciences and related engineering disciplines paved the way for advancements in such systems.

A high level of complexity can be realized by sophisticated silicon (Si)-based integrated circuits (ICs) ^8–10. High integration density enables high spatial resolution at the cost that only small areas of the brain can be covered. However, mechanically stiff Si induces a foreign body response leading to tissue encapsulation of implants and insulation of electrode contacts from neurons which, in return, compromises functionality in chronic studies and applications ^11,12. It has been shown that this effect can be minimized by matching the implant’s mechanical properties to the host tissue and making the implant conformable to adapt to the biological surface shape. Conformability leads to intimate contact and, therefore, high efficacy and minimal invasiveness ¹³. Polyimides (PI) exhibit excellent chemical stability and biocompatibility, desirable mechanical properties, low water uptake leading to reduced plasticization, high electric resistivity and dielectric strength and are among the most used substrate materials for new generation mechanically compliant implants ^14–17. They are excellent substrate materials for flexible neural implants as they are able to conform to curvilinear structures such as the brain ¹⁸, are compatible with microfabrication processes and can be manufactured in sub micrometer thickness, and their high performance and long-term stability have been shown in several studies ^19–25. Therefore the combination of mechanically compliant PI-based substrates and small Si-based circuits was proposed to develop bioelectronic devices with complex functionality and good adaption good to the biomechanics of the host tissue^26,27 that are able to cover larger brain areas. Intrinsically flexible electronics were made possible with organic transistors and circuits and improvements in stability, device performance, cost, and fabrication methods have been made in the past years. Utilizing such could meet requirements concerning mechanical flexibility, however, they currently do not offer the integration density and complexity possible with mature and reliable Sibased technology ^28–30.

Anatomical boundary conditions in terms of available space and desired electrode count or location for recording or stimulation of neural signals may vary between target regions, species or even individuals ^31,32. Thus, a modular arrangement of circuit components can be beneficial by enabling fast adaption of the implant design to address a large number of established and new fields of application. These comprise monitoring of large cortical areas, organ healing processes, or bridging functions in the peripheral nervous system - highly specialized prototypes for diverse scientific questions and medical treatments.

An implant should always be designed taking into account the mechanical and geometrical conditions of the target tissue ³³, such as the softness of the brain in general and the curvature of the brain region of interest. The radius of curvature in the sensorimotor cortex of a rat, for example, can range from several hundred microns to a few millimeters ³¹, while radii on the cortex of a human brain are much larger. Conformability is the ability of a thin membrane or foil to attach to a body and take on its shape ³⁴, and due to better conformability, closer contact between electrodes in the implant and the cortex can be achieved ⁵. In addition to foil thickness, conformability depends on many other factors: its mechanical properties, the geometries of the implant and the cortex and the surface tension of the wetting liquid ³⁵, in this application, the biological fluid between brain and implant. Therefore, the geometry, size and stiffness of implants should be tailored to the specific application to reduce the biotic/abiotic gap and achieve recordings with a high spatial resolution ³⁶. Hybrid bioelectronic devices, consisting of sophisticated ICs modularly distributed on miniaturized flexible polymeric substrates, are a promising approach to overcome a central challenge in neural implants: the combination of complex functionality at high channel density and longevity in the harsh biological environment. The virtue of the proposed method is the possibility of scaling down the size of Si-based dice and distributing tasks between dice connected by thin-film interconnect lines on the flexible substrate. This is accompanied by a lower bending stiffness compared to a single but larger die and good adaptability of the technical system in terms of the range of function, substrate dimensions and target anatomy. This work aims to formulate design rules regarding implant geometries and thin-film routing with the background question of the existence of critical points for mechanical stress in the implant. The question is concerning system integrity and good electrode-tissue contact essential to be asked with the scope of developing a chronic implant. Furthermore, contact pads on dice should be positioned in areas without critical stress/delamination due to the substrate conformation to the curvilinear brain surface. Reliable implant function can be achieved if no delamination of dice from the substrate or interconnect damage takes place.

A finite element model (FEM) simulation was developed to describe the hybrid implant's mechanical behavior when conformed to a brain surface with the aim of design optimization. The brain area exemplarily considered in this study was the somatosensory cortex of rats, approximated as a cylinder with a 3 mm radius of curvature ³¹. Furthermore was the development of a transfer process for Si-based dice into flexible PI-based substrates and their three-dimensional placement accuracy subject of investigation. Results were combined to manufacture prototype implants (Fig. 1).

2.1 Design Assessment: Implant Geometry

The mechanical stress at the die-substrate interface concerning the variation of edge fillet radius was determined. The maximum delamination height between the flexible PI substrate (thickness of 10 µm) and the die (thickness of 100 µm) was taken as an indicator of how large the effect due to the filleted die corners on the total delamination is. The possible deformation of the PI substrate increases with delamination. Therefore, a more considerable maximum delamination height can be associated with a larger delaminated area. With larger edge filleting (and lower maximum delamination), a lower distance between the delamination edge and the die edge was observed (in the XYplane). The effective area for a die varied with the fillet radius (Figure 2a). Most delamination (>1 nm) was observed at the die edge. However, delamination could be found further towards the die center. Maximum delamination of 24 nm occurred along the die edge with r_edge = 20 µm at a distance of up to 12.8 µm towards the die center, or 14.1 nm at 8.2 µm for r_edge = 50 µm. Since even minor delamination can damage the electrical contacts between pads on dice and interconnects in the PI substrate, the largest distance from the die edge to the individual points of delamination was subtracted from the edge length of the die to determine an effective die length. This resulted in a usable area of 87.8% of the total die area for r_edge = 20 µm and 92.3% for r_edge = 50 µm. The die with 50 µm fillet radius has a lower geometrical area (0.158 mm²) compared to the die with 20 µm radius.

However, the effective area of 0.146 mm², caused by rotation around α_Z along with delamination, is more extensive than for the die with lower fillet radius. Even larger fillets could lead to loss of die area that is not compensated by the gain in effective die area, which must be evaluated accordingly. Local stress concentration in the PI substrate occurred at the corners of the die (Figure 2b) and was distributed along with the edge fillets. The distribution of stress at the corners of the interface was similar in size and magnitude for different radii (between 1 µm and 50 µm). Based on the simulated stress distribution, conclusions can be drawn for interconnect placement on the PI substrate. Since areas with most prominent stress were observed in the vicinity of corners for all variations of the fillet radius, interconnect routing should be avoided in this area as much as possible. The assumption of a die misaligned by 10° to the curvatures orthogonal was made to include both process and implantation tolerances to establish design rules for a possible use case. A more considerable effective die length and area could be expected with lower misalignment.

2.2 Transfer Technology

The transfer of Si-based dice into mechanically flexible PI-based substrates was performed using a carrier equipped with cavities that dice were placed in and subsequent levelled microfabrication of mechanically conformable substrates directly on the carrier. Galvanic deposition of the sacrificial copper (Cu) layer on the cavity sidewalls for die release at the end of the process sequence was successfully performed (Fig. 3a). Cu (5.81 ± 1.45 µm, n = 75, determined with optical micrographs and ImageJ software) was deposited over the entire cavity height down to the carrier backside surface (inset in Fig. 3a). Deposition was intentionally limited to the cavity sidewalls; the dicing tape cover averted deposition on the frontside surface and no backside deposition occurred due to the bare Si surface. Frontside deposition was not carried out, as the electrodeposited layer had surface roughness orders of magnitude larger than the polished Si wafer ³⁷, which would be transferred to the subsequent PI and metallization layers. Dry etching of the cavities was done with an aluminum etch stop to prevent passivation gas deposition on the carrier backside that potentially lower the adhesion of the backside PI (intended for die fixation during steps 3 to 7 in Fig. 6). The cavities were designed slightly larger than the dice which ensured uncompromised insertion and tight fit. A substrate backside that had been levelled by mechanical grinding and sufficiently cleaned enabled a well-adhering backside PI, which was essential for subsequent process steps allowing fixation on vacuum chucks during spin coating, mask alignment or dry etching. Spin coating of negative tone photoresist over the PI-filled gap between dice and carrier was successful, and the photoresist mask for interconnects exhibited sharp undercuts on the surface of dice (Fig. 3b, position marked in Fig. 3c) and in the gap. A rather thick photoresist (4 µm maN1440) was chosen for patterning the metallization layer to enable interruptionfree coating over the height topography between dice and carrier substrate, which is essential for continuous interconnection. According to the simulation results from the design assessment in section 2.1 no interconnect lines were routed directly over die corners to circumvent an area of stress concentration in the substrate that could compromise the function of this particular interconnect. Lithography led to satisfactory results and PVD was able to reliably form all 35 intended interconnect lines (Fig. 3c). Metal interconnects had a minimum width of 30 µm and were deposited interrupt-free over the gap (between carrier and dice, Fig. 3d) and contact pads on dice by sputter deposition of a WTi/Au/WTi stack and subsequent lift-off technique. Metal interconnects were well-integrated into PI substrates (Fig. 3e) and formed continuous vias through the dry-etched PI for die contact (Fig. 3f). PI above the contact pads was opened with beveled sidewalls (α_Via = 61° ± 0.9°, n = 5) to facilitate a uniform PVD coating from the PI base layer surface down to the dice contacts. Thinning of dice and carrier backside was performed without displacement or tearing out of dice or impairment of Cu sacrificial layer between cavity and dice sidewalls (Fig. 3g). The UVtape was exposed after thinning and removed without any displacement of PI-based samples or damage to the carrier. Large dice on PI substrates with 35 sputtered interconnects were released from the carrier and conformed well to the spherical surface of a wetted glove (Fig. 3h). SEM on the PI surface of dice that were mechanically sheared off after thermally accelerated ageing at 60° C in phosphate-buffered saline showed seamless and void-free integration (Fig. 3h). The close-up view shows the imprint topography in the PI due to metal structures on the die after adhesive failure. No signs of air inclusions in the PI underneath the dice could be observed. As shown in the SEM images, the PI surface is smooth without entrapped voids. This indicates a properly cured polymer, most likely facilitated by allowing one side of the PI layer to exchange with the surrounding atmosphere while annealing. Good adhesion between dice and flexible substrate as well as tight and void-free contact is desirable to increase the system's robustness against mechanical stress. The main advantage between the process sequence presented here and other approaches is the embedding of dice into PI before annealing at high temperatures. This eliminated the need for additional adhesive material, such as BCB, which was used in several other approaches as a fixative on already cured PI/polymer substrates ^38,39. With this process, a maximum number of 34 dice were aligned, transferred and interconnected using a single carrier. Hybrid devices consisting of 10 µm thick PI-based substrates and three Si-based dice, each 390 x 390 µm² and 24 µm thin (interconnected by sputtered thin-film metal), could be released from the carrier and were highly flexible (Fig. 1c).

2.3 Accuracy Assessment: Die Placement

Deviation from the ideal die position in the X-, Y- and Z-direction was determined for variation of the cavity length in the transfer carrier (Figure 4a). Average deviations were lowest for the smallest cavity and tend to increase with cavity length, with the lowest mean in-plane translation for the 417 µm cavity of 5.13 µm ± 3.06 µm (n = 90) and a deviation in Zdirection of 4.26 µm ± 2.53 µm (n = 45). A similar trend was observed for die rotation (Figure 4b), with a mean in-plane rotation of 0.80° ± 0.49° (n = 45) and mean out-of-plane rotation of 0.90° ± 0.50° (n = 90) and the valley height h_valley between PI at substrate level and inside the gap between transfer carrier and die (Figure 4c) of 8.75 µm ± 4.93 µm (n = 180). The critical rotation angle α_Z,max for this case, was 8.97°, according to Equation (2), with a critical translation of 24.8 µm. The rotational misalignment around the X- and Y-axis was > 0° (Figure 4b), possibly due to the carrier bow in combination with a not perfectly flat surface during insertion of dice. If both were perfectly flat and parallel, out-of-plane die rotation would have been prevented. For the smallest cavity of 417 µm length, h_valley was significantly lower (0.05 level) than for the other cavity lengths. From the analysis of positioning parameters, a trend toward a cavity size of 417 µm for the lowest misalignment could be drawn. Critical rotation and translation were prevented by the geometrical design of dice- and cavity length and critical distance (in this case, between the edge of the contact pad opening and the adjacent interconnect line). Tailoring these parameters led to maximum possible values below critical rotation and translation. With increasing distance r_pad of pads from the die center, the risk of misalignment becomes more prominent. Therefore, well suited cavities will push the limits towards smaller pad sizes and pitches and thus function density of the embedded ICs.

A measure to improve die integrity on the flexible PI-based substrate and increase the area available for contact pads is to provide radius fillets in the design of the die base shape. Maximum delamination of 24 nm occurred along the die edge with r_edge = 20 µm at a distance of up to 12.8 µm towards the die center, or 14.1 nm at 8.2 µm for r_edge = 50 µm which resulted in an usable area of 87.8% of the total die area for r_edge = 20 µm and 92.3% for r_edge = 50 µm. Further adaption in the die shape could be made concerning the anatomical boundary conditions of a specific application. Since the flexible substrate in the immediate vicinity of die corners is prone to mechanical stress concentration, interconnect lines should not be routed in these areas. Additionally, contact pads on dice should be placed with a clearance from the die rim since the substrate was more affected by delamination in this area when the implant conformed to a curvilinear body. This clearance depends on the radii of curvature of the target region and die base shape. In the exemplary study provided here (for a 400 µm by 400 µm die on a cylindrical surface with a radius of curvature of 3 mm), a clearance of 10 µm was suggested.

The developed process was based on the placement of individual dice into cavities of a carrier wafer that allowed for subsequent levelled microfabrication of the PI- and metal thin-film-based substrate. Application of the PI in its liquid form by spin coating and the annealing directly on dice led to a strong (59.1 ± 13.7 MPa ⁴⁰, comparable with die fixation by polymeric adhesive) and void-free contact between dice and substrate without additional adhesion promoters. This enabled a thickness of the substrate in the lower micrometer range, which is a prerequisite for conformable PI substrates. The mean translation was as low as 5.13 µm ± 3.06 µm (n = 90) (in-plane) and 4.26 µm ± 2.53 µm (n = 45) (out-of-plane) and a mean rotation down to 0.80° ± 0.49° (n = 45) (in-plane) and 0.90° ± 0.50° (n = 45) (out-of-plane) was achieved for dice of 390 µm edge length. Handling thin and fragile dice was circumvented by grinding dice to the desired thickness at the end of the process sequence (minimum thickness of 24 µm).

Hybrid bioelectronic devices, consisting of sophisticated ICs modularly distributed on miniaturized flexible polymeric substrates, are a promising approach to overcome a central challenge in neural implants: the combination of complex functionality at high channel density and longevity in a biological environment. The virtue of this method is the possibility of scaling down the size of Si-based dice and distributing tasks between dice connected by thin-film interconnect lines on the flexible substrate. This is accompanied by a lower bending stiffness compared to a single but larger die and good adaptability of the technical system in terms of the range of function, substrate dimensions and target anatomy. It has now been defined how ASICs can be integrated into mechanically compliant PIbased substrates, taking into account the previously defined requirements. The next steps are to (a) test the function of the system acutely in vivo, (b) verify the stability in animal models, and (c) validate the assumptions on structural biocompatibility. The conclusions drawn in this work and the developed processes lay the foundation for further development in next-generation neural implants that allow the adaption to a wide range of anatomical boundary conditions and may have significant value for pre-clinical studies on the long-term application of hybrid implants.

5.1 Design Assessment: Implant Geometry.

Simulations were done with COMSOL Multiphysics 5.3 (COMSOL Inc., Sweden) under the solid mechanics module and studied under stationary conditions. The deformation of a rat´s cortex was modelled thoroughly by ¹⁸ with validation in vivo. This existing model was used as starting point for the approximation of the mechanical interaction between implant (substrate was modelled as a rectangle with a length of 2 mm and thickness of 10 µm) and cortex which was 2D approximated as a quarter circle (mirrored at x = 0) with a radius of 3 mm (a radius found in the somatosensory cortex of rats ³¹). Both bodies were assumed linearly elastic with Young's modulus E, Poisson’s ratio v and density ⍴ (Table 1 ^18,41). Deformation of the PI substrate due to surface tension and capillary forces was approximated as a point load of 0.21 mN to the outer lower edge of the substrate, perpendicular to the tangent of the deformed geometry. Domains were connected through triangular elements, which were refined gradually from 80 µm to an element size of 4 µm on the cortex surface with a conformability threshold set to 90% of the array’s length in contact with the cortex. Cortex depression, curvature along the surface and cortex shape after loading were studied.

A die was modelled with corners filleted by a radius of curvature r_edge (varied between 20 µm and 50 µm) on a PI substrate, and rotation of an angle α_Z around the die center in the XYplane was varied between 0° and 45° (Figure 5b). The PI substrate was deformed and the implant was dimensioned as follows: l_die = 400 µm, t_die = 100 µm, t_PI = 10 µm, gap length = 415 µm. Substrate and die were connected as assembly in COMSOL, adhesive strength was set to 58 MPa (according to experimental shear-test data ⁴⁰) and von Mises stress on the interface between substrate and die was investigated in a parametric study. A load of 50 Pa was applied on the entire substrate bottom surface to account for deformation due to surface tension of the cerebrospinal fluid alongside the substrate ¹⁸.

5.2 Transfer Technology.

The transfer process was based on placing individual dice into cavities in a carrier which allowed further levelled standard microfabrication steps, e.g. photolithography, dry etching, physical vapor deposition (PVD) and mechanical grinding. Precise positioning was guaranteed, and setting the relative contact pad position on all dice prevented contact shorting or pads not connected by interconnects deposited in batch. Handling of ultra-thin and fragile dice was averted by thinning at the end of the process sequence.

For carrier fabrication, a 4inch Si wafer with a thickness of 525 µm, low bow (< 20 µm) and low warp (< 20 µm) was used. Cavities were dry-etched through the wafer by deep reactive ion etching (DRIE, alternating SF₆ etching and C₄F₈ passivation) (SPTS Technologies, United Kingdom) using a photo-resist mask (26 µm AZ 10XT, MicroChemicals GmbH, Germany). Next, a gold (Au) seed layer for galvanic deposition of Cu was applied from the front side under carrier rotation. To ensure that galvanic deposition (and therefore increase in surface roughness) only takes place on the cavity sidewalls, the seed layer on the carrier frontside was protected by dicing tape (SWT 10 + R, Nitto Denko, Japan). Subsequently, 6 µm of Cu was deposited which was chosen as a sacrificial layer since it could be selectively etched against other materials employed to facilitate the die release after device finalization. After mechanical removal of the tape, the seed layer was removed from the front side by reactive ion etching (RIE) in argon plasma at 200 W (SPTS Technologies, United Kingdom). UV-light curable back grinding tape (E3125 KL, LINTEC Corporation Advanced Materials Operations, Japan) was mounted on the frontside to fixate dice throughout the following process steps. Dice were placed inside the cavities with their active side (with connection pads) fixated by the tape on the carrier frontside (Figure 6-1). Subsequently, the backside of carrier and dice were levelled with an automatic surface grinder (Figure 6-2) (DAG810, Disco Corporation, Japan). On the levelled backside, a 4 µm thin PI (BPDAPPD) layer was applied (Figure 6-3). Spin coating of PI was performed with a 2 min hold period before and after rotation to allow the PI to level and cover height topographies on the wafer. To exclude the grinding tape from the temperature impact of the annealing process, it was exposed to UV-light (RAD 2000, LINTEC Corporation Advanced Materials Operations, Japan) and peeled off after PI soft-curing at 90 °C for 180 s on a hotplate. Thus, the PI backside layer replaced the tape as die fixation throughout the following process steps. Finally, the annealing of PI on the backside was done at 450 °C for 10 min (YES PB HV, Yield Engineering Systems, USA) in nitrogen atmosphere (Figure 64). On the frontside, 4 µm of PI was applied as implant base layer (Figure 65) and opened above the die contact pads by RIE at 100 W in oxygen plasma (Figure 66) using a photoresist mask (AZ 10XT). To facilitate a uniform PVD coating from the PI base layer surface down to the die contacts resist underexposure was performed to transfer beveled sidewalls into the PI openings. After resist stripping, sputter deposition of the metal layer stack (50 nm / 300 nm / 50 nm) of tungsten titanium (WTi), Au and WTi was performed (Star 100 PentaCo, FHR Anlagenbau GmbH, Germany) using 4 µm of negative tone photoresist maN1440 (micro resist technology GmbH, Germany) and liftoff technique to remove excess material. The PI surface was activated in oxygen plasma in an RIE machine for 45 s at 80 W. An oxygen plasma surface activation was performed to enhance adhesion between the PI base and the hereafter applied PI top layer (4 µm), which was opened by RIE above the contact pads and between individual devices using a photoresist mask (AZ 10XT). At this point, device structuring was complete (Figure 67), and RIE at 100 W could remove the PI backside layer in oxygen plasma. Afterwards, the dice were thinned together with the carrier from the backside to the desired thickness (Figure 6-8) and cleaned. In correspondence with the FEM simulation grinding was carried out down to 100 µm, however, to validate the transfer process at the lower extreme value of the die thickness, grinding to 25 µm (thickness of the active electronics) was also tested. After exposure, the carrier was removed from the UV-curable tape (Figure 6-9), and finally, the devices were released by dissolving the Cu-sacrificial layer selectively with ammonium persulfate solution and rinsed with deionized water (Figure 6-10). Focused ion beam (FIB) cutting and scanning electron microscopy (SEM) of surfaces and cross-sections was performed in a Scios 2 HiVac (Thermo Fisher Scientific, Waltham, MA, USA). Methods for thermally accelerated ageing and mechanical shearing of dice are presented elsewhere⁴⁰.

5.3 Accuracy Assessment: Die Placement.

Assessments of possible die rotation with given process parameters were performed, which can be used to optimize die design in terms of edge, size and shape of contact pads. The risk of misalignment between different layers of the device during fabrication is aimed to be minimized to prevent possible misconnection. Figure 7a sketches a die and two contact pads with an outer edge at a distance r_pad from the center of the die (Fig. 7b). The die was placed in a cavity of length l_cavity and rotated by an angle α_Z from the ideal orientation with respect to the cavity sidewall.

Figure 7 Assessment of critical positioning errors. Schematic representation of (a) a die rotated by α_Z, which leads to an electrical shortage of two contact pads by an interconnect line, (b) location relations with the dashed line indicating the position of the contact pad after rotation and (c) translation error leading to an electrical shortage of two contact pads by an interconnect line.

The critical angle α_Z,crit, which would lead to a misalignment causing a shortage of contact pads by interconnects, was calculated as follows: x_crit was the deviation in Xdirection caused by rotation that would lead to contact pads shorted by an interconnect. Under the assumption of small angles, this distance could be approximated by the base length of the isosceles triangle l_b, spanned by rotation of r_pad around α_Z (Equation (1)). Solving for α_Z and applying the critical distance leads to the critical rotation angle α_Z,crit (Equation (2)). The critical inplane translation could be extracted as the distance between pads (Figure 7c).

$${l}_{b}=2\bullet {r}_{pad}\bullet \text{sin}\left(\frac{{\alpha }_{Z}}{2}\right)$$

$${\alpha }_{\text{Z},\text{c}\text{r}\text{i}\text{t}} =2\bullet {\text{sin}}^{-1}\left(\frac{{x}_{\text{c}\text{r}\text{i}\text{t}}}{2\bullet {r}_{\text{p}\text{a}\text{d}}}\right)$$

Inplane rotational error α_Z and translational errors Δx and Δy of dice were analyzed with optical micrographs and ImageJ software (imagej.net, Wayne Rasband, USA) for different cavity lengths l_cavity. The die center deviation in Zdirection from the ideal position in the XYplane Δz and the die tilt α_X and α_Y around the X and Yaxis were extracted from topography scans with a mechanical profilometer (Dektak 150, Veeco Instruments Inc., USA).

AUTHOR INFORMATION

Corresponding Authors

^*Calogero Gueli – Institut für Mikrosystemtechnik, Universität Freiburg, 79110 Freiburg, Germany, Email: [email protected]

Thomas Stieglitz – Institut für Mikrosystemtechnik, Universität Freiburg, 79110 Freiburg, Germany, Email: [email protected]

Author Contributions

C.G. conceptualized and designed the study, manufactured the devices, conducted experiments regarding manufacturing and mechanical characterization, analyzed the data, supervised and conceptualized the mechanical simulation and wrote the manuscript, M.S. designed the ASIC, M.E. and M.O. acquired funding, M.E. co-conceptualized the manufacturing process, co-designed the study, reviewed the manuscript and gave scientific advice, J.M. manufactured devices, conducted shear tests, C.G. and J.M. analyzed respective data, N.F. implemented the mechanical simulation, T.S. gave scientific advice, co-designed and reviewed the manuscript. All authors have given approval for the final version of the manuscript.

Funding Sources

This work was funded by the German Research Foundation (DFG, grant no. 401023906 and OR 245/15-1) and is part of BrainLinks-BrainTools which was funded by the German Research Foundation (DFG, grant no. EXC 1086) and is currently funded by the Federal Ministry of Economics, Science and Arts of Baden Württemberg within the sustainability program for projects of the excellence initiative.

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENT

The authors thank the colleagues at IMTEK and UUlm for their technical support.

Little, S. et al. Adaptive deep brain stimulation in advanced Parkinson disease. Ann. Neurol. 74, 449–457; 10.1002/ana.23951 (2013).
Shepherd, R. K., Hatsushika, S. & Clark, G. M. Electrical stimulation of the auditory nerve: The effect of electrode position on neural excitation. Hear. Res. 66, 108–120; 10.1016/0378-5955(93)90265-3 (1993).
Clark, G. A History. In Cochlear Implants, edited by G. Clark (Springer-Verlag, New York, 2003), Vol. 405, pp. 1–57.
Abiri, R., Borhani, S., Sellers, E. W., Jiang, Y. & Zhao, X. A comprehensive review of EEG-based brain-computer interface paradigms. J. Neural Eng. 16, 11001; 10.1088/1741-2552/aaf12e (2019).
Fallegger, F., Schiavone, G. & Lacour, S. P. Conformable hybrid systems for implantable bioelectronic interfaces. Adv. Mater. 32; 10.1002/adma.201903904 (2020).
Navarro, X. et al. A critical review of interfaces with the peripheral nervous system for the control of neuroprostheses and hybrid bionic systems. J. Peripher. Nerv. Syst. 10, 229–258; 10.1111/j.1085-9489.2005.10303.x (2005).
Sahyouni, R. et al. Interfacing with the nervous system: a review of current bioelectric technologies. Neurosurg. Rev. 42, 227–241; 10.1007/s10143-017-0920-2 (2019).
Jun, J. J. et al. Fully integrated silicon probes for high-density recording of neural activity. Nature 551, 232–236; 10.1038/nature24636 (2017).
Sayed Herbawi, A. et al. CMOS neural probe with 1600 close-packed recording sites and 32 analog output channels. J. Microelectromechanical Syst. 27, 1023–1034; 10.1109/JMEMS.2018.2872619 (2018).
Sporer, M., Reich, S., Kauffman, J. G. & Ortmanns, M. A direct digitizing chopped neural recorder using a body-induced offset based DC servo loop. IEEE Trans. Biomed. Circuits Syst. PP; 10.1109/TBCAS.2022.3177241 (2022).
Polikov, V. S., Tresco, P. A. & Reichert, W. M. Response of brain tissue to chronically implanted neural electrodes. J. Neurosci. Methods 148, 1–18; 10.1016/j.jneumeth.2005.08.015 (2005).
Barrese, J. C. et al. Failure mode analysis of silicon-based intracortical microelectrode arrays in non-human primates. J. Neural Eng. 10; 10.1088/1741-2560/10/6/066014 (2013).
Minev, I. R. et al. Electronic dura mater for long-term multimodal neural interfaces. Science 347, 159–163; 10.1126/science.1260318 (2015).
Schweitzer, P. A. Corrosion engineering handbook. Corrosion of polymers and elastomers. 2nd ed. (CRC Press, Boca Raton, 2007).
Boretius, T. et al. A transverse intrafascicular multichannel electrode (TIME) to interface with the peripheral nerve. Biosens. Bioelectron. 26, 62–69; 10.1016/j.bios.2010.05.010 (2010).
Mathews, A. S., Kim, I. & Ha, C.-S. Synthesis, characterization, and properties of fully aliphatic polyimides and their derivatives for microelectronics and optoelectronics applications. Macromol. Res. 15, 114–128; 10.1007/BF03218762 (2007).
Massey, L. K. Permeability properties of plastics and elastomers. A guide to packaging and barrier materials. 2nd ed. (Plastics Design Library/William Andrew Pub, Norwich, NY, USA, 2003).
Vomero, M. et al. Conformable polyimide-based μECoGs: bringing the electrodes closer to the signal source. Biomaterials 255, 120178; 10.1016/j.biomaterials.2020.120178 (2020).
Petrini, F. M. et al. Sensory feedback restoration in leg amputees improves walking speed, metabolic cost and phantom pain. Nat. Med. 25, 1356–1363; 10.1038/s41591-019-0567-3 (2019).
Rubehn, B. & Stieglitz, T. In vitro evaluation of the long-term stability of polyimide as a material for neural implants. Biomaterials 31, 3449–3458; 10.1016/j.biomaterials.2010.01.053 (2010).
Wurth, S. et al. Long-term usability and bio-integration of polyimide-based intra-neural stimulating electrodes. Biomaterials 122, 114–129; 10.1016/j.biomaterials.2017.01.014 (2017).
Badi, M. et al. Intrafascicular peripheral nerve stimulation produces fine functional hand movements in primates. Sci. Transl. Med. 13, eabg6463; 10.1126/scitranslmed.abg6463 (2021).
Harland, B. et al. A subdural bioelectronic implant to record electrical activity from the spinal cord in freely moving rats. Adv. Sci., e2105913; 10.1002/advs.202105913 (2022).
Guan, S. et al. Elastocapillary self-assembled neurotassels for stable neural activity recordings. Sci. adv. 5, eaav2842; 10.1126/sciadv.aav2842 (2019).
Vomero, M. et al. On the longevity of flexible neural interfaces: Establishing biostability of polyimide-based intracortical implants. Biomaterials 281, 121372; 10.1016/j.biomaterials.2022.121372 (2022).
Ohta, J., Tokuda, T., Sasagawa, K. & Noda, T. Implantable CMOS biomedical devices. Sensors 9, 9073–9093; 10.3390/s91109073 (2009).
Sporer, M., Gueli, C., Stieglitz, T. & Ortmanns, M. NeuroBus - Archtecture and communication bus for an ultra-flexible neural interface. submitted. IEEE International Symposium on Circuits and Systems (ISCAS) (2023).
Kleemann, H., Krechan, K., Fischer, A. & Leo, K. A review of vertical organic transistors. Adv. Funct. Mater. 30, 1907113; 10.1002/adfm.201907113 (2020).
Lüssem, B. et al. Doped organic transistors. Chem. Rev. 116, 13714–13751; 10.1021/acs.chemrev.6b00329 (2016).
Ji, D., Li, T., Hu, W. & Fuchs, H. Recent progress in aromatic polyimide dielectrics for organic electronic devices and circuits. Adv. Mater. 31, e1806070; 10.1002/adma.201806070 (2019).
Paxinos, G., Calabrese, E., Johnson, G. A., Watson, C. & Badea, A. MRI/DTI atlas of the rat brain (Elsevier/AP, London, 2015).
Mace, G. M., Harvey, P. H. & Clutton‐Brock, T. H. Brain size and ecology in small mammals. J. Zool. 193, 333–354; 10.1111/j.1469-7998.1981.tb03449.x (1981).
Prodanov, D. & Delbeke, J. Mechanical and biological interactions of implants with the brain and their impact on implant design. Front. Neurosci. 10, 11; 10.3389/fnins.2016.00011 (2016).
Porto Cruz, M. F. Conformable epicortical implants for brain mapping: A technological perspective on neural recordings (Verlag Dr. Hut, München, 2021).
Roman, B. & Bico, J. Elasto-capillarity: deforming an elastic structure with a liquid droplet. J. Phys. Condens. Matter 22, 493101; 10.1088/0953-8984/22/49/493101 (2010).
Buzsáki, G., Anastassiou, C. A. & Koch, C. The origin of extracellular fields and currents--EEG, ECoG, LFP and spikes. Nature reviews. Neuroscience 13, 407–420; 10.1038/nrn3241 (2012).
Teichert, C. et al. Comparison of surface roughness of polished silicon wafers measured by light scattering topography, soft‐x‐ray scattering, and atomic‐force microscopy. Appl. Phys. Lett. 66, 2346–2348; 10.1063/1.113978 (1995).
Burghartz, J. N. et al. Hybrid systems-in-foil—combining the merits of thin chips and of large-area electronics. IEEE J. Electron Devices Soc. 7, 776–783; 10.1109/JEDS.2019.2896188 (2019).
Christiaens, W., Bosman, E. & Vanfleteren, J. UTCP: A novel polyimide-based ultra-thin chip packaging technology. IEEE Trans. Comp. Packag. Technol. 33, 754–760; 10.1109/TCAPT.2010.2060198 (2010).
Gueli, C., Martens, J., Eickenscheidt, M. & Stieglitz, T. Scalable batch transfer of individual silicon dice for ultra-flexible polyimide-based bioelectronic devices. 43rd Annual International Conference of the IEEE Engineering in Medicine & Biology Society, 6880–6883; 10.1109/EMBC46164.2021.9630832 (2021).
Hopcroft, M. A., Nix, W. D. & Kenny, T. W. What is the Young's Modulus of Silicon? J. Microelectromechanical Syst. 19, 229–238; 10.1109/JMEMS.2009.2039697 (2010).

Table 1

Mechanical properties of polyimide, silicon and rat brain used for modelling (Young's modulus E, Poisson's ratio v density ⍴)

Material	E in Pa	ν	⍴ in kg/m3
Polyimide	9∙10⁹	0.34	1100
Silicon	170∙10⁹	0.28	2329
Brain Cortex	1.75∙10³	0.40	1027

(Not answered)

Download PDF

Journal Publication

published 09 May, 2023

Read the published version in Microsystems & Nanoengineering →

Editorial decision: revise
05 Feb, 2023
Review #3 received at journal
03 Feb, 2023
Review #2 received at journal
22 Jan, 2023
Reviewer #3 agreed at journal
20 Jan, 2023
Review #1 received at journal
07 Jan, 2023
Reviewer #2 agreed at journal
31 Dec, 2022
Reviewer #1 agreed at journal
28 Dec, 2022
Reviewers invited by journal
28 Dec, 2022
Submission checks completed at journal
13 Dec, 2022
Editor assigned by journal
13 Dec, 2022
First submitted to journal
13 Dec, 2022

You are reading this latest preprint version

Highly Conformable Chip-in-Foil Implants for Neural Applications

Status:

Journal Publication

Version 1

Abstract

Figures

1 Introduction

2 Results

2.1 Design Assessment: Implant Geometry

2.2 Transfer Technology

2.3 Accuracy Assessment: Die Placement

3 Discussion

4 Conclusion

5 Methods

5.1 Design Assessment: Implant Geometry.

5.2 Transfer Technology.

5.3 Accuracy Assessment: Die Placement.

Declarations

References

Tables

Additional Declarations

Status:

Journal Publication

Version 1