Making the carbon nanotube brain electrode

Carbon Nanotube Fibers

The materials used in artificial joints, like knees or hips, not only determine what you will be able to do with the joint, but ultimately for how long. If you’re lucky, you will be a central part in deciding what goes inside you. The same considerations apply when it comes time to visit your local brain interface outfitter. As in the joint business, it now appears that we can do much better than the metal.

Researchers at Rice University have perfected a technique they call ‘wet spinning’ to bundle millions of nanometer-sized carbon nanotubes into micron-sized threads. Depending on the precise mix used, they can tailor these threads for optimal strength, stiffness and conductivity. Group leader Matteo Pasquale had previously created nanotubes that have a higher ratio of both strength and conductivity, to weight, then copper. He made a name for himself in coming up with ways to measure and demonstrate the raw physical prowess of these composite nanotubes. A scaled-up carbon nanotube factory, if it existed, could provide cable that could replace the need for steel-reinforced copper in transmission lines. On smaller scales however, there are already applications that the current supply chain can meet.

The artificial joint business now recognizes that metal grinding on metal doesn’t last very long. It also generates a steady stream undesirable element that deposits themselves, courtesy of the bloodstream, in unfortunate places. Elements like cobalt used to fortify stainless steel, or the vanadium and other goodies used to alloy titanium. Supple plastics like PEEK, sliding on ultra-smooth ceramics have much more to offer. The reason we have gone on here about metals at interfaces in joints, is to set in our heads the fact that herein lies much the same problem we have with implanted electrodes.

In other words, despite their great properties of strength and intrinsic conductivity, they tend to fail at the boundaries. Traditional precious metals used in implants, like platinum and silver, hand off electrons with no problem when the neighbor is another metal. However the impedance shift across the gap to brain is not so smooth — over time there is generally both loss of material in some spots and buildup in others that eventually compromises the whole works. Impedance is a better way to characterize electrode performance than resistance for several reasons. Whereas resistance is only a measure of opposition to a DC current flow, we might imagine impedance as resistance to everything — resistance under different frequencies or waveforms of AC, to pulses, and across sundry materials, conditions, and charge carriers.

In making bidirectional electrodes (those capable of stimulating and recording) it can be difficult to do both well. We should point out that with metals, like the medieval-looking pincushion array to the right, a duty cycle where current is delivered by the electrode alternately in both directions might have some cleaning effect for a while. However when it is time to record, there will eventually be some significant performance loss. While there is no single best impedance, to isolate just a single nearby neuron from the background chatter, you want fairly high electrode impedance — especially if you have a dense array of other neighbor electrodes that can potentially sample the same field. In order to deliver current only where you intend to, the bulk of the electrode body is covered with insulating material. For metal electrodes, glass is often used for the coating, while the nanotubes here used a 3-micron layer of flexible biocompatible polymer.

On the other hand, to deliver current for stimulating neurons you want a low impedance for your electrode sources or sinks. The researchers found that their nanotubes could stimulate neurons using a much lower voltage than traditional electrodes. Nanotubes aren’t just a one-trick pony, though. Even more important to a viable implant technology is the ability to mechanically match the brain. Again the joint provides the perfect analogy: If the stem on your ceramic-coated titanium implant does not bend, i.e. has similar modulus of elasticity and/or geometry as the surrounding bone, you will quickly degrade the bond and the implant will fail.

For stiff electrodes, stiff enough to press down through the cortex anyway, it won’t necessarily be the implant that eventually fails, but rather your soft brain. Consider an average trip to the cross-fit gym. As you clean and jerk even modest weight about, your brain moves too. It’s not just the quickened beat of a pounding heart, but also any number of accelerations, impacts, and gravitational anomalies that you willfully subject your delicately suspended neural tissue to. The ensuing dehydration itself produces significant shrinkage effects, as even does the natural sleep-wake cycle — up to 10% by volume according to some measurements. All this can and would wreak havoc using a stiff implant, even if it is not rigidly bound to any internal or external fixation points.

The researchers demonstrated proof of principle using rats that had the rodent equivalent of Parkinson’s disease. While not qualified to discuss the more subtle points of what that means for the rat, we can report that the softer gentler electrodes stimulated neurons as effectively as metal electrodes that required 10 times the amount of current-delivering area. All without any detectable inflammatory response.

As with joints, electrode arrays cannot be a one-size-fits-all affair. Perhaps the first sign we will know these products will be ready for prime time is when researchers figure out not only where they want to put them, but also what stimulation protocols to use. Things like that are critical to determining the ideal electrode geometry and also how much ‘effective stimulation lifetime’ needs to built into it. As for location, we would suggest that a gray matter array cannot possibly be the same one you want to use for the white matter. Within gray itself you want different properties for unmyelinated versus the myelinated axons, and in the white — as in the cortex for example — you want a different style for the deep layer neuron bodies than you would want for the surface layer neurites.

There are also other promising materials on the horizon to use in these kinds of arrays. Graphene is one that already has shown potential not just in the electrical and optical interface environment, but also in the larger clinical setting. While many of the things raised here are issues for tomorrow, when we eventually start to see them being commonly discussed, we might know it is almost time to shop.