106 IEEE REVIEWS IN BIOMEDICAL ENGINEERING, VOL. 3, 2010
Dry-Contact and Noncontact Biopotential Electrodes:
Methodological Review
Yu Mike Chi, Student Member, IEEE, Tzyy-Ping Jung, Senior Member, IEEE, and
Gert Cauwenberghs, Senior Member, IEEE
Methodological Review
Abstract—Recent demand and interest in wireless, mobile-based
healthcare has driven significant interest towards developing alter-
native biopotential electrodes for patient physiological monitoring.
The conventional wet adhesive Ag/AgCl electrodes used almost
universally in clinical applications today provide an excellent
signal but are cumbersome and irritating for mobile use. While
electrodes that operate without gels, adhesives and even skin
contact have been known for many decades, they have yet to
achieve any acceptance for medical use. In addition, detailed
knowledge and comparisons between different electrodes are not
well known in the literature. In this paper, we explore the use
of dry/noncontact electrodes for clinical use by first explaining
the electrical models for dry, insulated and noncontact electrodes
and show the performance limits, along with measured data. The
theory and data show that the common practice of minimizing
electrode resistance may not always be necessary and actually lead
to increased noise depending on coupling capacitance. Theoretical
analysis is followed by an extensive review of the latest dry elec-
trode developments in the literature. The paper concludes with
highlighting some of the novel systems that dry electrode tech-
nology has enabled for cardiac and neural monitoring followed
by a discussion of the current challenges and a roadmap going
forward.
Index Terms—Biopotentials, electrocardiograms (ECG), electro-
encephalograms (EEG).
I. INTRODUCTION
B
IOPOTENTIAL recordings in the form of electrocardio-
grams (ECG), electroencephalograms (EEG), electroocu-
lograms (EOG) and electromyograms (EMG) are indispensable
and vital tools for both medical and research use. These well-
Manuscript received August 11, 2010; accepted September 21, 2010. Date of
publication October 11, 2010; current version December 08, 2010. This work
was supported in part by National Semiconductor, by the National Science
Foundation, by the NIH/NIA, and by the Defense Advanced Research Projects
Agency.
Y. M. Chi is with the Department of Electrical and Computer Engineering,
Jacobs School of Engineering, University of California, San Diego, CA 92093
USA (e-mail: m1chi@ucsd.edu).
T.-P. Jung is with the Swartz Center for Computational Neuroscience, Insti-
tute for Neural Computation, University of California, San Diego, CA 92093
USA (e-mail: tpjung@ucsd.edu).
G. Cauwenberghs is with the Department of Bioengineering, Jacobs School
of Engineering, and Institute for Neural Computation, University of California,
San Diego, CA 92093 USA (e-mail: gert@ucsd.edu).
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/RBME.2010.2084078
proven signal modalities provide a wealth of physiological in-
formation, which by virtue of modern bioinstrumentation tech-
nology can be harnessed noninvasively and inexpensively for
the emerging global health applications of clinical physiolog-
ical monitoring and medical treatment [1], [2].
Traditionally, Ag/AgCl electrodes with wet conductive gels
are used for biopotential recordings. The standard Ag/AgCl
electrode has been well-characterized and studied over many
decades [3]–[5]. Most of its properties are well understood [6],
and sufficient empirical data exists for mechanism that are not,
such as low-frequency noise and drift [4]. Nevertheless, with
proper preparation, the signal is excellent.
The basic principles behind gel-less electrodes are also well
known. Despite decades of research in alternative biopotential
sensor technologies [7]–[10] for ECG and EEG applications,
the standard wet Ag/AgCl electrode is still almost universally
used for clinical and research applications. Each year billions
of disposable adhesive ECG clinical electrodes are produced,
while dry electrodes are limited to niche, nonmedical/scientific,
applications like fitness monitoring and toys.
The usefulness and performance of dry and noncontact elec-
trodes can be divided in to two categories. The first relates to
the to the signal quality of the device in terms of noise and
motion sensitivity. Second, because electrodes interface to the
skin either in contact or close proximity to the body, the spe-
cific electrode must also be evaluated for comfort and utility at
the system level. This paper aims to critically address the latest
developments in dry and noncontact electrodes accounting for
both of these considerations. One chief advantage of the stan-
dard clinical wet electrode is the fact that it adheres very well
to skin. While problematic from a patient comfort standpoint
for long-term use, adhesive wet electrodes stay fixed to spe-
cific, clinical-standard locations on the body. Dry and noncon-
tact electrodes address the comfort issues with the adhesive wet
electrode, but are much more difficult to secure against the pa-
tient. Thus for these technologies to be clinically useful, me-
chanical solutions must be devised to place the electrodes in the
proper position (such as the 12-lead ECG) or an alternative ap-
plication niche must be found. It is for these reasons, that dry
and noncontact electrodes are unlikely to replace the standard
hospital ECG or EEG electrode.
The literature around dry electrode technology is quite vast,
but dispersed across multiple, semi-isolated, research groups
and publications. In addition, the amount of information is com-
pounded by all of the possible applications (ECG, EEG, etc).
1937-3333/$26.00 © 2010 IEEE
CHI et al.: DRY-CONTACT AND NONCONTACT BIOPOTENTIAL ELECTRODES: METHODOLOGICAL REVIEW 107
Fig. 1. Electrical coupling of skin-electrode interface for various electrode topologies, including wet-contact gel-based Ag/AgCl, dry-contact
MEMS and metal
plate, thin-film insulated metal plate, and noncontact metal plate coupling through hair or clothing such as cotton. Insets show examples of practica
l electrodes for
each category as described in Section III.
With that in mind, this paper reviews the latest developments in
dry/noncontact electrodes while providing a historical context
and a discussion of the challenges and future directions for this
field. In 2000, Searle et al. [3] published a detailed comparison
between standard wet Ag/AgCl and their specific implementa-
tion of a dry and insulating electrodes from an impedance, in-
terference motion artifact rejection perspective. In contrast to
conventional wisdom, their paper demonstrated that dry and in-
sulate electrodes (if buffered and shielded) can perform as well,
if not better than, standard wet Ag/AgCl electrodes in each of
these respects. However, the intrinsic noise properties of the
electrode were not discussed and the paper was limited to only
two, specific dry and insulated electrode implementations.
This paper presents a systematic comparison between the var-
ious contact and noncontact electrode technologies with a focus
on quantifying the noise performance and motion sensitivity
as a function of physical and electrical parameters, as well as
their unobtrusiveness and ease for clinical use. The following
section presents a general model of the electrode interface, de-
scribed and characterized with measurements from an electrical
perspective. This establishes the fundamental principles for dry
and noncontact electrodes and describes the fundamental signal
quality limits. The different electrode technologies and their
properties are surveyed next, and the paper concludes with a
discussion of the latest developments in the literature along with
future directions and challenges.
II. S
KIN-ELECTRODE INTERFACE
The concept of “electrode” is rooted in the study of electro-
chemical cells where electrical transport is governed by oxida-
tion and reduction reactions taking place at the interface be-
tween a metal and an electrolyte. A conventional wet-contact
electrode fits this description, since the metal conductor of the
electrode is bathed in an electrolyte gel or solution that buffers
the electrolytic composition through the outer and inner layers
of the skin. Therefore, a wet-contact electrode is well character-
ized by a half-cell potential, a double layer capacitance, and par-
allel and series resistances as shown in Fig. 1. For a dry-contact
or noncontact electrode, however, the interface is more complex
and other processes enter the electrical interactions in skin-elec-
trode coupling. The performance of the electrode is critical, es-
pecially given the small signal amplitude of ECG (1 mV) and
EEG signals (10–100
V).
In general, the coupling between skin and electrode can be de-
scribed as a layered conductive and capacitive structure, with se-
ries combinations of parallel RC elements. The type of electrode
and skin coupling results in several such structures, as shown in
Fig. 1, with different conductance and capacitance values. For
each of these electrode types, typically one of the RC sections
dominates and the electrical coupling may be represented as a
single element with conductance
in parallel with capacitance
, or a simplified coupling admittance .
It is important to realize that both conductance and capaci-
tance are important in characterizing electrode performance. In
what follows we will show that the conventional notion that
low resistance (high conductance) is essential for good elec-
trode performance could be misleading, and that maximizing
resistance (minimizing conductance) in electrode-skin coupling
is actually beneficial in certain important limiting cases. This
unconventional and seemingly counter-intuitive observation de-
rives from simple circuit theory validated by experimental data,
which we offer here for the benefit of the reader who may have
missed this important point from previous literature coverage on
electrode interfaces. Thereby, we hope to rectify misunderstand-
ings in the role of coupling conductance on noise performance
and sensitivity to guide better and more informed decisions in
the design of the electrode and the skin coupling medium.
A. Electrical Model
To accurately model the effect of the skin-electrode coupling
admittance
on the quality and robustness of the received
108 IEEE REVIEWS IN BIOMEDICAL ENGINEERING, VOL. 3, 2010
Fig. 2. (Left) Simplified topology and circuit model of a general, actively shielded biopotential amplifier [11]. Active shield guards high-impedanc
e input from
interference by other sources and implies capacitive coupling between source and amplifier output. (Right) Simple implementation for dry active ele
ctrode made
from standard PCB [14]. Exposed metal on bottom surface contacts skin. The electrode can also work as a noncontact through insulation such as cotton. M
ore
complex designs can be found in [11]–[13].
Fig. 3. Dry/noncontact amplifier circuit noise model along (a) with a simplified plot of frequency behavior of (b) various noise sources. (c) For each RC layer,
noise contribution can be decreased by either drastically increasing resistance towards infinity, increasing capacitance, or reducing the resistance towards zero.
signal, it is necessary to account for the electrical coupling be-
tween the skin and the amplifier connected to the electrode to
acquire the signal. We consider the general, actively shielded
amplifier topology shown in Fig. 2 (left), chosen for its relative
immunity to interference from other sources and line noise [3].
This topology conforms to many of the published amplifier cir-
cuits for dry-contact and noncontact electrodes, e.g., [11]–[13].
A particularly simple low-power and compact realization, which
is used in the experimental data presented in this survey, is il-
lustrated in Fig. 2 (right).
We define the following electrical signals and parameters in
reference to the circuit topology in Fig. 2 (left) and its noise
model in Fig. 3 (left):
signal source on skin surface;
signal recorded at amplifier output;
input referred amplifier voltage noise;
net current noise at amplifier input;
, skin-electrode coupling admittance;
, amplifier input admittance;
active shield to electrode capacitance;
amplifier voltage gain.
As shown in the Appendix, the resulting received output signal
can be written as
(1)
with a source-to-output signal voltage gain
(2)
and source input-referred voltage noise
(3)
CHI et al.: DRY-CONTACT AND NONCONTACT BIOPOTENTIAL ELECTRODES: METHODOLOGICAL REVIEW 109
Fig. 4. Measured noise spectrum of various electrode types, placed at close proximity on forearm at rest, along with predicted (dotted lines) thermal noise
limits (6) from measured skin-electrode coupling impedance data. (Top) The instrumentation noise floor of the amplifier (Fig. 2) is also shown for reference.
(Bottom) Time-domain noise plots are also shown.
These expressions give a quantitative means to analyze the
noise performance as well as the motion and friction sensitivity
of various electrode topologies in terms of physical and elec-
trical circuit parameters, presented in the following.
B. Noise
The source input-referred noise power density follows di-
rectly from (3) where
and represent the power
(RMS squared) of the two input noise generators,
and
(4)
(5)
The relative contributions of the two noise components are
illustrated in Fig. 3. The first noise component, proportional to
the amplifier voltage input noise
, is scaled by a factor
inversely proportional to the electrode coupling efficiency. For
low-impedance contact sensors, this voltage noise component
reduces to the amplifier noise floor, while for high-impedance
contact sensors such as noncontact geometries, the amplifier
voltage noise floor is amplified by a factor
.
This noise amplification could be reduced by minimizing the
active shield capacitance as well as amplifier input capacitance.
However, as shown in Fig. 3, this first noise contribution does
not typically dominate at frequencies of interest, except for
noncontact electrodes at large distance with poor electrode
coupling. The second, and typically more significant noise
component, is proportional to the net current noise
into the coupling impedance. This net current noise combines
thermal noise contributed from the skin-electrode coupling
conductance
and amplifier input conductance , besides
amplifier input current noise
. This noise component is
fundamental to the skin-electrode coupling interface which
typically dominates contributions from the amplifier. In the
limit of a perfect noiseless, infinite input impedance amplifier,
the source input-referred noise power density (5) reduces to
(6)
Paradoxically, (6) shows that fundamentally the source
input-referred noise can be reduced to zero in two limits of
particular interest: either infinite coupling conductance (low-re-
sistance contact sensing), or infinite coupling impedance
(capacitive noncontact sensing). This presents a rather inter-
esting dichotomy—either of the two extreme cases of zero
resistance and infinite resistance of skin-electrode contact are
actually optimal for low-noise signal reception.
110 IEEE REVIEWS IN BIOMEDICAL ENGINEERING, VOL. 3, 2010
Fig. 5. ECG samples taken from various dry-contact and noncontact test electrodes (metal contact, thin film insulation, cotton noncontact), plotted against signal
taken simultaneously from wet Ag/AgCl electrode. Data is shown from .7 to 100 Hz bandwidth without 50/60 Hz notch. Increased noise floor of plastic and cotton
electrodes are not readily visible at ECG scales. Signal distortion can be seen on R-wave for cotton electrode due to increased source impedance.
TABLE I
M
EASURED ELECTRODE IMPEDANCES
Measured data on noise obtained from the differential signal
between two closely spaced electrodes on the forearm at rest are
given in Fig. 4, showing general agreement with the noise model
(6) with measured values of coupling resistance and capacitance
(Table I). As expected, the instrumentation noise floor of the am-
plifier (Fig. 2) is dominated by the measured data, confirming
that the conditions for the limit model (6) are satisfied. Interest-
ingly, the only electrode type with consistently higher observed
noise than the predicted thermal noise from the skin-electrode
coupling noise model are the wet-contact Ag/AgCl electrodes
at lower frequencies. Elevated
-like low-frequency drifts of
the Ag/AgCl offset (half potential mismatch) voltage were con-
firmed in extended (1-hour) recordings and are consistent with
observations in Huigen et al. [4].
One interesting result from this experiment is that for “ca-
pacitive” noncontact electrodes operating through clothing [14],
[15] , the noise performance and electrode coupling is actually
dominated by the resistive component of the cotton layer rather
than a capacitance. In many cases, dry contact electrodes are
much more capacitively dominated than noncontact electrodes
through clothing. Although difficult to imagine, cotton actually
acts as a poorly conductive electrode (
200 ), and is espe-
cially harmful for biopotential measurements. The impedance
of cotton is such that the coupling is mostly resistive in the fre-
quencies of interest, and amounts to adding a large and noisy
series resistor in the signal path. Had the resistance been higher
(i.e., very dry), or the shirt been thinner (increased capacitance),
the noise floor would have been lower. However, the increased
noise did not prevent some acceptable ECG measurements.
Sample ECG data recorded from the same system with
metal-plate electrodes mounted on the chest is shown in Fig. 5,
showing reasonably accurate correspondence between the
dry-contact as well as noncontact electrodes against a wet
Ag/AgCl electrode reference, even for electrodes placed over
a shirt. The capability to continuously record ECG without
direct skin contact opens the door to long-term clinical home
diagnosis and care applications (Section IV).
C. Motion and Friction
Relative motion of electrodes with respect to the body, as well
as friction of electrodes against the body surface, give rise to
artifacts in the received signals that are one of the main imped-
iments against the acceptance of dry-electrode and noncontact
biopotential sensors in mobile clinical settings. These artifacts,
![Fig. 9. Noncontact EEG headband and data from both frontal and occipital electrodes [14].](/figures/fig-9-noncontact-eeg-headband-and-data-from-both-frontal-and-1drwxfsy.webp)
![Fig. 6. A 10-s comparison of noise and drift from wet Ag/AgCl (red trace) versus noncontact electrodes (black trace) during various activities, inducing motion and friction. Noncontact electrodes were fixed in a tight wireless chest band on top of a cotton shirt [14].](/figures/fig-6-a-10-s-comparison-of-noise-and-drift-from-wet-ag-agcl-2wphkfju.webp)
![Fig. 8. Dry and noncontact electrode systems. ECG: (a) chest harness [41], (b) polar heart strap, (c) noncontact vest [14], (d) chair [42], [43], (e) wireless band-aid [20] and (f) dry chest strap [15]. EEG: (g) Neurosky single channel headset, (h) dry MEMs cap [44], (i) fingered dry EEG harness [15], (j) dry/noncontact EEG headband [14], (k) dry active electrode [45] and (l) ENOBIO wireless dry sensor.](/figures/fig-8-dry-and-noncontact-electrode-systems-ecg-a-chest-1w4t1ngj.webp)