Viscoelastic surface electrode arrays to interface with viscoelastic tissues

Designing implantable electrode arrays that mimic their target biological tissue(s)
Viscoelastic surface electrode arrays to interface with viscoelastic tissues

Medical devices diagnose, treat, and/or provide therapies, all to benefit the quality of life of patients. Described as early as the ancient Greeks, medical devices have undergone significant technological advancement and regulation over the past ~3000 years, but one subtype that has gained a lot of attention in the last few decades is implantable neural devices. Composed of an array of micro-electrodes (MEAs) that aim to restore the lost or damaged functions of the nervous system, neural implants communicate with neural tissue by applying electrical stimulation to, or recording from, various neurons. These devices have many functions in the clinic (for example: auditory sensation restoration in deaf patients, relief from chronic pain). Despite their broad uses, existing MEAs have significant limitations; they suffer from minimal electrode recording stability over time, and cause damage to the nervous system. The lack of long-term reliable interfaces is especially detrimental for complex applications such as brain-computer interfaces that allow tetraplegic patients to interact with their environment by recording brain signals from the motor cortex of the brain to guess their intentions.

One major limitation of clinical neural implants is the materials used to fabricate the MEAs. Currently, a rigid plastic film surrounds the thick and rigid metal disks used for the electrodes and electrical tracks. These materials present mechanical mismatch compared to the soft and curved tissues with which they interface, thereby constraining the physiological motion dynamics of the nervous system. Ultimately, this mismatch causes minimal contact between the electrodes of the implant to the underlying tissue, which results in poor recording capabilities as well as chronic scarring and decreased benefits to patients. Recent efforts have aimed to reduce mismatch by developing MEAs from materials with an elastic modulus more similar to that of the brain, using silicone elastomers or hydrogel coatings instead of thin films of plastics (Fig 1).

Fig 1: Comparison of the mechanical moduli of various classes of materials. Metals, used for the thin electrical tracks of existing MEAs are more than 109 times more stiff than neural tissues. Plastic films are 106 times stiffer, and elastomers, such as silicones which have been more widely used in the last 5-10 years still have 103 increased stiffness. By contrast, hydrogels offer comparable mechanical modulus to that of neural tissues.

Even though the reduction of the elastic modulus has offered better biointegration with the brain, we hypothesized other mechanical properties of brain tissues could play an equal or more important role.  Most notably, we realized that while all living tissues are viscoelastic in nature, no MEA are viscoelastic.

Viscoelasticity is a material property that means the materials continue to flow when a force is applied. Viscoelastic hydrogels have been used by our group, the Mooney lab at Harvard, for various bio-inspired applications. Alginate, a seaweed extract (Fig 2) has been widely explored in the group due to its highly tunable and viscoelastic nature: by altering the molecular weight of the alginate, and/or the amount of crosslinker which is used to turn the viscous alginate liquid into a gel, both the degree of viscoelasticity and mechanical modulus can be carefully controlled to recapitulate the native cell environment of various tissues.  

Fig 2: Alginate is a seaweed extract, with a mechanical profile that can be highly tuned for applications to single cell encapsulation and the development of bioinspired artificial matrices. Photo credit: SNP Inc.

We realized that a viscoelastic array might allow for a more conformability, i.e. access regions of the brain that cannot currently be targeted without causing damage or compression to the underlying tissue. A hypothetical 2D film, either from a rigid or elastic material (green) or a viscoelastic alginate gel (blue), can be placed on the same cortical region. While the green elastic film is unable to flow over time and remains static, the viscoelastic material can flow and precisely conform to the exact cortical geometry (Fig 3).

Fig 3: A schematic showing a hypothetical elastic (green) surface array, and a hypothetical viscoelastic (blue) surface array. When both implants are placed at the same location of the brain (top row), they have a 2D architecture. Over time, the elastic array is unable to flow to match the underlying geometry. The viscoelastic array, however, is able to continuously remodel to precisely match the underlying structure of the brain.

If we think about a fully assembled arrays instead of 2D sheets of materials, there are 2 main components: (1) the encapsulation layer that is handled by the surgeon, which embeds the (2) electrical tracks that record signals from the neurons. As none of the materials in the 'implantable surface array toolbox' were viscoelastic, we introduced new alternatives to fabricate the first fully viscoelastic array (Fig 4).  

Fig 4: A schematic of the fully viscoelastic array, showing all the different layers. In turquoise is a soft layer of alginate with the same mechanical modulus and viscoelasticity as the mammalian brain. In magenta is a viscoelastic insulation layer, based on a previously described self-healing PDMS layer. In black are the described viscoelastic conductive tracks, that are based on a microporous alginate network which are embedded with carbon conductive fillers.

Our first step in fabricating fully viscoelastic arrays was to mechanically characterize the mammalian brain, which highlighted its soft nature (~1 kPa), as well as its viscoelasticity (tan(delta)~0.4). We then found the right concentration of crosslinker for our alginate gels to match the brain mechanics. When we compared our viscoelastic substrate with plastic and elastic substrates to see if a viscoelastic material would indeed be more conformable, we found that the viscoelastic alginate had more than a 2x coverage of the underlying complex surface of a porcine brain model made of gelatin. Additionally-and importantly!-the alginate substrates were much thicker (a few hundred µm) compared to the plastic films (<10µm), which meant they were easier to handle, while conforming better to the complex curvature of the brain. We also found that the viscoelastic nature of the alginate allowed it to flow and intimately match the curvature of the brain, as well as enter the sulci. When removed from the underlying tissue, the alginate retained its shape for a few hours, before flowing to its initial 2D geometry (Fig 5). This enables rapid ‘personalized medicine’, as the films can flow precisely to their underlying surfaces, without damaging any of the fragile neurological structures.

Fig 5: Alginate (left, yellow material) flows to conform to the underlying geometry of the tissue. In contrast, an elastomer (right, blue material) returns to its initial flat shape. Scale bar is 5 mm.

Next, we wanted to replace the existing rigid metal electrodes with a viscoelastic alternative which was compatible with the processing of highly hydrated materials. Although alginate is ionically conductive, we wanted to enhance the electrical behavior of the gel by mechanically suspending conductive particles. The aspect ratio and composition of the additive are both important: small aspect ratio materials need a higher loading composition and are more likely to counter the brain-like mechanical profile (viscoelasticity, mechanical modulus) of the alginate matrix. However, carbon nanomaterials, such as graphene flakes (GF) and carbon nanotubes (CNT), are highly conductive, metal-free, and have a high aspect ratio. By incorporating them in the alginate, we were able to create viscoelastic conductors with tunable electrical properties.

As we were simply trapping the carbon nanomaterials in the alginate matrix, we had little control of the distribution of the materials and thus had variation between batches. When thinking about percolation-a path from one end of the gel to the other that is continuously conductive-we realized that we could engineer percolation by creating pores in the matrix. Indeed, the analytical solution predicts a continuous matrix vs a porous matrix, with particles of aspect ratio=1, needs ~16% to <4% of filler to ensure percolation. As the alginate is 98%+ water, by freezing the gels and then lyophilizing (drying), we are able to form ice crystals that localize the conductive additives to the gel wall. Then, we can remove the ice in the drying step and crosslink the composite gels to preserve the highly porous networks with carbon nanomaterials trapped in the wall (Fig 6).

Fig 6: A schematic of the conductive alginate gels (light blue) with conductive additives (black circles), comparing the nanoporous and microporous formulations for a gel of the same size with the same number of additives. On the far left is the nanoporous composites, with no freezing steps. The additives are mechanically distributed, randomly, throughout the gel matrix and the conductive path (outlined in red) is limited. In the middle image, the gels are frozen to form ice crystals (gray stars), which displace the additives to the space surrounding the ice crystals. On the far right, the ice is dried by lyophilization. The conductive path (red line) is able to span the entire length of the gel and form a percolating path.

At this point, we were just missing an insulation component of the device. A lot of effort was put to create a biocompatible approach to couple elastomers with hydrogels, but also with themselves without the use of vacuum processes (like a O2 plasma cleaner).  In December 2019, we heard Professor Zhenan Bao give a talk (at Fall MRS 2019) about a self-healing PDMS that could attach to itself even in aqueous environments. We synthesized the material and then entangled an amine-terminated PDMS monomer into the solution so that we could covalently attach alginate. Our modified insulating material, which we call PEVM, can self-heal around the viscoelastic tracks with limited current leakage.

It didn’t take long to combine all the introduced materials and fabricate a full multielectrode array. Some 'stats': from start to finish, it takes ~3 days and no clean room processes to make the devices. We can easily produce ~7-10 devices/batch with reliable behavior between both conductive hydrogel batches and device performance. Our first assembled devices were ready in March 2020-right when the pandemic broke out-but when we returned to the lab in June 2020, we were able to quickly restart the fabrication pipeline and by the end of the month had our first validation recording EKG from the epicardium of the mouse heart. Our collaborators in Switzerland at EPFL further validated the technology by recording from the motor cortex and then demonstrated some exciting recordings from the auditory cortex of a rat brain. Since then, we have been working on scaling the technology to larger number of electrodes while minimizing overall thickness, and continued demonstration of device conformability.

Fig 7: The fully assembled described viscoelastic array, hydrated and with 15 electrodes and a total thickness of ~250 μm, designed for the brain cortex (left). A scaled up version of the array with 15 electrodes, placed on a mock porcine brain shows how conformable the technology is, and its ability to follow the underlying complex geometry of the tissue (right).

We are excited to explore different applications of these viscoelastic arrays, especially to larger mammals with more complex brain geometries. Our platform is highly tunable, and the arrangement of electrodes can be adjusted for a particular application (physiological location and/or question). Ultimately, we have provided a tool that can be used by researchers broadly in the biomedicine and biomedical engineering space to broadly interface with tissues. If you have any suggestions for the next-generation devices, or a particular application interest in mind, we would love to hear from you!

To read more about our work, click here:


Please sign in or register for FREE

If you are a registered user on Nature Portfolio Engineering Community, please sign in