7 Years Later: Revolutionizing Neural Research with SiC

NeuroNexus is committed to continuous improvement to uphold the highest quality standards and ensure our customers remain at the forefront of scientific advancement. In 2017, driven by innovation, we introduced amorphous silicon carbide (a-SiC) into specific probe designs that demanded longer implant times, leveraging its superior dielectric properties. For this reason, a-SiC is well-suited for the time spans typically required in chronic non-human primate studies.

By 2022, we had developed an in-house process to deposit a-SiC, subsequently replacing SiO2 across all silicon probes. This development was a result of our Phase I NIH SBIR initiative, titled “Advanced three-dimensional microscale electrode array for high-density volumetric neural interfacing.” Beyond serving as an exceptional insulator and boasting excellent biocompatibility—essential for chronic implants—a-SiC surpasses SiO2 with its higher breakdown voltage, important for stimulation, and higher fracture resistance.

Incorporating a-SiC into our probe technology marks a significant advancement in our pursuit of innovative neural interface solutions. This transition underscores our commitment to pushing boundaries in neural research, ensuring reliability and performance in long-term scientific investigations.

A brief history of electrode technology

Our history of microelectrode technology starts in the 1950s.  

It wasn’t until 1953 that the first glass pipet microelectrode was used for extracellular recording (Rose, Science 1953). The glass pipette had a platinum wire inside, and it had a 10-micron tip. From there, there was a big development where Hubel developed the first microwire-based electrode (Hubel, Science 1957). This was a significant improvement because of the mechanical durability of wires compared to glass. He was able to record from the cat dorsal root for periods up to an hour. Later in that decade, Strumwasser reported using the first microwire bundle, an array consisting of four wires (Strumwasser, Science 1958). He was able to record for up to seven days.  

Those involved in operant conditioning studies may be familiar with the Olds work from 1965 which used nine nichrome wires for recordings lasting up to seven days and with 50% yield of action potentials (Olds, Brain Research 1965). In 1967, Naka reported the use of tungsten microwires (Naka, Brain Research 1967). These were 127-micron diameter with 1-micron tips, and he was able to record up to one month with them. 

In 1969 there began a revolution.  

Wise reimagined what a neural electrode would look like, using silicon as a substrate and using MEMS-based technologies (Wise, Conf on Engineering in Med and Biol 1969). His was the first report of an electrode developed with silicon substrate and silicon dioxide dielectrics. Then, in 1971, the National Institutes of Health (NIH) formed the Neural Prosthesis Program to push the field forward by bringing researchers together to collaborate and to disseminate new developments within the area of neural engineering and neural interfaces. In 1973, the Bak electrode was reported, contributing the first flexible interconnect for microelectrode packaging (Salcman & Bak, IEEE Trans BME 1973). Schmidt then reported recordings with iridium wires coated with parylene C. He was able to record 223 days (about 7 and a half months) in monkey cortex (Schmidt, Experimental Biology 1976). Finally, in 1988 was the first report of the Michigan probe, a silicon probe in vivo (Drake et al., IEEE Trans BME 1988), which demonstrated the value of a laminar array: simultaneous recordings at multiple depths. 

In 1994, the Center for Neural Communication Technologies (CNCT) was launched. This was an NIH-funded center that was in place for 10 years. It really amplified the efforts that were put into developing and refining the silicon probe technology, with a focus on acute probes. Acute probes are used temporarily and then taken out of the subject. The goal of CNCT was to disseminate silicon probes throughout the world for researchers that could use them. 

Williams tested chronically implanted silicon probes in 1999 (Williams & Kipke, Brain Res Prot 1999). This work brought into context the electrode-electrolyte interface, as well as the value of impedance spectroscopy, which was used to understand that the failure mechanisms of silicon probes are both device or abiotic failure and biological foreign body response or biotic failure. Then, in 2003 and 2004 Kipke and Vetter reported the first chronic performance assessment of the Michigan probe, and were able to get reliable recordings for fifty-five weeks (over 1 year) in rats (Kipke & Vetter, IEEE Neural Sys & Rehab Eng 2003; Vetter & Kipke, IEEE Trans BME 2004). 

In 2004, NeuroNexus was officially launched. 


  • 1953 = first glass pipet microelectrode for extracellular recording  
  • 1957 = first microwire-based electrode, one hour recording 
  • 1958 = first microwire bundle (4 wires), 7 days recording 
  • 1965 = nichrome wire bundle (9 wires), 7 days recording 
  • 1967 = tungsten microwires, 1 month recording 
  • 1969 = silicon, MEMS-based technology 
  • 1971 = NIH Neural Prosthesis Pogram 
  • 1973 = flexible interconnect packaging 
  • 1976 = parylene C-coated iridium wires, 7+ months recording 
  • 1988 = silicon probe 
  • 1994 = CNCT launched with acute silicon probes 
  • 1999 = chronic silicon probes 
  • 2003-4 = chronic silicon probes, 12+ months recording 
  • 2004 = NeuroNexus 

Silicon or Wires

The day the lab acquired simultaneous single-unit recordings for the first time, it felt like a jump to hyper speed. That graduation from a single tungsten electrode to multi-channel arrays meant a whole new kind of data was available about the region of interest. Once a lab commits to multi-channel recordings as a specific aim, one of the first decisions to make is whether to record with a wire bundle like the early handmade tetrodes, or if silicon probes are the way to go.

Silicon or Wires


What are the main differences between silicon probes and wire electrodes for single-unit recordings?

1. Durability. Wire bundles might be malleable, but insulation can wear away over time. Silicon probes are rigid (and therefore fragile), but with proper cleaning are reusable indefinitely.

2. Recording quality optimization. Most users seek high impedance on wires, but small features on silicon probes set the goal for ever-lower impedance.

3. Channel count. Wire bundles get fatter with every channel added. In contrast, silicon probes maintain a comparatively small profile with orders of magnitude higher channel counts.

To back this list up, here are some additional thoughts.


First, consider material properties of the multi-electrode array. Being able to bend the array out of and back into shape can be seen as an advantage, a “safety net” of a wire microelectrodes. If the objective of an experiment is to get multiple channels into the region of interest at once, this works great. The advantage of a rigid device like a silicon probe is that the layout of the electrode sites never changes. Data analysis that takes distance between channels into account can be hard coded and trusted for as long as one uses the same probe layout across any number of devices.

Recording Quality Optimization

When preparing an electrode for extracellular recording, there are details to double check before that electrode enters tissue. On a single tungsten electrode, most labs look for the highest impedance possible. This is comparable to the smallest wire diameter possible for wire bundles. The high impedance is equated with a sharp electrode, which in turn indicates a small electrode site. That is how the recordings will provide selectivity required to isolate single units. When an electrode site gets larger due to insulation wearing away over time, one is more likely to capture multi-unit activity or “hash”. On silicon probes, low impedance has advantages. This is because the electrode site size is pre-fabricated onto the silicon substrate. The electrode site does not, and can not, get larger over time. So, silicon probe manufacturers are constantly innovating to lower impedance, which is equated with thermal noise in each channel. With lower impedance, a recording’s background noise amplitude gets smaller.

Channel Count

The number of channels per wire bundle was addressed above as being limited, in a way, by the diameter of device researchers are willing to insert into their region of interest. Silicon probe technology has evolved to allow for ever-higher numbers of electrode sites on ever-smaller silicon substrates. A more detailed description of the difference between wire bundles and silicon probes, as it relates to channel count, is that silicon probes are planar devices. They can be made with one shank (i.e., “tooth” of a comb) or with 2, 4, 8, 16+ shanks each with multiple electrode sites on them. It is easy to group electrode sites at a single depth in tissue, or to spread electrode sites across depths in a laminar array. With higher channel counts, both can be done with the same device in a single insertion.


A laboratory group must consider scientific, mechanical, budgetary and other constraints when planning an experiment. Dr. Gyorgy Buzsaki (NYU) once said, “Since [silicon probe pioneer] NeuroNexus began fabricating probes with high reliability and reasonable costs, we virtually stopped using wire electrodes and monitor electrical activity with silicon probes. It is a one-way process: once one begins to record with silicon probes, he/she never goes back to wires.”

Recording Quality Optimization

From the vault…

At the start of 2019, our first newsletter featured three new publications on the auditory system - a nod to the upcoming Association for Research in Otolaryngology meeting that took place in person in Baltimore that month.

The benefit of NeuroNexus probes for this work is the consistent layout of electrode sites on silicon probes. The researchers all used linear arrays, for example spaced every 50um for recordings in mouse, and spaced every 100um for recordings in guinea pig. Both of these options are standard in the NeuroNexus catalog.

In two of the studies, the linear arrays were inserted along the tonotopic axis in the inferior colliculus, enabling recording of single neurons from multiple isofrequency bands. What this means is that NeuroNexus probes enabled the researchers to sample neuronal activity throughout a particular region of the brain that is sensitive to sound. Different neurons in that region are activated by different frequencies; each individual neuron responds to a narrow frequency band and is characterized by just one “best frequency”. The neurons in the inferior colliculus are arranged in order by their best frequency, and a linear microelectrode array can be inserted along that arrangement to sample from neurons with as many different best frequencies as possible.

With this setup, one study compared neuronal activity in response to frequencies above and below a particular frequency that was used to induce hearing loss. The other study determined that the neuronal arrangement in the inferior colliculus was altered if a specific step in auditory system development was disrupted.

The third publication we featured was a great reminder to our staff in the office and our Neuroscientist colleagues alike that NeuroNexus probes can be used in a very wide range of research models. The researchers used several probe designs to record from auditory cortices in the zebra finch and studied the processing stages of song.

Bonus Tip: Probe Insertion

Our January 2019 product feature addressed the challenge of penetrating the dense surface of the brain. While we recommend performing a durotomy prior to probe insertion, the pia exhibits some resistance and so causes the underlying tissue to dimple beneath a probe. In our experience, if you pause insertion and wait a few minutes, the probe will work itself into the tissue and you can be on your way. We also get asked to sharpen probe tips for certain applications.

The NeuralGlider by Actuated Medical is a device that produces tiny vibrations along the axis of the probe to facilitate insertion without dimpling. This is especially advantageous for recordings in superficial regions where dimpling could put pressure on the tissue of interest, or could cause the probe to insert past the target and have to be retracted. We are happy to be able to rent out a NeuralGlider to interested labs.