Note: Descriptions are shown in the official language in which they were submitted.
CA 02619000 2007-12-28
BioFET Based Microfluidic System
FIELD OF THE INVENTION
[001] This invention pertains to integrated bio-sensor systems and in
particular to those based on label-free detection of oligonucleotides or
protein fragments in a multiplexed microarray format, namely an
electronic DNA microarray chip in an industry standard silicon CMOS
technology.
BACKGROUND TO THE INVENTION
[002] Food and water-borne pathogens constitute a major public health
problem globally. Therefore, early detection and containment of
pathogens, such as Salmonella, E-coli and Campylobacter Jejuni, a
bacterium responsible for gastroenteritis, will prevent the rapid spread of
this communicable diseases and the otherwise inevitable contamination of
other consumable goods. Fast, reliable, sensitive and inexpensive methods
of detecting biological contaminants are therefore of great importance and
appreciated by many areas of research, including but not limited to, water
and food monitoring for public health.
[003] The use of mainstream electronics is an appealing technique for
DNA hybridization detection. Metal-oxide-semiconductor field-effect
transistors (MOSFETs) can be modified to act as DNA hybridization
biosensors. By removing the gate metal and polysilicon off a FET device,
and functionalizing the underlying dielectric with a tether molecule,
ssDNA probes can be immobilized onto the dielectric surface. If target
DNA molecules hybridize to these probes, the underlying FET structure
will be affected by this increase of DNA density, due to the intrinsic
negative charges on the DNA backbone. The excess charge induces a
counter-charge from the surroundings, including the underlying silicon
1
CA 02619000 2007-12-28
substrate. Electronically, the result of this action is as a shift in the
threshold voltage of the biosensor. Such a modified transistor is referred
to herein as a biological field-effect transistor (bioFET).
[004] There are many advantages of this technique. A direct method of
label-free electrical detection of DNA hybridization is provided using
relatively inexpensive, mainstream complementary metal-oxide-
semiconductor (CMOS) silicon technology. As such, precise and
controlled silicon-based fabrication environments are carried over to the
biosensor. The biosensor array is integrated with the accompanying
electronics for signal processing on the same chip. In addition, the
continued scaling down of CMOS technology allows DNA microarrays
with higher sensitivity, and higher density, to be realized. CMOS
technology can also be used to manufacture low-power components.
Combined with dense integration capabilities, DNA microarrays built in
this manner are good candidates for a potentially portable DNA analysis
and pathogen detection device.
[005] A bioFET consists of a field effect transistor, the current through
which can be modulated by the gate potential. The gate provides the
interface between the electronic and biological domains. A solution of
biomolecules, such as single stranded DNA, RNA and proteins ligands, is
put in contact with the bioFET and are covalently fixed on a gate
electrode. The association of complimentary DNA, RNA or protein from
the sample solution increases the charge and hence the potential of the
gate electrode. This is reflected as a change in the current and sensing of
the association. This type of detector is widely used for sensing ions,
organic compounds both in the gaseous and liquid phase. However the
sensitivity of the device is based solely on the non-linear I-V
characteristics of the transistor.
2
CA 02619000 2007-12-28
[006] There remains a need for DNA microsensors that operate with very
low power, and can be incorporated into handheld devices for direct
detection of pathogens from environmental samples. A system using
label-free electrical detection would eliminate the need for expensive and
cumbersome laser scanning systems required in current microarray
technology and would lead to advantages in terms of portability and cost,
making it more amenable to deployment in the field.
SUMMARY OF THE INVENTION
[007] An object of the present invention is to provide a biological sensor
system which is cost-effective, easy to operate and maintain, and which
employs highly sensitive means for detecting biological molecules in
water or food.
[008] Another object of the present invention is to provide a circuit
interface system to the bioFET which produces linear, temperature-
insensitive, and environment insensitive, biosensors that are highly
sensitive to DNA charges.
[009] Another object of the present invention is to provide a model and
interface of CMOS-based electronic biosensors to develop highly sensitive
label-free DNA microarrays compatible with mainstream silicon-based
microelectronics.
[0010] Another object of the present invention is to provide different
interface circuit topologies to enhance the sensitivity and performance of
the bioFET. Such circuit techniques are utilized to enhance the sensitivity
of the bioFET sensor, and are suitable for implementation in DNA
microarrays. The use of feedback systems, both positive and negative, as
3
CA 02619000 2007-12-28
well as non-conventional techniques, such as floating nodes and clocked
circuits, is provided.
[0011] Another object of the present invention is to provide a design of a
biochip DNA array in an industry standard silicon CMOS technology.
BRIEF DESCRIPTION OF THE DRAWINGS
[0012] Fig. 1 is a configuration of a CMOS inverter, biased as an inverting
amplifier.
[0013] Fig. 2 is a configuration of a CMOS inverter as an inverting
amplifier with active, diode connected load.
[0014] Fig. 3 is a configuration of a CMOS inverter as an inverting
amplifier with active, independently biased load.
[0015] Fig. 4 is a configuration of a CMOS common-source amplifier
with negative resistance load.
[0016] Fig. 5 is a configuration of a CMOS ring oscillator with bioFET as
an inverter stage.
[0017] Fig. 6 is a configuration of a feedback operational amplifier with
offset voltage due to hybridization.
[0018] Fig. 7 is a schematic representation of a charge sensing system,
including an embodiment of a bioFET of the present invention.
[0019] Fig. 8 shows schematic diagrams of circuits implemented on the
microarray chip of the present invention.
[0020] Fig. 9 shows the design layout of the microarray chip of the
present invention.
[0021] Fig. 10 depicts a schematic diagram of a folded cascode
operational amplifier of the present invention.
[0022] Fig. 11 shows the a) Frequency response of the operational
amplifier, and b) Voltage transfer characteristics of the designed
operational amplifier of the present invention.
4
CA 02619000 2007-12-28
[0023] Fig. 12 depicts the schematic diagram of Fig. 10 with the addition
of compensation capacitor.
[0024] Fig. 13 shows a compact layout of an operational amplifier of the
present invention.
DETAILED DESCRIPTION OF THE INVENTION
[0025] BioFETs are based on the principle that adsorption of charged
biomolecules, such as DNA fragments (oligonucleotides), on the gate
dielectric of a FET will modulate the transistor's surface potential and
ultimately its current. If the dielectric is appropriately functionalized with
an oligonucleotide of a given sequence, hybridization of an
oligonucleotide with the complementary sequence will give rise to an
electrical signal in the FET. The fabrication of bioFETs requires the
controlled deposition of high quality, high dielectric constant insulators of
nano-scale thickness and methods for the chemical ftinctionalization of
these ultrathin dielectric layers for biomolecule attachment.
[0026] DNA hybridization causes a shift in the threshold voltage of the
bioFET. This shift is equivalent to a change in the voltage of the reference
electrode. For constant biasing, the change in the threshold voltage causes
a change in the amount of drain current, via the transconductance of the
bioFET. In addition, the threshold change in the bioFET causes a shift in
its electrical characteristics such as the C-V characteristics or the I-V
characteristics.
[0027] The following characteristics of the bioFET sensor allow a
determination of three different sensor circuit families:
[0028] Potentiometric sensors: These sensors aim at sensing the change in
the threshold voltage by maintaining a constant current for the drain of the
5
CA 02619000 2007-12-28
BioFET, while measuring the shift in the gate-source voltage (VGs)
directly via a high input impedance readout circuit, such as an operational
amplifier. These sensors can offer better linearity in measurements, as the
output can be fed to linear amplifiers for sensitivity enhancement. Another
characteristic of this type of sensor is that the sensitivity is limited by
the
limits of validity of the one dimensional solution, and not by the device
sizing.
[0029] Amperometric sensors: In this regime, the change in the current of
the bioFET due to DNA hybridization is sensed while VGs is kept
constant. If the bioFET is in the linear region of operation, the sensor
works as a resistance sensor, monitoring the change in the channel
resistance as the DNA is immobilized. In saturation, amperometric sensors
would be equivalent to common-source amplifiers, offering a high degree
of sensitivity to hybridization. Higher aspect ratios of the bioFET will, in
general, cause higher sensitivities due to the higher transconductance
values. However, the output resistance can also be lowered by this, and a
compromise must be made to maintain a high sensitivity value.
[0030] Capacitance sensors: The change in the small signal capacitance
due to a shift in the threshold voltage can affect the operation of different
AC circuits, particularly amplifiers and oscillators. DNA hybridization
can, therefore, change the oscillation frequency of an oscillator or the gain
of an amplifier. The sensitivity of the capacitance due to the DNA
hybridization scales as the area of the bioFET, which may be challenging
in terms of miniaturization.
[0031] Many different interface circuits have been proposed for ion
sensitive field effect transistors, including both potentiometric and
amperometric sensors, which are equally applicable for the bioFETs.
These circuits are concerned with interfacing to the transistorm, as
6
CA 02619000 2007-12-28
opposed to the actual sensitivity of the device. The more sensitive a
circuit is, the more possible it is to reduce the density of DNA probes,
thereby allowing the sensor to detect very dilute concentrations of the
DNA molecules.
[0032] The configurations of bioFETs shown in Figs. 1-6 are
implemented to fabricate hybrid devices employing functionalized carbon
nanotubes or nanowires on the surface of the gate oxide to increase the
effective surface area for hybridization and signal generation. The
resulting biosensor can be integrated within a commercial CMOS
technology, leading to low-cost devices integrated with on-chip amplifiers
and signal conditioning/ processing electronics, further enhancing the
overall sensitivity of the bioFET system.
[0033] The bioFET configuration of the invention comprises several basic
bioFET units combined in networks to enhance the sensitivity of the
biomolecule association.
1. CMOS inverter, biased as an inverting amplifier
[0034] A set of bioFETs (PMOS and NMOS) are configured as an
inverter and share a reference electrode. The inverter is biased at the point
where both transistors are saturated for maximum gain. Hybridization on
both transistors results in the output voltage changing by a large amount.
[0035] As shown in Fig. 1, a set of BioFETs (PMOS and NMOS) are
configured as an inverter, sharing the reference electrode. The inverter
can then be biased at the point where both transistors are saturated for
maximum gain. Hybridization on both transistors results in the output
voltage changing by a significant amount.
7
CA 02619000 2007-12-28
[0036] The amplification in this circuit is dependant on the
transconductances of the transistors, the output conductance, and the
amount of threshold voltage change due to DNA hybridization. Assuming
that AVill is constant, which is the case for specific parameters of the
device (electrolyte concentration, pH, probe density), then only the
transconductances and output conductances will affect the amplifier's gain.
Changing the transconductance by changing the device size or changing
the DC bias current will not help much in increasing the gain, because the
output resistance will decrease as well.
=21R-17)
rout = (XID )-1
[0037] One method that would enhance the gain in this stage would be to
operate the inverter in subthreshold, since such a high output swing is not
needed, and since subthreshold bias would give extremely high output
resistances. A simple simulation has shown that gains of around 100 (40
dB) are obtainable.
[0038] The output of such an amplifier would have to be taken from a
low-output impedance buffer. A unity gain opamp can be used for this
purpose.
2. CMOS inverter configured as an inverting amplifier with active, diode
connected to load
[0039] This design provides the same performance as the previous one but
reduces the area of hybridization from two transistors to one.
8
CA 02619000 2007-12-28
[0040] In an attempt to reduce the bio-sensitive area, one might use a
simple amplifier configuration with a diode connected load. This can also
be seen as a MOSFET inverter biased to operate as an inverting amplifier.
Fig. 2 shows this configuration.
[0041] This can be a very small gain, because g1 is generally much
smaller than the output resistances of the two transistors and will therefore
dominate, reducing the overall gain of the circuit. Thus, this circuit shares
the advantages of the previous one.
3. CMOS inverter configured as an inverting amplifier with active
independently biased load
[0042] In this configuration, shown in Fig. 3, only one of the transistors is
a bioFET which can undergo hybridization while the other transistor is
biased at a constant point. This configuration allows a wider choice of
biasing points for the bioFET, which is somewhat independent of the
power supply, but lower gain than the first configuration.
[0043] The gain in this case can be much higher than that in the second
configuration, but not as high as the first configuration, described above.
4. CMOS common-source amplifier with negative resistance load
[0044] In this configuration, the gain of the amplifier is made high over a
voltage range by using negative resistance loads, such as resonant
tunneling diodes or other MOSFET circuits.
[0045] The gain of a single-stage bioFET amplifier can be increased using
negative-resistance loads. Using one of the several negative-resistance
techniques (resonant tunneling diodes or MOSFET circuits for negative
9
CA 02619000 2007-12-28
resistance), the gain of the amplifier can be made very high (ideally
infinite) over a voltage range. If the transistor is biased around that range,
then the sensitivity to the bioFET is superb. Fig. 4 depicts a schematic
diagram of one implementation.
[0046] In this configuration, extremely high gains can be obtained through
optimization of the component values and biasing conditions.
[0047] Similar to the previously described circuits, this circuit would also
require an output buffer op amp to read the voltage. This circuit uses only a
single bioFET and may achieve gains higher than that of the first
configuration.
5. CMOS ring oscillator with bioFET as an inverter stage
[0048] In this configuration, all the inverters of a ring oscillator are
replaced by bioFET inverters. The hybridization of the biomolecule
causes a capacitance change which affects the oscillation frequency.
[0049] In order to establish the effect of DNA hybridization on the ring
oscillator's characteristics, it is important to find the relationship between
both the threshold voltage and the input capacitance, and frequency of
oscillation.
[0050] A ring oscillator can be seen as a regular phase shift oscillator, but
with loop gain that is higher than 1, such that the circuit oscillates between
the rail voltages. One can think of the ring oscillator as starting at the
equilibrium point, and then starting to oscillate due to some disturbance.
Under small-signal conditions, the Barkhausen criteria for oscillation can
be evaluated. This will give the frequency of oscillation of the ring
oscillator. Of course, this will only apply under small-signal conditions.
CA 02619000 2007-12-28
This is why the best sensitivity of the ring oscillator to hybridization might
be under small-signal oscillation. This can be done with heavy enough
capacitive loads. Fig. 5 shows a schematic diagram of a small-signal ring
oscillator's simplified equivalent circuit.
[0051] In Fig. 5, the resistance R is the parallel combination of the
saturated output resistances of the pullup PMOS and pulldown NMOS,
and the capacitance C is the parallel combination of the gate capacitances
for the PMOS and NMOS of the next stage. Without these delay elements,
the circuit would not oscillate, but would rather stabilize at an equilibrium
voltage where the input and output of the inverters has the same voltage
(called the inverter's threshold voltage).
[0052] From determining the mathematical condition, it can be guaranteed
that the ring oscillator will oscillate.
[0053] By replacing the standard inverters with bioFET inverters, the
oscillation frequency will be affected by the capacitance change due to
hybridization, and not by the threshold voltage shift. The threshold voltage
shift will merely cause the DC average stable point around the ring to
deviate. Biasing the bioFETs around its threshold voltage results in a
change in the inverter's oscillation properties due to the capacitance
changes upon hybridization.
[0054] Subthreshold operation will allow higher gains of the inverter
stage, allowing more frequency deviation per change in capacitance due to
DNA hybridization.
11
CA 02619000 2007-12-28
6. CMOS ring oscillator with bioFET as a load capacitor
,
[0055] Another manifestation of the first configuration is where a
MOSFET inverter stage is replaced by a bioFET inverter. The expression
for the sensitivity is again:
do) ¨C 1+G3 _______________
11
dCi R 1 (2 -v-,ri1) \3
'4 -'
[0056] Another manifestation is to have only one of the inverter
MOSFETs functionalized as a bioFET. This way, upon hybridization, the
gain of the inverting amplifier stage will change, and this will cause a
change in the frequency of the output. In this configuration, if the
oscillator is operated in its oscillation mode, then hybridization will
change it oscillation properties:
7. General LC oscillator with bioFET as a load capacitor
[0057] In this configuration, tunable oscillators such as RF VCO, Colpitts,
Hartley or Clapp oscillators are used with the capacitance obtained from
the bioFET. Hybridization changes the capacitance and hence the
oscillation frequency of the device. Similar analysis as that of the ring
oscillator can be carried out to find the frequency deviation.
8. Mismatched analog circuits
[0058] The effects of mismatch can be studied in particular circuits. In this
configuration, an operational amplifier in open loop is provided a stable
input and another input from the bioFET. The inputs are matched in the
unhybridized state. Upon hybridization the inputs become mismatched and
the mismatch is amplified in the output signal by the amplifier gain.
12
CA 02619000 2007-12-28
[0059] It seems that operational amplifiers might have the best sensitivity
to mismatch, even under feedback. An entire analysis of opamp-based
mismatch can be carried out, depending on the opamp's topology.
Mismatch would introduce an output offset voltage, whose value would be
dependant on the open-loop gain of the amplifier. Under closed loop unity
feedback, the change in the output voltage would be:
dVout = ¨A
dVoffset 1 A
[0060] This value is less than 1. See Fig. 6. Thus, under closed-loop, a
feedback opamp might not provide that much gain. However, one can
utilize the opamp in open-loop. Then, the gain due to mismatch will be
that of the open-loop gain A.
9. Leakage component modulation for a floating charge device (eg.
Dynamic logic circuit)
[0061] This configuration utilizes the effects of hybridization in the
leakage current of MOSFET devices. In this
configuration, the
modulation of the leakage current by the hybridization is used as a signal.
When a charge is stored in a floating node, substrate leakage currents lead
to its depletion over time. Hybridization of the biomolecule causes further
inversion of a channel and an increase of leakage current and hence
discharge in a shorter time. This leakage is used as an indication of
hybridization.
[0062] This is because the inversion might not cause conduction along the
channel, but rather increase the area of the effective PN-junction and thus
increase the leakage current. This is be very noticeable for floating
charges.
13
CA 02619000 2007-12-28
[0063] These configurations lead to significant enhancement in the
sensitivity of the device. An individual configuration is what is termed as
a BioPixel. The device itself consists of an array of BioPixels. Each pixel
is configured by the biomolecule present in its sensing layer to identify its
complementary target associated with a particular disease biomarker.
[0064] One example of a bioFET in a 1 milliMolar solution shows
electrical current changes of 20 nA for hybridization of around 4,080
molecules of 17 base pairs on a bioFET with 12,000 square micron gate
area fabricated in an industry standard silicon complementary metal-
oxide-semiconductor (CMOS). Experiments with the bioFET sensor are
with purified and amplified nucleic acids in order to demonstrate
functionality, determine sensitivity limits and to calibrate the existing
theoretical model. In practical applications, the bioFET sensor is
packaged with microfluidic pre-processing modules. These preprocessing
modules perform a number of steps to transform the analyte from its
naturally occurring state to a form that is suitable for detection by the
bioFET sensor.
[0065] An example of a microfluidic/MEMS based system that performs
all the required preprocessing steps including filtration, concentration, cell
lysing, DNA extraction, processing and amplification is represented
schematically at Fig. 7.
[0066] This system combines the physical and chemical modeling of the
electrolyte/DNA system with semiconductor technology to establish a
physical model that relates the change in current due to DNA
hybridization to the density of DNA probes. This measure gives
quantitative characterization of the sensitivity of the bioFET, which is
used to compare bioFET-based sensing to other mainstream methods, such
as optical DNA microarrays.
14
CA 02619000 2007-12-28
[0067] The linearization of the entire physical model of the bioFET gives
good approximations to the response of the bioFET to DNA hybridization.
Moreover, the simplification shows that the response to DNA
hybridization can be quantified as a shift in the threshold potential, the
amount of which can be given in closed form, as a function of the ionic
strength, DNA strand length, DNA probe density, degree of hybridization,
adsorption affinity of the insulator surface, and permittivity of the
electrolyte and membrane areas.
[0068] Since these biosensors require that the sample presented to them is
pure, an integrated microfluidic sample purification system is essential for
the device's operation. Microfluidic chips are attractive for processing
field samples because many of the sample processing steps can be
integrated into a single module, minimizing the dilution of intracellular
components following cell lysis. Furthermore, sample handling in
microfluidics is gentle which minimizes DNA shearing effects, which are
important for high molecular weight DNA fragments. The pre-processing
module performs a number of essential steps including sample filtration,
cell lysing, biomolecule extraction and concentration.
[0069] Sample filtration is critical because particulate matter present in
the environment pose a serious challenge to the functioning of
microfluidic system due to channel blockage and also provide sites for
accumulation of other particulates. Selective filtration and concentration
of the biological material of interest can be done using size-based
retention of cells on microfabricated filters. Microfabricated porous
monoliths with defined pore sizes allow construction of a series of filters
which are used to isolate biological materials like cells, which are in a
characteristic size range, from other particulate matter. This methodology
is further used to isolate sub-populations in cells. For example white blood
cells can be isolated from red blood cells using this method due to their
CA 02619000 2007-12-28
size difference. However, microfabricated filters has small capacity and
are prone to become clogged. Active filter regeneration is achieved by
generating counter flow using an electrokinetic mechanism. In
electrokinetics, electrical fields are used to direct flows through
microchannels and microporous materials such as filters. Since electrical
fields are easily controlled, modulated and reversed using computer
generated signals, they could be used for dynamic cleaning of a clogged
filter. The active filter consists of a microporous membrane of a defined
pore size with a pair of electrodes on either side of the membrane.
Application of a positive potential across the membrane drives flow
through the membrane and leads to accumulation of particulate matter.
Subsequently, reversal of potential across the membrane generates counter
flow, clearing the filter. The concentrated particulate matter is moved to a
waste reservoir and the filtering process continued. This process
regenerates the filter and allows higher capacity. A series of stacked filter
designs can be developed to size-separate the particles for further analysis.
[0070] The same design can be used for concentration of the collected
biological material. The collected material (cells) consist of the
information containing biomolecule packaged along with other
biomaterials such as DNA, RNA, proteins, lipids and cellular materials.
The cell membrane is broken and its contents exposed for the biomolecule
to be extracted and analyzed. Cell lysis is performed using osmotic shock
on exposure to pure water. In this method, a porous material inside a
microchannel is used to concentrate the cells of interest. Subsequently, the
buffer solution is replaced with distilled water using the microfluidic
solvent exchange. Cells present in DI water undergo osmotic shock due to
the large concentration gradient present across their cell membrane.
Subsequently, the solvent is once again replaced by buffer solution.
16
CA 02619000 2007-12-28
[0071] Since the cell lysate consists of the biomolecule of interest along
with other contaminants, a microscale solid phase extraction is performed
to concentrate and isolate the DNA. This is especially important when a
subsequent PCR stage is included as PCR is inhibited by certain proteins.
The cell lysate is electrokinetically passed through another silica sol-gel
porous region microfabricated on chip. Silica surfaces have preferential
adsorption of DNA over other proteins, lipids and other containments. The
concentrator is subsequently washed with a propanol/water mixture to
further purify the extracted DNA. Release of concentrated DNA is
performed using the low ionic concentration solution or DI water.
Amplification of the section of DNA of interest from the extracted DNA is
performed using standard PCR protocol subsequent to which the DNA is
passed on to the detector stage.
Design of a DNA Microarray Chip in CMOS 0.8 pm Technology
[0072] The chip was designed using an industry standard silicon CMOS
technology which also supports microelectromechanical systems
(MEMS)/microfluidic device control and high-voltage integrated circuit
(HVIC) control and power systems.
[0073] All circuit simulations were conducted using simulation software.
All layouts were built using a custom design platform. The models for the
transistors and the design layers were included in the design kit and
technology file for the industry-standard silicon process. The transistor
models were based on a standard industry modeling architecture. The only
components to be simulated were the operational amplifiers. The BioFET
transistors were simulated once with a gate contact to verify normal
transistor operation and proper construction, but the gate contact was
subsequently removed to facilitate direct exposure of the polysilicon,
which would be etched during post-processing.
17
CA 02619000 2007-12-28
[0074] Several different microarrays were required in the construction.
For example, for each microarray cell, the bioFET transistor was expected
to occupy a square area of 50 mx50 gm. This size is needed to
accommodate the array spotter requirements. The spotter resolution
limitations also required that the spacing between the cells be set at 200
gm, edge to edge. Such spacing allows a lot of unused space between the
cells that is sufficient to incorporate any auxiliary circuits. Therefore, all
operational amplifiers and differential circuits were placed within these
spaces. Several different microarrays were implemented in the design.
Each of these microarrays was required to be electrically isolated from all
others. Examples of the arrays constructed were the following:
1. A 7x7 Array of 50 mx50 pm P-type bioFETs, with separate drain
connections but shared source and body connections.
2. A 7x7 Array of 50 gmx50 gm N-type bioFETs, with separate drain
connections but shared source and body connections.
3. A 7x7 Array of 50 gmx50 gm P-type bioFETs, with separate drain
connections but shared source and body connections. Each bioFET
was connected in a common-source fashion to a high-impedance
active load. The output was read using an operational amplifier
configured as a unity gain buffer.
4. A 7x7 Array of 50 ptmxi gmx21 P-type multi-fingered bioFETs, with
separate drain connections but shared source and body connections.
Each bioFET was connected in a common-source fashion to a high-
impedance active load. The output is read using an operational
amplifier configured as a unity gain buffer.
5. A 14x7 array, consisting of 7x7 pairs of bioFETs. Each bioFET was
connected in common source mode, and the outputs of each pair were
passed to differential amplifiers. The output of the differential was
read by a unity gain buffer. Only one of the two bioFETs would
18
CA 02619000 2007-12-28
undergo hybridization. This array is designed to reject any signal
fluctuations that are common to both bioFET devices, such as those
created by temperature and pH variations.
[0075] Fig. 8 shows schematic diagrams of the circuits that were
implemented in these arrays. In addition, a large square area of
2.5mmx2.5mm was required for monitoring the etch process during post-
processing. This region consisted of bare polysilicon deposited on top of
the substrate, with the insulator in between. A connection was provided to
the periphery of this area to drain away any charges that might leak into
the substrate during the etch process. Etch monitoring is important to
ensure that the insulator underneath the polysilicon is not damaged during
the etching process.
[0076] The layout of the final chip is shown in Fig. 9. The chip has a size
of 1Ommx10mm, and is surrounded by 260 pads, in addition to two pads
that are placed closer to the center of the chip. These two pads are
introduced to serve as reference electrodes when the electrolyte is applied
to the surface of the chip. The number of pads was not sufficient for all
different arrays and some of them had to be shared by two different arrays.
Specifically, the drain connections of each of the bioFETs in arrays 1 and
2 share a single pad.
[0077] To facilitate post processing, several alignment marks were placed
on the chip. These consisted of squares of the top metal layer, located in
different areas around the chip. These marks provided landmarks that were
used for aligning the etch mask with the chip during post-processing.
Finally, it is important that there be clearance spacing between the
outermost pads and the sliced edge of the chip. A clearance of 0.75mm
was required to guard against edge bead formation during the deposition
of PDMS onto the chip during post-processing.
19
CA 02619000 2007-12-28
[0078] The main circuit component that required proper design in this
chip was the operational amplifier. Many different topologies for CMOS
operational amplifiers exist. The topology chosen was that of a folded
cascode op-amp. This topology allowed for higher gains to be achieved
from a single stage. The body effect in the cascode transistor enhances the
gain performance of the amplifier, contrary to the case of cascaded
amplifiers. Finally, compensation capacitors in a cascade transistor need
not be as large as that for a regular two stage op-amp. This is because the
capacitance can be placed across two stages, which are the output node
and the cascode stage, as opposed to one stage in the cascaded two stage
amplifier case. One challenge of using this topology was that it allowed
for some common-mode gain, even with perfectly matched differential
transistors and current sources.
[0079] Fig. 10 shows the schematic diagram of the operational amplifier.
Transistors M1 and M2 form the input differential pair, with M3 and M4
as active loads. Transistors M5 and M6 form the cascode pair, and a
feedback connection to the gates of the active load provide the differential
to single-ended conversion. The single ended output was applied to an
output common source stage, which provided the final output pin of the
amplifier. Using a common source output stage instead of a common-drain
mode is justified because the operational amplifier operates in unity
feedback mode. Thus, its output resistance is roughly divided by the open-
loop gain, which will be quite high. The reason for using an amplifying
stage at the output is to obtain a high output dynamic range, which will be
limited only by the onset of linear operation of the output transistors. If
this last stage were a common-drain stage, then the factor that would
determine the dynamic range would be the overdrive voltage of the output
transistor, which would be determined by the gate bias. If high dynamic
range is required, then the gate bias of the output PMOS would have to be
CA 02619000 2007-12-28
quite high, necessitating a higher gate bias for the cascode transistors to
keep them saturated. However, it was desired to have a cascode bias of
zero volts to minimize the extra circuitry required to bias voltage
generation.
[0080] The current driving requirement for the operational amplifier was
chosen at 100 A. Given the large number of op-amps in the chip, higher
output drive capabilities may heat up the chip and cause undesirable
effects to the bioFET sensors. To find the required gate bias voltage of the
output PMOS transistor, a value of 100 A was used as the quiescent drain
current, and set the output voltage to zero. This minimizes the offset
voltage of the op-amp. Recognizing the presence of large-signal variations
at the output, the calculated bias was adjusted to keep the output transistor
in saturation under the harshest signal swings (chosen here as 1 volt).
From hereon, the analysis continued backwards, with the cascode
transistors, active loads, and finally the input differential pair. In each
step,
conditions were imposed to guarantee that the device stayed saturated, the
appropriate voltage values were determined, and the sizes were calculated
based on the currents that flowed in the branches. The total current
consumption in the internal stages of the op-amp was also chosen as 100
A, divided equally among the different conduction paths. Based on these
current values and the DC voltages on the nodes that would guarantee
saturation, the sizing of the different devices was deduced. The current
sources in Fig. 10 were implemented using a stacked transistor pair. This
had the effect of increasing the output resistance and made the circuit
behave more like an ideal current source. The biasing of these transistors
was chosen so as to maintain the saturation region of operation. The
biasing was kept as low as possible so that the current source had a higher
range of operation. This further enhanced the dynamic range of the device.
21
CA 02619000 2007-12-28
[0081] Figs. 11a) and b) show the simulated frequency response and the
voltage transfer characteristics of the designed op-amp, respectively. The
dynamic range spanned around 4 volts, and the low frequency gain was
simulated at 83 dB. The bandwidth for the uncompensated op-amp is
shown to be around 35kHz. However, to guarantee stability of the op-amp
with unity feedback, a Miller compensation capacitor was added at the
terminals shown in Fig. 12. The cascode configuration allowed for the
placement of the capacitor over a large inverting gain, consisting of the
output stage and the cascode stage. This resulted in a smaller capacitance
value and higher stability. The compensated op-amp had a bandwidth of
81(Hz, a gain margin of 7.633 dB, and a phase margin of 35 . Fig. 13
shows a compact layout of the op-amp, in which it is clear that the
compensating capacitor occupies a considerable area of the design. The
following table shows the performance parameters of a sample operational
amplifier of the invention, compared with typical values of general
purpose CMOS op-amps.
Sample Performance Parameters of a Designed Operational Amplifier
Parameter Designed Op-Amp Typical
Open-loop gain 85dB 74dB
CMRR 124dB 80dB
Open-loop bandwidth 35kHz (5kHz with 100Hz-
100kHz
compensation)
Output resistance 7.8k0 3000-10kC2
Slew rate 12.5V/i.ts (9.5 V/ps with 2-20V/ps
10pF load)
Offset voltage -0.58mV 0.1 -1mV
Gain margin 7.63dB
Phase margin 35.3 >60
Power consumption lmW 0.5-10mW
P SRR 74dB 80dB
22
CA 02619000 2015-02-26
[0082] The foregoing has constituted a description of specific
embodiments showing how the invention may be applied and put into use.
These embodiments are only exemplary. The invention is defined in the
claims which now follow.
23
