Three dimensional microelectrodes enable high signal and spatial resolution for neural seizure recordings in brain slices and freely behaving animals

Abstract

Neural recordings made to date through various approaches—both in-vitro or in-vivo—lack high spatial resolution and a high signal-to-noise ratio (SNR) required for detailed understanding of brain function, synaptic plasticity, and dysfunction. These shortcomings in turn deter the ability to further design diagnostic, therapeutic strategies and the fabrication of neuro-modulatory devices with various feedback loop systems. We report here on the simulation and fabrication of fully configurable neural micro-electrodes that can be used for both in vitro and in vivo applications, with three-dimensional semi-insulated structures patterned onto custom, fine-pitch, high density arrays. These microelectrodes were interfaced with isolated brain slices as well as implanted in brains of freely behaving rats to demonstrate their ability to maintain a high SNR. Moreover, the electrodes enabled the detection of epileptiform events and high frequency oscillations in an epilepsy model thus offering a diagnostic potential for neurological disorders such as epilepsy. These microelectrodes provide unique opportunities to study brain activity under normal and various pathological conditions, both in-vivo and in in-vitro, thus furthering the ability to develop drug screening and neuromodulation systems that could accurately record and map the activity of large neural networks over an extended time period.

Full text

1
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports
Three dimensional microelectrodes
enable high signal and spatial
resolution for neural seizure
recordings in brain slices and freely
behaving animals
P. Wijdenes1,2, K. Haider1, C. Gavrilovici3, B. Gunning4, M. D. Wol4, T. Lijnse5, R. Armstrong1,
G. C. Teskey1,3, J. M. Rho3,6, C. Dalton2,5 & Naweed I. Syed1,3,4,7*
Neural recordings made to date through various approaches—both in-vitro or in-vivo—lack high
spatial resolution and a high signal-to-noise ratio (SNR) required for detailed understanding of brain
function, synaptic plasticity, and dysfunction. These shortcomings in turn deter the ability to further
design diagnostic, therapeutic strategies and the fabrication of neuro-modulatory devices with
various feedback loop systems. We report here on the simulation and fabrication of fully congurable
neural micro-electrodes that can be used for both in vitro and in vivo applications, with three-
dimensional semi-insulated structures patterned onto custom, ne-pitch, high density arrays. These
microelectrodes were interfaced with isolated brain slices as well as implanted in brains of freely
behaving rats to demonstrate their ability to maintain a high SNR. Moreover, the electrodes enabled
the detection of epileptiform events and high frequency oscillations in an epilepsy model thus oering
a diagnostic potential for neurological disorders such as epilepsy. These microelectrodes provide
unique opportunities to study brain activity under normal and various pathological conditions, both
in-vivo and in in-vitro, thus furthering the ability to develop drug screening and neuromodulation
systems that could accurately record and map the activity of large neural networks over an extended
time period.
Our understanding of brain functions under normal and pathological conditions remains limited, due in large
part, to barriers in monitoring subtle electrical signals from large networks of interconnected neurons. As such,
the underlying neuronal dynamics of many neurological disorders remain unknown, precluding our ability to
treat perturbed brain function using neuro-stimulation approaches1–3. is lack of fundamental knowledge
stems from our inability to monitor neural activities from complex, synaptically connected networks at a high
spatial–temporal resolution, and over extended time periods. In an attempt to ll these gaps, microelectrodes
embedded in Multi-Electrode Arrays (MEAs) are now routinely interfaced with a variety of homogeneous cell
culture and brain slice preparations maintained in-vitro4–7. However, as the neural networks established in
culture are articial, the connectivity patterns studied in any given experiment may vary from preparation to
preparation and are likely not representative of invivo network phenomena. While MEAs are useful in their
utility to monitor neural activity simultaneously at multiple sites, this approach is limited in its spatio-temporal
resolution8,9, even more so when tentatively interfaced with brain slices. e signal-to-noise ratio (SNR) oered
by traditional MEAs is oen low10 and precludes longer-term recording of spontaneously active networks at a
resolution, high enough to decipher excitatory and inhibitory synaptic potentials related to neural activity and
OPEN
1Faculty of Medicine, Hotchkiss Brain Institute, University of Calgary, 2500 University Dr. NW, Calgary, AB T2N
1N4, Canada. 2Biomedical Engineering Graduate Program, University of Calgary, 2500 University Dr. NW, Calgary,
AB T2N 1N4, Canada. 3Alberta Children’s Hospital Research Institute, University of Calgary, 2500 University
Dr. NW, Calgary, AB T2N 1N4, Canada. 4Department of Cell Biology and Anatomy, University of Calgary, 2500
University Dr. NW, Calgary, AB T2N 1N4, Canada. 5Department of Electrical and Computer Engineering, University
of Calgary, 2500 University Dr. NW, Calgary, AB T2N 1N4, Canada. 6Departments of Neurosciences and Pediatrics,
University of California San Diego, Rady Children’s Hospital, San Diego, CA, USA. 7Cumming School of Medicine,
University of Calgary, 3330-Hospital Drive, NW, Calgary, AB T2N 4N1, Canada. *email: nisyed@ucalgary.ca
Content courtesy of Springer Nature, terms of use apply. Rights reserved

2
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
connectivity. ese limitations are due mainly to three factors: (1) traditional planar electrodes used in MEAs
only record neural activity from the outer and “traumatized” layers of brain slices, which harbour a larger
population of either dead or dying cells4; (2) the surface of the tissue damaged during slicing oen releases
proteases and ions, such as potassium and sodium, which in turn result in excitotoxicity of the adjacent area,
thus causing further tissue damage at the recording site resulting in electrical artefacts; (3) the perfusion system
required to provide a continuous ow of nutrients and oxygen at a set temperature to maintain the brain slice
creates a ow of charged ions within the recording chamber and around the microelectrodes, generating electri-
cal noise11,12 (Fig.1a). Even in those instances where recordings are possible, the above stated challenges make
data collection and interpretation inconsistent and unreliable13. In contrast, while several types of protruding
three-dimensional microelectrodes have been developed to oer alternatives to planar electrodes which record
higher delity activity in brain slices (e.g. spine-shaped protrusion electrodes14, carbon nanotube electrodes15
or metal-transfer-micromolded electrodes16,17) the ecacy of the recorded potentials is oen suboptimal due
to tissue damage and low SNR9.
Most of the above-mentioned challenges are carried forward in the eld of implantable neural electrodes
designed for the recording and stimulation purposes. In these situations, the lack of signal delity precludes the
development of “smart” implants with fully responsive and adaptable feedback loop system. Moreover, current
neural interfaces do not provide high enough SNR, granularity or coverage required to generate sucient and
reliable data deemed essential for articial intelligence and machine learning applications in clinical practices.
e existing devices although tend to provide a decent coverage of the cortical brain structures, albeit with low
SNR and spatial resolution (e.g. traditional electrocorticograms (ECoGs) used in the evaluation for epilepsy
surgery18 or higher density alternative MicroECoGs, which suer from low SNR), or a nominal coverage of
the brain with higher SNR and spatial resolution (e.g. Utah Array19, Paradromic20, Neuralink21. Additionally,
the implanted electrodes oen result in scar formation around the implant and as such not only aect the SNR
but also perturb neuronal functions. Ideally, all of the above problems would need to be concurrently solved to
maximize our ability to map the brain, develop eective machine learning algorithms that can improve smart
feedback loop systems for neuromodulation, and ultimately improve patient outcomes.
Based on an electrical and structural computational simulation paradigm, we describe here a novel fabrication
method that enabled the development of a reusable invitro multisite array of three-dimensional microelectrodes
permitting high-resolution, long-term ECoG-like cortical recording with minimal tissue invasion and damage.
Unlike most previously published work on microelectrodes, the fabrication process presented here is scalable
and easily adaptable for clinical monitoring and therapeutic applications such as intracranial monitoring and
brain machine interfaces.
Results
Preliminary design and computational validation. A computational simulation was constructed to
investigate the spectrum of three-dimensional electrode geometries attainable with our scalable fabrication pro-
cess. e electrode array performance can be characterized by several interdependent parameters: the neural
selectivity, sensitivity, and long-term durability. Whereas the selectivity increases by minimizing the geomet-
ric area (GA) of the electrode to achieve more localized recordings, it does nevertheless augment electrode
impedance while reducing signal quality and the attenuated sensitivity. e electrode sensitivity can, however, be
increased by maximizing the electrochemical surface area (ESA) through surface modications like conductive
coatings, which reduce electrode impedance and improve signal quality9. Electrode durability mainly depends
on the structure, material biocompatibility and usage such as electrical stimulation, which has proven to reduce
sensitivity thereby aecting the longevity of many neurostimulation medical devices22,23. Assuming that the
material biocompatibility and stimulation protocols have been properly selected to maximize the longevity
of the electrodes, optimizing GA and ESA for a given electrode structure and composition is a fundamental
requirement of MEA and neuro-implant designs.
In the work presented here, the GA of three-dimensional microelectrodes was controlled by the electrode
height and diameter while the ESA was calibrated by the uninsulated electrode surface area. In addition to its
insulating properties, an SU-8 (biocompatible polymer) coating surrounding the gold wires was also added to
improve electrode strength and durability. To characterize the electrical and mechanical contributions of the SU-8
coating, our group developed two computational simulations using the Electric Currents and Solid Mechanics
modules in COMSOL Multiphysics (COMSOL Inc., Burlington MA). e goals of these simulations were to char-
acterize the electrode impedance and relative mechanical behavior of 3D electrodes in response to axial loading.
e impedances calculated using our model ranged from 45kΩ to 12MΩ (Fig.1c., 50µm gold wire with
maximum ESA and 10µm gold wire with minimum ESA). We however found relatively small variations in
electrode impedance with varying SU-8 thickness. e trend in our simulated impedances indicates decreasing
impedance with increasing ESA, which is consistent with trends reported in literature24.Furthermore, our simu-
lated impedances are consistent with values reported for commercially available 3D electrodes in 0.9% saline,
which validated the accuracy of our simulation3,4,25.
is computational analysis also demonstrated that the SU-8 thickness played a prominent role in our
mechanical simulations (Fig.1b). Simulated critical load factors varied between 1.38 and 169, which suggests
that thicker SU-8 coatings delay the onset of buckling. e critical load factor is a safety factor describing the
ratio between the buckling load to the applied load. e relative trend in critical load factors for the same load
applied to dierent geometries can be used as a proxy for characterizing electrode strength and stability. e
mechanical simulations suggested that the SU-8 coating improved the structural strength and stability of our
3D electrodes, which was important for mitigating dynamic factors in an invitro and invivo environment and
achieving stable recordings.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

3
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
Figure1. (a) Isometric view of two 3D electrodes showing the simulated 3D electrode geometry. e
geometry of the green SU-8 insulation layer surrounding the gold 3D electrodes was approximated from the
true appearance of a 3D electrode resulting from our fabrication process: a thick base that gradually turns into
a uniform coating around the sha due to the “wicking eect” from surface tension. Both electrodes have the
same overall electrode height, SU-8 insulation height, and 3D electrode diameter, but dierent SU-8 insulation
thicknesses. (b) Buckling simulation showing two 3D electrodes with dierent SU-8 thicknesses that are both
subject to an identical, arbitrarily chosen, load at the tip. e body of the electrode with thicker SU-8 is deected
less. (c) Simulated electrode impedance at 1kHz for dierent electrode diameters and insulation heights. Each
curve represents the impedance of a 300µm tall 3D electrode with various diameters of SU-8 insulation at a
constant height. e trends in this plot indicate the correlation between uninsulated electrode surface area and
the impedance of the electrode. As electrode diameter increases, impedance decreases due to a larger surface
area. As the insulation height decreases, impedance decreases due to a larger uninsulated electrode surface area.
(d) Simulated critical load factors for 300µm tall 3D electrodes of dierent diameters, SU-8 thickness, and
SU-8 height. Each plane represents a dierent gold electrode diameter. An increasing gold electrode diameter
increases the load bearing capacity of the electrode structure. e load bearing capacity of a particular electrode
diameter can be optimized by increasing the SU-8 height and thickness.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

4
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
e results of these simulations provided a basis for future experimental validation of the range of three-
dimensional electrode geometries achievable with our fabrication process. Optimizing the SU-8 geometry for
gold electrode diameters of 25µm (~ 1mil) and 50µm (~ 2mil) was of particular interest as these are considered
standard wire sizes in the electronics industry for wire bonding, a common and inexpensive method of creating
integrated circuitry connections, allowing for increased scalability. Furthermore, our simulations provided a
foundation for optimizing the electrode design in terms of low impedance and mechanical robustness for any
custom wire diameter, and/or wire material, thus allowing us to maximize reusability and electrical performance
and create durable, high density, and high-performance three-dimensional invitro electrode arrays.
Flexible fabrication process of three-dimensional gold microelectrodes for in vitro record-
ings. Based on the initial computation and simulation parameters, micro-electrodes were designed and fab-
ricated to record from brain tissue invitro. Standard gold planar microelectrodes were rst fabricated using
conventional photolithography techniques26 (Fig.2a, i to iii). Microelectrode sizes and intervals were adjusted
according to experimental needs by modifying the photomask designs thus allowing us to keep the design rela-
tively simple, economical, and scalable. ree-dimensional gold microwires were then individually bonded onto
the surface of the planar electrodes and coated with SU-8 (Fig.2a, iv to vi). Only the bases and edges of the
microwires were insulated with SU-8, leaving the conductive tips electrically exposed. is method yielded full
Figure2. (a) Side view schematic of the fabrication process of the three dimensional gold microelectrodes:
i to iii, metal electrode deposition and patterning; iv to v, ree-dimensional spike electrode fabrication onto
planar electrodes, with the dashed line indicating break height; vi, insulation layer deposition and vii, schematic
showing only the bare tips of the electrodes directly interfacing with neural cells located inside the brain slice,
i.e. within the remaining healthy tissue. (b) From le to right; Le: picture of the whole 49mm × 49mm square
Multi-Electrode Array chip. e rectangular gold pads on the four sides allow the connection between the
recording setup and the three-dimensional microelectrodes present in the center. Middle: magnied image
showing a section of the array of gold three-dimensional electrodes under an optical microscope. e three-
dimensional microelectrodes shown are separated by a 500µm gap from each other and are connected to
the outer rectangular pads described previously by mean of their respective micro-wires. Right: picture of an
uncoated tip from a three-dimensional electrode obtained with a Scanning Electron Microscope (SEM). e
electrode diameter and the sharpness of the tip allows for optimal penetration within the brain slice, which
limits the invasiveness.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

5
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
control and exibility in the size of each single three-dimensional electrode (diameter and height of wires used),
inter-electrode spacing, and materials used (gold in the present example, but palladium, platinum etc. could also
be implemented to improve electrical stimulation properties). e electrodes were individually fabricated and
electronically addressable. e three-dimensional gold electrodes reported here were purposefully fabricated
with a targeted height of 300μm (average = 266μm, standard deviation = 27μm, n = 30). Gold wire diameters of
either 17μm or 25µm were used. ese heights, spacing and materials were chosen specically for recording
from 400µm thick acute brain slices of mice (Fig.2a, vii), adjustable to suit any other experimental requirement.
Following the fabrication, we characterized and validated the morphological attributes of the spike microelec-
trodes with optical microscopy and SEM (Fig.2b).
Aer cleaning and sterilizing these three-dimensional MEAs, poly-D-lysine was used as a coating material to
enhance the interfacing between electrodes and neurons27. Acute hippocampal brain slices (400µm thick) from
C3HeB/FeJ mice littermates (P35) were then positioned with the help of an optical microscope in a recording
chamber and anchored onto the electrodes arrays by means of a weighted mesh to prevent movement and to
facilitate the penetration of the three-dimensional electrodes into the slice (Fig.3a,b). Penetration depth was
not measured, but is assumed to be consistent across experiments. Aer positioning mouse acute hippocampal
brain slices over the electrodes in an activity triggering articial cerebrospinal uid (aCSF, 0mM Mg2+ or high
8.5mM K+), spontaneous neural activity was consistently recorded (n = 55, 98% of the time) invitro at multiple
electrode sites and across the dierent channels. e recorded activity could be tracked within the entire brain
slice and the patterns of propagation were identied (Fig.3c–e; video available in supplementary material). is
activity consisted of bursting ictal events, either localized or spread across the slices, with high frequency activity
(~ 80Hz) oen observed in mammalian brain systems expressing hypersynchronous activity28–32 (Fig.3c). is
provided us with the opportunity to track high frequency bursting activity between dierent areas of the brain
slice and to analyse its overall excitability. For example, while we anticipated recording neural activity within
the well-studied region of the hippocampus (Fig.3e), we also recorded high-frequency bursting activity in the
mid-brain and thalamic regions (Fig.3d). Unlike previously reported data obtained using three-dimensional
electrodes13, this bursting activity was reliably and consistently traceable within a brain slice, thus permitting
the identication of neural network pathways.
e SNR of the gold three-dimensional microelectrodes was compared with earlier reported devices includ-
ing well characterized9,26,27 traditional planar microelectrodes. ese new 3D microelectrodes oered a reduced
mean noise of 20μV (compared to 50–70μV for traditional planar microelectrodes), which we attribute to the
insulating coating present at the bases and edges of the 3D structure. e highest recorded eld potential activity
peak-to-peak with the 3D microelectrodes was 3.2mV, compared with signals of < 1mV with traditional planar
microelectrodes9. Overall, these new 3D microelectrodes oered a much higher SNR (> 300% increase) than
previously reported devices using traditional planar microelectrodes. Finally, as the bases and edges of these 3D
microelectrodes were insulated, we recorded activity from within the brain slice where most electrically active
cells remain viable4, thus enabling continuous recordings over a period up to 4h. As expected, based on the
computational simulation, the insulation on the electrode edges provided structural support, which reduced the
physical degradation of the three-dimensional microelectrodes over time and made the MEAs reusable for at
least 15 times in an invitro setting. e MEAs showed no signs of damage aer 15 tests, but no further destruc-
tive testing was performed. An impedance check is completed by MC_Rack prior to each experiment wherein
each electrode is veried to be functional to ensure proper connection and recording capabilities. is serves as
a baseline metric for reusability.
We next sorted to determine whether our newly fabricated electrodes would also have utility as implantable
devices in-vivo. We therefore transferred the fabrication process to develop implantable exible substrate for
recording induced seizures in freely behaving animals to seek further the electrodes clinical potential.
Transfer onto exible substrate for in vivo epilepsy recording. e use of three-dimensional elec-
trode array in-vivo in humans and other animals has been demonstrated in multiple ways, oen using devices
such as the Utah Array or Michigan Array19,33. ese penetrating recording tools have been used to improve both
the signal quality and spatial resolution simultaneously. Several publications have however reported that 3D
electrodes developed on a rigid substrate result in scar formation around the recording or stimulating sites34,35.
ese scars in turn dramatically reduce the signal resolution oered by the recording electrodes and thus aect
the longevity of the implantable neuromodulation devices. In addition to recording from only a very limited
area of brain slice and therefore not providing enough coverage (e.g. intracranial monitoring for the evaluation
of epilepsy surgery), the three-dimensional electrode arrays use has remained limited to research applications
and not extended to clinical practice. We deemed it essential that in order to limit tissue scar formation, which
is a deterrent to long-term recordings, the substrate would need to be modied and made exible such as tradi-
tionally used ECoGs36 in clinical practice, or exible arrays for animal research18. We therefore next took on the
challenge of embedding three-dimensional electrode array onto a exible substrate to merge both properties:
high signal resolution and improved longevity.
Adapting a similar fabrication process to the one described in the previous section, we modied the wire
bonding parameters to create three-dimensional electrodes on exible substrates from the printed circuit board
(PCB) industry. Following the design of a neocortical implant tailored to record activity from a freely behav-
ing rodent, two surface nishes were tested: electroless nickel immersion gold (ENIG) and electroless nickel
electroless palladium immersion gold (ENEPIG). Such processes oen use polyimide or polyethylene tere-
phthalate (PET) as a base which has already been demonstrated to be biocompatible and relatively inert37.
ree-dimensional micro-electrodes with similar dimensions to the one developed on the glass substrate were
then added to the exible substrates (Fig.4a). No dierence in bonding success was noticed between ENIG and
Content courtesy of Springer Nature, terms of use apply. Rights reserved

6
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
ENEPIG. e bases and edges of the electrodes were then coated with SU-8 using a similar process to the one
used for the MEA fabrication (Fig.4b). e animal implants were then rinsed with autoclaved deionized water
and sterilized with 70% ethanol.
Adult Sprague–Dawley (SD) rats (n = 4) were anesthetized to proceed with the implantation (Fig.4c). Holes
were drilled into the skull for a ground screw, 2 anchor screws, and a twisted bipolar electrode for chronic implan-
tation in the corpus callosum of the right hemisphere. e bipolar electrode was used to electrically stimulate
the rat brain in order to elicit electrographic seizures, a well-established technique called electrical kindling38.
A small craniotomy was performed to expose the le caudal forelimb area of the motor cortex and the dura was
removed to expose the neocortex (Fig.4c). e exible substrate with embedded three-dimensional electrodes
was then placed on the exposed motor cortex and the entire assembly on the skull was cemented in place with
dental cement (Fig.4d).
Figure3. (a) MEA positioned into the recording system (MEA1060; Multichannel Systems, Reutlingen,
Germany) with an acute hippocampal brain slice positioned at the center of the circular chamber. A mesh is
placed on top of the slice to prevent any movement and facilitate the penetration of the three-dimensional
electrodes into the slice. aCSF saturated with carbogen at a temperature of 33.0°C was perfused inside the
chamber (right tube) and siphoned at the bath level interface. (b) Picture taken with an optical microscope
of an acute hippocampal brain slice positioned on an array of electrodes. Areas such as the mid-brain (MD),
thalamus (TH) and hippocampus (H) are clearly visible. (c) Example of bursting neural eld potential recorded
with the three-dimensional gold microelectrodes (high 8.5mM K+). Various patterns of activity were recorded,
including ictal like events typically seen during seizure-like events. (d) Activity recorded simultaneously within
the thalamus and mid-brain from the same brain slice. Notice moving ictal timeframes indicative of seizure
propagation (b). (e) Simultaneously, activity was recorded within the hippocampus (CA1 (H3/H4/H5), CA2
(H2) and CA3 (H1)).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

7
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
Following a minimum 7days recovery period, an electrographic seizure (aer discharge) was elicited with
the stimulating electrode using a Grass S88 stimulator (Natus Neurology, Warwick, RI) and the seizure duration
measured and severity scored based on Racine’s scale39. All rats consistently had a minimum behavioral stage 4
seizure, indicated by a generalized (evolved) seizure involving bilateral forelimb clonus. Simultaneously, local eld
potentials were recorded through the exible implant using a custom-made wireless recording unit (Neuraura
Biotech Inc., Canada). While the implant could record from up to 8 channels, electrical activity was recorded
through four channels at a time in this application at a sample frequency of 1kHz. Electrical aer discharges
characteristic of seizure activity were recorded through the three-dimensional micro-electrodes (Fig.4e,f). e
signal-to-noise ratio was again calculated, by taking the ratio of the maximum peak-to-peak amplitude of the
signal during seizure onset to the mean noise during inter-ictal periods, (max SNR above 10) and compared to
current standards9,26, providing conrmation that the three-dimensional electrodes used in-vivo have a SNR
signicantly higher than traditional stainless-steel electrode traditionally used in research settings. Seizure propa-
gation could also be traced between the three-dimensional electrodes demonstrating that electrographic seizures
can be detected at a sub-millimeter level. Resulting from this increased SNR, HFOs were also detected in 100%
of recorded seizures with ripples (blue, 80–250Hz) and fast ripples (red, 250–500Hz) bands being clearly dis-
tinguishable from background activity.
Discussion
e eld of neuroscience and its clinical applications are in dire need to design and develop neural-interfaces and
neuro-prosthetic devices that would enable high resolution recordings of brain activity over an extended time
period. In this study, we rst generated a computational simulation paradigm. is enabled us to demonstrate that
the addition of SU-8 with dierent thicknesses and height around a conductive wire would dramatically aect the
electrical characteristic as well as the structural stability of the micro-electrode. As a result, an appropriate balance
needed to be realized when fabricating micro-electrodes to optimize the impedance and structural robustness.
e ideal electrode would have the lowest impedance while being the most mechanically robust. Based on the
results of our simulation (Fig.1c,d), a 300µm tall and 50µm diameter gold electrode with 10µm SU-8 thick-
ness provides the largest critical load factor and is therefore the most mechanically robust. e impedance of
the electrode is a function of the SU-8 insulation height which determines the electrochemical sensing area of
the electrode. For a 300µm tall and 50µm diameter electrode with 10µm SU-8 thickness, the impedance values
vary between ~ 50 and 150kΩ for SU-8 insulation heights between 50 and 250µm. Furthermore, increasing
SU-8 insulation heights also resulted in increased critical load factors. Due to the highly compliant nature of
our three-dimensional gold micro-electrodes, a larger critical load factor was preferred to make the electrode
more mechanically robust. However, as we did not have access to 50µm diameter gold wire, 25µm and 17µm
gold wire were used in our experimental work. Because of the highly adaptable and scalable characteristics of
this fabrication process, unique parameters could be adapted to meet dierent experimental and clinical needs.
As demonstrated in the in-vivo and in-vitro experiments, the three-dimensional micro-electrodes presented
here oer an improved recording capability when considering signal-to-noise ratio, detection of HFOs and
traceability of neural activity.
e three-dimensional microelectrodes presented here were fabricated using thermosonic gold wire bond-
ing—a widely used process in the electronics industry. is approach resulted in improved manufacturability and
fabrication in comparison to previously reported methods of fabricating 3D electrode arrays like the Utah array.
Previously reported methods involved multi-step micromachining processes comprising of multiple stages of
material deposition followed by selective material removal40. ese methods are relatively costly, and inherently
susceptible to inconsistencies in the supply chain due to multiple processes involved and the individual complexi-
ties of each fabrication step. is hinders process scalability for large scale manufacturing. In contrast, while the
wire bonder used in this work was manual, our primarily additive wire bonding process could be performed
with an industrial scale fully automated wire bonder which can form between 5 and 12 wires (3D electrodes) per
second once fully optimized for a particular electrode geometry, with each electrode as ne as 40µm apart. With
our additive wire bonding process, we could match and/or exceed the electrode density (electrodes per area) of
any commercially available in-vitro 3D microelectrode array presently available (to the best of our knowledge9).
e electrode pitch on the exible probes used in our in-vivo experiments was ner than previously reported
in-vivo exible 3D arrays, thus allowing higher spatial resolution41. e SNR achieved in our invivo experiments
was also much higher than reported for other exible 3D arrays implanted in the rat motor cortex41.
e three-dimensional microelectrodes presented here were fabricated using readily available materials and
processes used in the electronics industry and optimized for recording neural signals. Our implanted electrodes
not only remained in place and monitored induced seizures but they also did not cause scarring. Although we did
not attempt to record spontaneous activity of either individual or populations of neurons but we are condent
that these electrodes should be able to pick eld potentials from cortical neurons. is may however, require
further renement and the use of gold wires with larger diameter to create gold electrodes; parenthetically, this
may also further decrease the impedance. Furthermore, while the electrodes presented here are made of gold for
its lower impedance and higher recording capabilities, the fabrication process can also be modied to include
other materials that are better suited for electrical stimulation capabilities. For instance, platinum and iridium
can be investigated rst as these materials are wire bondable, biocompatible and known to be superior to gold
for neurostimulation42,43. is would however require a wire bonder used in large scale industrial processes, and
such was not available for this work. Such an approach will eliminate the need for having to stimulate neurons
using external electrodes and could thus also be used to perturb neurons during various behaviors in freely
moving animals. Finally, conductive coatings such as poly(3,4-ethylenedioxythiophene) polystyrene sulfonate
(PEDOT:PSS) and carbon nanotubes can be applied to three-dimensional gold electrodes to improve the ESA
Content courtesy of Springer Nature, terms of use apply. Rights reserved

8
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
even further while simultaneously improving the charge injection capacity of the electrodes thus optimizing the
geometry for stimulation26,44 even further.
Finally, because the advances in the eld of neuro-technology are currently hampered by our inability to
record brain activity at a high SNR and to stimulate neurons safely, many opportunities in the area of neuromodu-
lation and brain machine interface (BMI) can be unlocked due to our improved and exible neural electrodes. By
improving the SNR of the recording electrodes in feedback loop systems, this would facilitate the data cleaning
and analysis steps, and therefore permit improved stimulation output of smart neuro-implants. For example,
when considering the practice of epilepsy surgery and neuro -stimulation treatment, the electrodes presented in
this study can provide 20 × increase spatial resolution versus existing ECoGs, maintain large coverage of brain
areas versus the Utah Array, and provide at least 3 × improved SNR compared to traditional ECoGs. In the case
of epilepsy surgery, these improved capabilities will enable neurosurgeons to resect pathological tissue from the
patients with more precision. is will also facilitate brain mapping and fasten diagnostic process/foci identi-
cation, allowing for the full spectrum of neural activity to be studied, including the dierentiation between
“normal” and “pathological” HFOs45. In turn, pathological HFOs have been recognized as potential biomarkers
for epileptogenic zones and resection or stimulation of brain tissue eliciting such activity has been shown to lead
to better post-surgical outcomes46,47.
is new technology allows for a far greater capture of near real time neural activity. With improved record-
ing capabilities, responsive neurostimulation can be further improved, leading to signicant impacts in medi-
cal implants ranging from epilepsy and seizure management48,49, occipital recording/stimulation for visual
prosthesis50 (Second Sight, US), language prosthetics51, or limbs prosthetics52. In turn, an improved feedback
loop system that stimulates the brain only when required has the opportunity to reduce the formation of scar
tissue and improve longevity of implants, making them more widespread and successful (if not more) in a man-
ner analogous to that of a cardiac pacemaker.
Manufacturing and methods
Simulation. To reduce computational time, the simulated 3D electrode geometries were treated as cylinders.
A 300µm tall gold cylinder of xed height and diameters between 10 and 50µm was used to model the 3D elec-
trode, and a 200µm diameter gold pad was used to model the electrode base. e SU-8 coating was modelled in
two parts to approximate the SU-8 structure achieved with our fabrication process. e approximation consists
of a conical lower portion with a height of 50µm, a xed 100µm base diameter and varying upper diameters
such that the thickness of SU-8 around the gold wire varied between 2.5 and 10µm. e upper portion of
the insulation was modelled as annular cylinder, continuous with the lower portion, with thicknesses varying
between 2.5 and 10µm, and heights varying between 0 and 250µm (only a conical SU-8 base present and only
the tip of the gold wire is exposed, leaving and area of 20–30µm).
For the impedance simulation, the 3D electrode was separated 500µm from a 500µm diameter × 500µm tall
AgCl counter-electrode and surrounded by a 1mm diameter × 1mm tall cylinder of physiological saline (0.9%
NaCl). e upper and lower boundaries of the cylinder were assigned insulating boundary conditions (no nor-
mal current ow) while the circumferential boundaries were set as ground (V = 0V) to represent a semi-innite
medium. e counter-electrode was assigned a xed potential of V = 1V and a current of 10nA was applied at
the 3D electrode. e properties of the electrode–electrolyte interfacial layer were approximated using litera-
ture values for Au in 0.9% NaCl1, and electrode impedances were calculated at 1kHz as the ratio between the
RMS voltage and current. For the mechanical simulations, the lower boundaries of the gold wire and SU-8 base
were assigned xed conditions, while the upper boundary of the wire was treated as a free end. e interior Au/
SU-8 boundaries were conditioned to prevent penetration between the boundaries and move together during
deformation. To characterize the relative trend in load bearing capacity of dierent 3D electrode structures, an
arbitrarily chosen 1
mN
m
2 downwards longitudinal load was applied to the tip of the electrode and the resulting
critical load factors were compared. A default physics-controlled ne mesh was used for all cases, with relatively
small variations with ner mesh elements.
MEA fabrication. Planar microelectrodes in an 8 × 8 array conguration were fabricated using standard
photolithography onto a 49 × 49mm, 1mm thick glass substrate. e glass was coated with ~ 400nm of gold,
deposited on a 50nm chrome adhesion layer using a sputter deposition process (CMS-18, Kurt K Lesker Co.,
Pennsylvania, USA). A positive photoresist, (HPR504, Fujilm, USA) was then spin coated onto the chrome/
gold coated glass slide (3500rpm, WS-650-23B, Laurell Technologies Corp., North Wales, Pennsylvania, USA).
Figure4. Fabrication process, implantation, and results of in-vivo electrodes. (a) Bonding on a planar array
of micro-electrodes positioned 350µm apart to create three-dimensional electrodes. (b) Array of three-
dimensional micro-electrodes with coating at the base and edges of the structures. (c) Adult Sprague–Dawley
rats where anesthetized and (d) exible implants were positioned on the motor cortex following a craniotomy.
(e) Bipolar stimulation electrodes implanted in the corpus callosum of the rat right hemisphere were used to
induce seizures through electrical stimulation (electrical kindling) and recorded through Neuraura’s micro-
sensors. Only the four channels recording relevant activity are shown here. Raw LFP recordings were bandpass
ltered in the HFO frequency ranges using a nite impulse response (FIR) lter. Raw recording are depicted in
green, ltered ripple (80–250Hz lter) in blue, and ltered fast ripple (250–500Hz lter) in red. Purple markers
show the beginning of seizure recording across channels. Amplier used for in-vivo recordings outputted data
in volts. (f) Detailed view of one channel showing action potentials during the baseline, ictal and postictal
periods.
▸
Content courtesy of Springer Nature, terms of use apply. Rights reserved

9
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
Content courtesy of Springer Nature, terms of use apply. Rights reserved

10
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
e photo resist was so baked on a hot plate for 90s (110°C) and then exposed using a positive photomask
(Photoscience Inc., California, USA) using a Mask Aligner (MA/BA 6, Suss Microtec, Corona, California, USA),
and developed using Microposit developer 354 (Dow Chemical Corp., Midland, Michigan, US).
A Cr/Au wet chemical etch was performed to dene the multielectrode array and any remaining photoresist
was removed using acetone. Sizes and intervals between the circular microelectrodes were adjusted according to
experimental needs using dierent photomask designs in the photolithography process. For the work reported
here, electrodes of 100µm diameter, with inter-electrode spacing of 500µm, were fabricated (Fig.2a, ii). An
epoxy photoresist (SU-8) was then spin coated over the entire electrode array. e SU-8 layer was patterned
with a second photomask to leave the planar microelectrodes bare of SU-8 insulator, but with their connecting
wires insulated (Fig.2a, iii).
Forming three-dimensional microelectrodes. Spike-shaped three-dimensional gold microelectrodes
were added onto the surface of the planar microelectrodes using a manually programmable wire bonder (West-
Bond model 454647E, West-Bond Inc., USA). ese spikes were created by bonding gold wires onto the planar
electrodes, manually extending the wires to a set height and then severing them. Finally, a second insulating
layer (SU-8) was locally deposited around the bases and on the sides of the newly formed three-dimensional
gold microelectrodes, covering approximately 85% of the electrode height, using a micropipette and a manually
operated micromanipulator. is process allowed the three-dimensional electrode tips to be le bare of insula-
tor for the last few micrometers, so that they would be in direct contact with the healthy neural cells inside the
brain slice. As with the rest of this highly customizable process (substrate footprint, electrode layout, electrode
pitch, electrode height, electrode surface area), the amount of surface area in direct contact with the cells could
be adjusted depending on experimental needs.
ree-dimensional electrodes were wire bonded to gold planar pads using a WestBond model 454647E
manual wire bonder set to ball bonding mode (West-Bond Inc., USA). A 0.7mil (17.78µm) diameter or 1mil
(25.4µm) diameter gold wire was used. ermosonic gold ball bonding was chosen, over other biocompatible
wirebondable metals like platinum, due to limitations on the model of wire bonder that was used. Ultrasonic
power of 0.6W was applied for a period of 200ms. Aer bonding gold wire to a planar gold pad, the bonding
capillary was moved upwards 350µm and the wire was severed manually using microscissors. is process was
repeated to create each three-dimensional microelectrode.
MEA washing, cleaning and sterilization procedure. Aer inspection for any defects or debris under
an optical microscope, MEAs were rinsed extensively (30–50 times) with ltered, autoclaved distilled water. is
ensured complete removal of any fabrication residues deemed harmful for biological tests. Also, following every
experiment, MEAs were washed in a similar way to remove cell residues and previous coatings. To completely
remove any cell residues, MEAs followed a cleaning procedure using a 10% tergazyme solution. Again, MEAs
were extensively rinsed with ltered, autoclaved distilled water to ensure complete removal of any enzymatic
solution that would be considered harmful to the cells. On rare occasions, diluted household bleach (5.25%
Sodium Hypochlorite diluted with a ratio of 1:500) was used to remove residues or potential fungus. MEAs were
then sterilized with 70% ethanol for 30min and rinsed 3 times with ltered autoclaved distilled water.
Mice and brain slices. C3HeB/FeJ mice were purchased from Jackson Laboratories (Bar Harbor, ME,
U.S.A.), and the colony was maintained in the Animal Resource Facility at the Cumming School of Medicine,
University of Calgary. Mice were given food and water adlibitum and kept on a 12-h light/dark cycle. Wild type
(+ / +) mice at P35 (postnatal day 35) were used in this study and all procedures in this study were performed in
accordance with the recommendations in the Canadian Council for Animal Care. All animal research protocol
of this study were approved by the Health Sciences Animal Care Committee of the University of Calgary.
A detailed protocol is published elsewhere53. Briey, on the day of the experiment, animals were anesthetized
(Ketamine–Dormitor mixture; 0.1ml/100g; i.p.), sacriced, and their brains were removed and quickly placed
into ice cold, oxygenated (95% O2 / 5% CO2) articial cerebrospinal uid (aCSF; all in mM: 86 NaCl, 3 KCl, 4
MgCl2, 1 NaH2PO4, 75 sucrose, 25 glucose, 1 CaCl2, and 25 NaHCO3). Coronal slices (Approx. 400μm; 1.5
to − 0.3mm relative to Bregma) containing the hippocampus were prepared using a Leica VT 1000S vibratome.
ese slices were then placed in aCSF containing (in mM) 124 NaCl, 4.5 KCl, 1 MgCl2, 10 glucose, 1 CaCl2,
and 26 NaHCO3 at 32°C for 30min to recover, and remained aerwards in a aCSF bath at room temperature
(22–24°C) until used. All solutions used during this process were maintained at pH 7.4 and bubbled with 5%
CO2 / 95% O2 (carbogen).
Neural activity recordings with MEAs and analysis. Neural activity was recorded by an MEA ampli-
er and PCI acquisition card (MEA1060; Multichannel Systems, Reutlingen, Germany) and visualization was
made using the soware MC_Rack and MEA_Select (Multichannel Systems, Reutlingen, Germany). Soware
TCX-Control was used to control the temperature regulator and heated cannula (respectively TC02 and PH01,
Multichannel Systems, Reutlingen, Germany). e recordings were compiled and processed using MC_DataTool
and MC_Rack soware respectively. More precisely, a spike detector present in MC_Rack allowed the extraction
of timestamp associated with each individual action potential. ese outputs were then imported and processed
by Excel (Microso; Redmond, WA, USA) to better analyze activity frequency or the time elapsed between two
adjacent action potentials, known as inter-spike intervals (ISI). Noise was measured using the baseline recording
in an aCSF solution without the presence of a brain slice where the electrodes were le for approximately 5min.
e maximum peak-to-peak amplitudes were then analyzed to calculate the signal to noise ratio.
Content courtesy of Springer Nature, terms of use apply. Rights reserved

11
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
Implant fabrication and preparation. Implants were rst designed using Altium Designer (Altium
Ltd., NSW, Australia), a common soware in the printed circuit boards (PCB) industry. e substrate footprint
was approximately 1.75mm × 8mm and designed to t within the available surface area on the motor cortex
of Adult Sprague–Dawley (SD) rats. Electrode pads were designed to be 150µm diameter and interelectrode
spacing was dened as 350µm to accommodate our 8-channel data acquisition hardware. e design can be
adapted to increase the number of electrodes and density depending on the number of channel inputs oered
by the amplication system being used. e design was sent to a ex PCB manufacturer for fabrication. Using
polyimide as a substrate, two processes were followed: electroless nickel immersion gold (ENIG) and electroless
nickel electroless palladium immersion gold (ENEPIG), both of which are surface nishes intended to provide
a gold wire bondable surface for 3D electrode formation to be added on top of the planar electrode pads. ENIG
applies a layer of nickel over the copper, followed by gold over the nickel. ENEPIG applies a layer of nickel over
the copper, followed by a layer of palladium over the nickel, followed by a layer of gold over the palladium. Once
received by our facility, the fabrication of spike-shaped three-dimensional gold microelectrodes and coating of
their edges and bases was done following the same protocol as for MEA fabrication.
Rats and surgery. Adult Sprague–Dawley (SD) rats (n = 4) weighing 252–300g at the time of surgery were
used in this study. All rats were obtained from Charles River (Quebec, Canada) and experiments were approved
by the Health Sciences Animal Care Committee at the University of Calgary. Experiments were carried out in
compliance with the ARRIVE guidelines. Rats were maintained on a 12-h light/dark cycle with lights on at 7am.
All experiments were conducted during the light phase. Rats received food and water adlibitum.
Rats were initially anesthetized with isourane at 5% and maintained at 1.5–2%54. e level of anesthesia was
monitored by assessing breathing rate and a withdrawal reex from a light foot pinch. Lidocaine 2% (20mg/kg,
volume of 0.5ml) was injected subcutaneously at the incision site. A single injection of buprenorphine (dosage
of 0.03mg/kg, volume of 0.05ml subcutaneous) and baytril (10mg/kg, 0.2ml subcutaneous) was given prior to
surgery. In addition, twice per day for 3days post-surgery, animals were individually given buprenorphine jello
(oral dose of 0.17ml); and once a day for 3days post-surgery, animals were given Baytril. Holes were drilled into
the skull for a ground screw (placed above the cerebellum), 2 anchor screws (located in proximity to craniotomy
to provide structural stability), and a twisted bipolar electrode (Teon-coated, stainless steel wire, A-M Systems,
Sequim, WA) crimped with gold male amphenol pins with a tip separation of ~ 1mm for chronic implantation
in the corpus callosum of the right hemisphere (− 5.2mm anterior, 1.0mm lateral, − 2.5mm ventral). A small
craniotomy was performed to expose the le caudal forelimb area of the motor cortex (2mm anterior from
bregma, 2mm posterior, 3mm lateral from midline). e dura was removed to expose the cortex. Body tempera-
ture saline was used to keep the exposed cortex moist. e exible 8 channel cortical grid electrode measuring
1.75mm × 2mm was placed on the exposed motor cortex. e entire assembly on the skull was cemented in
place with dental cement. An FFC connector and standard tip jack connectors (Cinch Connectivity Solutions
Johnson) were used to connect the array to the recording system. A minimum of 7days was provided for recovery
before experiments began.
Electrical kindling. An electrographic seizure (aer discharge) was elicited with a stimulation (60 Hz
biphasic square wave; 1s train; 1ms pulse widths) through the electrode in the corpus callosum using a Grass
S88 stimulator (Natus Neurology, Warwick, RI). e seizure duration and severity observed were recorded based
on Racine’s scale39. All animals consistently had a minimum behavioral stage 4 when kindled.
Received: 1 March 2021; Accepted: 22 October 2021
References
1. McNeill, A. S. Neuroscience Research Progress 1 Online Resource (Nova Science Publishers, 2011).
2. Brodie, M. J. Medical therapy of epilepsy: When to initiate treatment and when to combine?. J. Neurol. 252, 125–130. https:// doi.
org/ 10. 1007/ s00415- 005- 0735-x (2005).
3. Markram, H. Seven challenges for neuroscience. Funct. Neurol. 28, 145–151. https:// doi . o r g/ 10. 11138/ FNeur/ 2013. 28.3. 144 (2013).
4. Steidl, E.-M., Neveu, E., Bertrand, D. & Buisson, B. e adult rat hippocampal slice revisited with multi-electrode arrays. Brain
Res. 1096, 70–84. https:// doi. org/ 10. 1016/j. brain res. 2006. 04. 034 (2006).
5. Jun, J. J. et al. Fully integrated silicon probes for high-density recording of neural activity. Nature 551, 232–236. https:// doi. org/
10. 1038/ natur e24636 (2017).
6. Raducanu, B. C. et al. Time multiplexed active neural probe with 1356 parallel recording sites. Sensors 17, 2388 (2017).
7. Gong, X. W., Li, J. B., Lu, Q. C., Liang, P. J. & Zhang, P. M. Eective connectivity of hippocampal neural network and its alteration
in Mg2+-free epilepsy model. PLoS ONE 9, e92961. https:// doi. org/ 10. 1371/ journ al. pone. 00929 61 (2014).
8. Fromherz, P. ree levels of neuroelectronic interfacing: Silicon chips with ion channels, nerve cells, and brain tissue. Ann. N Y
Acad. Sci. 1093, 143–160. https:// doi. org/ 10. 1196/ annals. 1382. 011 (2006).
9. Spira, M. E. & Hai, A. Multi-electrode array technologies for neuroscience and cardiology. Nat. Nanotechnol. 8, 83–94. https:// doi.
org/ 10. 1038/ nnano. 2012. 265 (2013).
10. Kim, G. H. et al. Recent progress on microelectrodes in neural interfaces. Materials (Basel, Switzerland) 11, 2018. https:// doi. org/
10. 3390/ ma111 01995 (1995).
11. Buskila, Y. et al. Extending the viability of acute brain slices. Sci. Rep. 4, 5309. https:// doi. org/ 10. 1038/ srep0 5309 (2014).
12. Ting, J. T., Daigle, T. L., Chen, Q. & Feng, G. Acute brain slice methods for adult and aging animals: Application of targeted patch
clamp analysis and optogenetics. Methods Mol. Biol. 1183, 221–242. https:// doi. org/ 10. 1007/ 978-1- 4939- 1096-0_ 14 (2014).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

12
Vol:.(1234567890)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
13. Heuschkel, M. O., Fejtl, M., Raggenbass, M., Bertrand, D. & Renaud, P. A three-dimensional multi-electrode array for multi-site
stimulation and recording in acute brain slices. J. Neurosci. Methods 114, 135–148. https:// doi. org/ 10. 1016/ s0165- 0270(01) 00514-3
(2002).
14. Hai, A. et al. Spine-shaped gold protrusions improve the adherence and electrical coupling of neurons with the surface of micro-
electronic devices. J. R. Soc. Interface 6, 1153–1165. https:// doi. org/ 10. 1098/ rsif. 2009. 0087 (2009).
15. Suzuki, I., Fukuda, M., Shirakawa, K., Jiko, H. & Gotoh, M. Carbon nanotube multi-electrode array chips for noninvasive real-time
measurement of dopamine, action potentials, and postsynaptic potentials. Biosens. Bioelectron. 49, 270–275. https:// doi. org/ 10.
1016/j. bios. 2013. 05. 023 (2013).
16. Rajaraman, S., Bragg, J. A., Ross, J. D. & Allen, M. G. Micromachined three-dimensional electrode arrays for transcutaneous nerve
tracking. J. Micromech. Microeng. 21, 085014. https:// doi. org/ 10. 1088/ 0960- 1317/ 21/8/ 085014 (2011).
17. Rajaraman, S. et al. Metal-transfer-micromolded three-dimensional microelectrode arrays for in-vitro brain-slice recordings. J.
Microelectromech. Syst. 20, 396–409. https:// doi. org/ 10. 1109/ JMEMS. 2011. 21052 53 (2011).
18. Kaiju, T. et al. High spatiotemporal resolution ECoG recording of somatosensory evoked potentials with exible micro-electrode
arrays. Front. Neural Circuits 11, 20 (2017).
19. Maynard, E. M., Nordhausen, C. T. & Normann, R. A. e Utah Intracortical Electrode Array: A recording structure for potential
brain-computer interfaces. Electroencephalogr. Clin. Neurophysiol. 102, 228–239. https:// doi. org/ 10. 1016/ S0013- 4694(96) 95176-0
(1997).
20. Mitrasinovic, S., Brown, A. P. Y., Schaefer, A. T., Chang, S. D. & Appelboom, G. Silicon Valley new focus on brain computer inter-
face: Hype or hope for new applications?. F1000Research 7, 1327–1327. https:// doi. org/ 10. 12688/ f1000 resea rch. 15726.1 (2018).
21. Musk, E. An integrated brain-machine interface platform with thousands of channels. J. Med. Internet Res. 21, e16194–e16194.
https:// doi. org/ 10. 2196/ 16194 (2019).
22. Ghafoor, U., Kim, S. & Hong, K.-S. Selectivity and longevity of peripheral-nerve and machine interfaces: A review. Front. Neuro-
robot. 11, 59–59. https:// doi. org/ 10. 3389/ fnbot. 2017. 00059 (2017).
23. Warren, D. J. et al. Recording and decoding for neural prostheses. Proc. IEEE 104, 374–391. https:// doi. org/ 10. 1109/ JPROC. 2015.
25071 80 (2016).
24. Cui, H., Xie, X., Xu, S., Chan, L. L. H. & Hu, Y. Electrochemical characteristics of microelectrode designed for electrical stimula-
tion. Biomed. Eng. Online 18, 86. https:// doi. org/ 10. 1186/ s12938- 019- 0704-8 (2019).
25. Microsystem, B. NeuroPort Array PN 4382, 4383, 6248, and 6249 Instructions for Use. (Blackrock Microsystem, https:// www.
black rockm icro. com/ wp- conte nt/ ifu/ LB- 0612_ Neuro Port_ Array_ IFU. pdf, 2018).
26. Wijdenes, P. et al. A novel bio-mimicking, planar nano-edge microelectrode enables enhanced long-term neural recording. Sci.
Rep. 6, 34553. https:// doi. org/ 10. 1038/ srep3 4553 (2016).
27. Luk, C. C. et al. Trophic factor-induced activity ‘signature’ regulates the functional expression of postsynaptic excitatory acetyl-
choline receptors required for synaptogenesis. Sci. Rep. 5, 9523. https:// doi. org/ 10. 1038/ srep0 9523 (2015).
28. Gavrilovici, C., Pollock, E., Everest, M. & Poulter, M. O. e loss of interneuron functional diversity in the piriform cortex aer
induction of experimental epilepsy. Neurobiol. Dis. 48, 317–328. https:// doi. org/ 10. 1016/j. nbd. 2012. 07. 002 (2012).
29. Jones, N. A. et al. Cannabidiol displays antiepileptiform and antiseizure properties invitro and invivo. J. Pharmacol. Exp. er.
332, 569–577. https:// doi. org/ 10. 1124/ jpet. 109. 159145 (2010).
30. Simeone, T. A., Simeone, K. A., Samson, K. K., Kim, D. Y. & Rho, J. M. Loss of the Kv1.1 potassium channel promotes pathologic
sharp waves and high frequency oscillations in invitro hippocampal slices. Neurobiol. Dis. 54, 68–81. https:// doi. org/ 10. 1016/j.
nbd. 2013. 02. 009 (2013).
31. Maehara, T. Epilepsy and high frequency oscillation. No Shinkei Geka 43, 285–295. h t t ps:// do i . o r g/ 10. 11477/ mf . 14362 03010 (2015).
32. González Otárula, K. A., von Ellenrieder, N., Cuello-Oderiz, C., Dubeau, F. & Gotman, J. High-frequency oscillation networks
and surgical outcome in adult focal epilepsy. Ann. Neurol. 85, 485–494. https:// doi. org/ 10. 1002/ ana. 25442 (2019).
33. Ferguson, M., Sharma, D., Ross, D. & Zhao, F. A critical review of microelectrode arrays and strategies for improving neural
interfaces. Adv. Healthcare Mater. 8, 1900558. https:// doi. org/ 10. 1002/ adhm. 20190 0558 (2019).
34. Weltman, A., Yoo, J. & Meng, E. Flexible, penetrating brain probes enabled by advances in polymer microfabrication. Micromachines
https:// doi. org/ 10. 3390/ mi710 0180 (2016).
35. Campbell, A. & Wu, C. Chronically implanted intracranial electrodes: Tissue reaction and electrical changes. Micromachines 9,
430. https:// doi. org/ 10. 3390/ mi909 0430 (2018).
36. Amino, M. J. & Daro, R. B. Encyclopedia of the Neurological Sciences 2nd edn. (Academic Press, 2014).
37. Coombs, C. & Holden, H. Printed Circuits Handbook (McGraw-Hill, 2016).
38. Teskey, G. C. (Oxford University Press, 2020).
39. Racine, R. J. Modication of seizure activity by electrical stimulation: II. Motor seizure. Electroencephalogr. Clin. Neurophysiol. 32,
281–294. https:// doi. org/ 10. 1016/ 0013- 4694(72) 90177-0 (1972).
40. Bhandari, R., Negi, S. & Solzbacher, F. Wafer-scale fabrication of penetrating neural microelectrode arrays. Biomed. Microdevice
12, 797–807. https:// doi. org/ 10. 1007/ s10544- 010- 9434-1 (2010).
41. Xiang, Z., Liu, J. & Lee, C. A exible three-dimensional electrode mesh: An enabling technology for wireless brain–computer
interface prostheses. Microsyst. Nanoeng. 2, 16012. https:// doi. org/ 10. 1038/ micro nano. 2016. 12 (2016).
42. Yamagiwa, S. et al. Layer-by-layer assembled nanorough iridium-oxide/platinum-black for low-voltage microscale electrode neu-
rostimulation. Sens. Actuat. B Chem. 206, 205–211. https:// doi. org/ 10. 1016/j. snb. 2014. 09. 048 (2015).
43. Chung, T.-W. et al. Fabrication of iridium oxide/platinum composite lm on titanium substrate for high-performance neurostimu-
lation electrodes. Coatings 8, 420. https:// doi. org/ 10. 3390/ coati ngs81 20420 (2018).
44. Ghazavi, A. et al. Eect of planar microelectrode geometry on neuron stimulation: Finite element modeling and experimental
validation of the ecient electrode shape. J. Neurosci. Methods 248, 51–58. https:// do i . org/ 10. 1016/j. jneum eth. 2015. 03. 024 (2015).
45. Weiss, S. A. et al. Localizing epileptogenic regions using high-frequency oscillations and machine learning. Biomark. Med. 13,
409–418. https:// doi. org/ 10. 2217/ bmm- 2018- 0335 (2019).
46. omschewski, A., Hincapié, A.-S. & Frauscher, B. Localization of the epileptogenic zone using high frequency oscillations. Front.
Neurol. 10, 94 (2019).
47. Fedele, T. et al. Resection of high frequency oscillations predicts seizure outcome in the individual patient. Sci. Rep. 7, 13836.
https:// doi. org/ 10. 1038/ s41598- 017- 13064-1 (2017).
48. Sisterson, N. D., Wozny, T. A., Kokkinos, V., Constantino, A. & Richardson, R. M. Closed-loop brain stimulation for drug-resistant
epilepsy: Towards an evidence-based approach to personalized medicine. Neurotherapeutics 16, 119–127. https:// doi. org/ 10. 1007/
s13311- 018- 00682-4 (2019).
49. Wong, S., Mani, R. & Danish, S. Comparison and selection of current implantable anti-epileptic devices. Neurotherapeutics 16,
369–380. https:// doi. org/ 10. 1007/ s13311- 019- 00727-2 (2019).
50. Stronks, H. C. & Dagnelie, G. e functional performance of the Argus II retinal prosthesis. Expert Rev. Med. Devices 11, 23–30.
https:// doi. org/ 10. 1586/ 17434 440. 2014. 862494 (2014).
51. Anumanchipalli, G. K., Chartier, J. & Chang, E. F. Speech synthesis from neural decoding of spoken sentences. Nature 568, 493–498.
https:// doi. org/ 10. 1038/ s41586- 019- 1119-1 (2019).
52. Loeb, G. E. Neural prosthetics: A review of empirical vs. systems engineering strategies. Appl. Bionics Biomech. 2018, 1435030–
1435017. https:// doi. org/ 10. 1155/ 2018/ 14350 30 (2018).
Content courtesy of Springer Nature, terms of use apply. Rights reserved

13
Vol.:(0123456789)
Scientic Reports | (2021) 11:21952 | https://doi.org/10.1038/s41598-021-01528-4
www.nature.com/scientificreports/
53. Gavrilovici, C., D’Alfonso, S., Dann, M. & Poulter, M. O. Kindling-induced alterations in GABAA receptor-mediated inhibition
and neurosteroid activity in the rat piriform cortex. Eur. J. Neurosci. 24, 1373–1384. https:// doi. org/ 10. 1111/j. 1460- 9568. 2006.
05012.x (2006).
54. Brown, A. R., Coughlin, G. M. & Teskey, G. C. Seizures alter cortical representations for complex movements. Neuroscience 449,
134–146. https:// doi. org/ 10. 1016/j. neuro scien ce. 2020. 09. 002 (2020).
Acknowledgements
Devices were fabricated with the support of the Microsystems Hub at the University of Calgary. We also thank the
following for supporting this work: Alberta Innovates; Canadian Institute of Health Research (CIHR); Natural
Sciences and Engineering Research Council of Canada (NSERC), Hotchkiss Brain Institute (HBI), Biomedical
Engineering Program at the University of Calgary (Summer Internship Support), and CMC microsystems (design
soware support). We also acknowledge the support of Dr. Belostotski and Dr. Zailer at the University of Calgary
for their expertise and advice on wire bonder operation, as well as Wali Zaidi and Jean Kawasoe (operational sup-
port), Mark Simpson (wireless EEG) and Dr. Dunn’s lab (MRI testing). is work would not be possible without
the generous support of the Antje Graupe Pryor Foundation and its Werner Graupe International Fellowship in
Engineering, as well as Mitacs Accelerate program for PW.
Author contributions
P.W., K.H and C.D. fabricated, analyzed and tested the devices and their electrodes congurations. C.G. pre-
pared the mice brain slices and B.G. implanted electrodes in animals. P.W., T.L., R.A., M.D.W. conducted the
biological experiments. Figures and main manuscript text were prepared and written by P.W., C.D., K.H., T.L.
and N.S. In-vitro experiments were performed in the labs of N.S, in-vivo experiments were performed in the labs
of G.C.T. J.R. and N.S. participated in data interpretation and experimental design. All authors have reviewed
the manuscript.
Competing interests
Disclosure and competing nancial interests: is technology has been protected and a patent has been led
(Application Number: WO 2018/184104) Since this work was completed, a spin out company from the university
of Calgary has been formed (Neuraura Biotech Inc). P.W., B.G., K.H. and C.D. are involved in some capacity in
the development of this spin out. e remaining authors have no conicts of interest.
Additional information
Supplementary Information e online version contains supplementary material available at https:// doi. org/
10. 1038/ s41598- 021- 01528-4.
Correspondence and requests for materials should be addressed to N.I.S.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.
© e Author(s) 2021
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com