COLUMN ADC

[0001] This application is a divisional of U.S. patent application Ser. No. 13/174,542 filed Jun. 30, 2011, which claims the benefit of priority to previously filed U.S. provisional patent application Ser. No. 61/360,493 filed Jun. 30, 2010, U.S. provisional application Ser. No. 61/360,495 filed Jul. 1, 2010, U.S. provisional application Ser. No. 61/361,403 filed Jul. 3, 2010, and U.S. provisional application Ser. No. 61/365,327 filed Jul. 17, 2010, the disclosures of all of which are incorporated herein by reference in their entirety.

[0002] Electronic devices and components have found numerous applications in chemistry and biology (more generally, “life sciences”), especially for detection and measurement of various chemical and biological reactions and identification, detection and measurement of various compounds. One such electronic device is referred to as an ion-sensitive field effect transistor, often denoted in the relevant literature as an “ISFET” (or pHFET). ISFETs conventionally have been explored, primarily in the academic and research community, to facilitate measurement of the hydrogen ion concentration of a solution (commonly denoted as “pH”).

[0003] More specifically, an ISFET is an impedance transformation device that operates in a manner similar to that of a Metal Oxide Semiconductor Field Effect Transistor (MOSFET), and is particularly configured to selectively measure ion activity in a solution (e.g., hydrogen ions in the solution are “analytes”). A detailed theory of operation of an ISFET is given in “Thirty years of ISFETOLOGY: what happened in the past 30 years and what may happen in the next 30 years,” P. Bergveld, Sens. Actuators, 88 (2003), pp. 1-20 (“Bergveld”), which publication is hereby incorporated herein by reference in its entirety.

[0004] Details of fabricating an ISFET using a conventional Complementary Metal Oxide Semiconductor (CMOS) process may be found in Rothberg, et al., U.S. Patent Publication No. 2010/0301398, Rothberg, et al., U.S. Patent Publication No. 2010/0282617, and Rothberg et al., U.S. Patent Publication 2009/0026082; these patent publications are collectively referred to as “Rothberg,” and are all incorporated herein by reference in their entireties. In addition to CMOS, bipolar and CMOS (biCMOS) processing may be used, such as a process that would include a p-channel MOS FET array with bipolar structures on the periphery. Alternatively, other technologies may be employed where a sensing element can be made with a three-terminal devices in which a sensed ion leads to the development of a signal that controls one of the three terminals; such technologies may also include, for example, Gallium Arenides (GaAs) and carbon nanotube technologies.

[0005] Taking a CMOS example, a P-type ISFET fabrication is based on a p-type silicon substrate, in which an n-type well forms a “body” of the transistor. Highly-doped p-type (p+) source (S) and drain (D) regions, are formed within the n-type well. A highly-doped n-type (n+) region B may also be formed within the n-type well to provide a conductive body (or “bulk”) connection to the n-type well. An oxide layer may be disposed above the source, drain and body connection regions, through which openings are made to provide electrical connections (via electrical conductors) to these regions. A polysilicon gate may be formed above the oxide layer at a location above a region of the n-type well, between the source and the drain. Because it is disposed between the polysilicon gate and the transistor body (i.e., the n-type well), the oxide layer often is referred to as the “gate oxide.”

[0006] Like a MOSFET, the operation of an ISFET is based on the modulation of charge concentration (and thus channel conductance) caused by a Metal-Oxide-Semiconductor (MOS) capacitance. This capacitance is constituted by a polysilicon gate, a gate oxide and a region of the well (e.g., n-type well) between the source and the drain. When a negative voltage is applied across the gate and source regions, a channel is created at the interface of the region and the gate oxide by depleting this area of electrons. For an n-well, the channel would be a p-channel. In the case of an n-well, the p-channel would extend between the source and the drain, and electric current is conducted through the p-channel when the gate-source potential is negative enough to attract holes from the source into the channel. The gate-source potential at which the channel begins to conduct current is referred to as the transistor's threshold voltage VTH (the transistor conducts when VGS has an absolute value greater than the threshold voltage VTH). The source is so named because it is the source of the charge carriers (holes for a p-channel) that flow through the channel; similarly, the drain is where the charge carriers leave the channel.

[0007] As described in Rothberg, an ISFET may be fabricated with a floating gate structure, formed by coupling a polysilicon gate to multiple metal layers disposed within one or more additional oxide layers disposed above the gate oxide. The floating gate structure is so named because it is electrically isolated from other conductors associated with the ISFET; namely, it is sandwiched between the gate oxide and a passivation layer that is disposed over a metal layer (e.g., top metal layer) of the floating gage.

[0008] As further described in Rothberg, the ISFET passivation layer constitutes an ion-sensitive membrane that gives rise to the ion-sensitivity of the device. The presence of analytes such as ions in an analyte solution (i.e., a solution containing analytes (including ions) of interest or being tested for the presence of analytes of interest), in contact with the passivation layer, particularly in a sensitive area that may lie above the floating gate structure, alters the electrical characteristics of the ISFET so as to modulate a current flowing through the channel between the source and the drain of the ISFET. The passivation layer may comprise any one of a variety of different materials to facilitate sensitivity to particular ions; for example, passivation layers comprising silicon nitride or silicon oxynitride, as well as metal oxides such as silicon, aluminum or tantalum oxides, generally provide sensitivity to hydrogen ion concentration (pH) in an analyte solution, whereas passivation layers comprising polyvinyl chloride containing valinomycin provide sensitivity to potassium ion concentration in an analyte solution. Materials suitable for passivation layers and sensitive to other ions such as sodium, silver, iron, bromine, iodine, calcium, and nitrate, for example, are known, and passivation layers may comprise various materials (e.g., metal oxides, metal nitrides, and metal oxynitrides). Regarding the chemical reactions at the analyte solution/passivation layer interface, the surface of a given material employed for the passivation layer of the ISFET may include chemical groups that may donate protons to or accept protons from the analyte solution, leaving at any given time negatively charged, positively charged, and neutral sites on the surface of the passivation layer at the interface with the analyte solution.

[0009] With respect to ion sensitivity, an electric potential difference, commonly referred to as a “surface potential,” arises at the solid/liquid interface of the passivation layer and the analyte solution as a function of the ion concentration in the sensitive area due to a chemical reaction (e.g., usually involving the dissociation of oxide surface groups by the ions in the analyte solution in proximity to the sensitive area). This surface potential in turn affects the threshold voltage of the ISFET; thus, it is the threshold voltage of the ISFET that varies with changes in ion concentration in the analyte solution in proximity to the sensitive area. As described in Rothberg, since the threshold voltage VTH of the ISFET is sensitive to ion concentration, the source voltage VS provides a signal that is directly related to the ion concentration in the analyte solution in proximity to the sensitive area of the ISFET.

[0010] Arrays of chemically-sensitive FETs (“chemFETs”), or more specifically ISFETs, may be used for monitoring reactions—including, for example, nucleic acid (e.g., DNA) sequencing reactions, based on monitoring analytes present, generated or used during a reaction. More generally, arrays including large arrays of chemFETs may be employed to detect and measure static and/or dynamic amounts or concentrations of a variety of analytes (e.g., hydrogen ions, other ions, non-ionic molecules or compounds, etc.) in a variety of chemical and/or biological processes (e.g., biological or chemical reactions, cell or tissue cultures or monitoring, neural activity, nucleic acid sequencing, etc.) in which valuable information may be obtained based on such analyte measurements. Such chemFET arrays may be employed in methods that detect analytes and/or methods that monitor biological or chemical processes via changes in charge at the chemFET surface. Such use of chemFET (or ISFET) arrays involves detection of analytes in solution and/or detection of change in charge bound to the chemFET surface (e.g. ISFET passivation layer).

[0011] Research concerning ISFET array fabrication is reported in the publications “A large transistor-based sensor array chip for direct extracellular imaging,” M. J. Milgrew, M. O. Riehle, and D. R. S. Cumming, Sensors and Actuators, B: Chemical, 111-112, (2005), pp. 347-353, and “The development of scalable sensor arrays using standard CMOS technology,” M. J. Milgrew, P. A. Hammond, and D. R. S. Cumming, Sensors and Actuators, B: Chemical, 103, (2004), pp. 37-42, which publications are incorporated herein by reference and collectively referred to hereafter as “Milgrew et al.” Descriptions of fabricating and using ChemFET or ISFET arrays for chemical detection, including detection of ions in connection with DNA sequencing, are contained in Rothberg. More specifically, Rothberg describes using a chemFET array (in particular ISFETs) for sequencing a nucleic acid involving incorporation of known nucleotides into a plurality of identical nucleic acids in a reaction chamber in contact with or capacitively coupled to the chemFET, wherein the nucleic acids are bound to a single bead in the reaction chamber, and detecting a signal at the chemFET, wherein detection of the signal indicates release of one or more hydrogen ions resulting from incorporation of the known nucleotide triphosphate into the synthesized nucleic acid.

[0012] However, noise and signal deterioration are big concerns for analog data signals from the chemical detection circuit with a lot of pixels, and that the chemical detection circuit must be placed close to its peripheral devices. Thus, there is a need in the art to provide digital outputs and also a need for on-chip data compression.

[0075] A floating gate (FG) transistor may be used to detect ions in close proximity to the gate electrode. The transistor may be configured with other transistors to form a pixel that can be placed into an array for addressable readout. In the simplest form, the ancillary transistors are used solely to isolate and select the floating gate transistor for readout in an array. The floating gate transistor may be a chemically-sensitive transistor, and more specifically, a chemically-sensitive field effect transistor (ChemFET). The ChemFET may be designed with a metal-oxide-semiconductor field-effect transistor (MOSFET) containing self-aligned source and drain implants fabricated using standard complementary metal-oxidesemi-conductor (CMOS) processing. The ChemFET may be an ion sensitive FET (ISFET), and may be a PMOS or an NMOS device.

[0076] A floating gate (FG) transistor may be used to detect ions in close proximity to the gate electrode. The transistor may be configured with other transistors to form a pixel that can be placed into an array for addressable readout. In the simplest form, the ancillary transistors are used solely to isolate and select the floating gate transistor for readout in an array. The floating gate transistor may be a chemically-sensitive transistor, and more specifically, a chemically-sensitive field effect transistor (ChemFET). The ChemFET may be designed with a metaloxidesemiconductor field-effect transistor (MOSFET) containing self-aligned source and drain implants fabricated using standard complementary metaloxidesemiconductor (CMOS) processing. The ChemFET may be an ion sensitive FET (ISFET), and may be a PMOS or an NMOS device.

[0077] To reduce the pixel size to the smallest dimensions and simplest form of operation, the ancillary transistors may be eliminated to form an ion sensitive field-effect transistor (ISFET) using one transistor. This one-transistor, or 1T, pixel can provide gain by converting the drain current to voltage in the column. Parasitic overlap capacitance between terminals of the transistor limits the gain. The capacitance ratios also allow consistent pixel-to-pixel gain matching and relatively constant current operation which justifies the use of a row selection line which can sink the necessary current without causing unacceptable variation. Derivatives of this allow for increased programmable gain through a cascoded transistor enabled during readout. Configurable pixels can be created to allow both common source read out as well as source follower read out.

[0078] FIG. 1 illustrates a 1T ion sensitive pixel according to one embodiment of the present invention. As shown, the pixel 100 may have one and only one transistor 101, one and only one row line R and one and only one column line C. The transistor 101 is shown as an n-channel MOSFET (NMOS) transistor in a p-type epitaxial substrate available using standard CMOS processes in this embodiment. It should be understood that NMOS is only used as an example in the present invention, and the transistor 101 may be a PMOS as well. The selection of NMOS or PMOS as a preferred device depends on which device does not require a top-side bulk contact for a given process. Typically NMOS is preferred when using a P+ wafer with P− epitaxy layer (called an epi-wafer) because the underlying P+ substrate biases the bulk on an array of pixels without the need to wire in a bulk contact at each pixel location. Therefore, a global bulk contact is an attractive combination for use with a 1T pixel where a small pixel pitch is required. The floating gate G of the transistor 101 may contain trapped charge, which may be properly discharged such that the electrode is at approximately the same potential as the substrate when all other terminals are also biased to the substrate potential. The row line R may be capacitively coupled to the drain D of the transistor 101, and the column line may be coupled to the source S of the transistor 101. A gate to drain overlap capacitance Cgd may form between the gate G and the drain D. The pixel 100 may be addressable from the row line R, which supplies the column current (i.e., drain-to-source current of the transistor 101) and boosts the potential at the floating gate.

[0079] In a one-transistor pixel array, such as the one shown in FIG. 3, row selection may be facilitated by boosting the FG nodes for a particular row. In one embodiment, the readout of the pixel is a winner-take-all circuit, which will be described below.

[0080] FIG. 2 illustrates the cross section of a 1T pixel according to one embodiment of the present invention. The transistor in the 1T pixel may be formed using an n-channel FET device by having a drain D and a source S formed using n-type implants within a p-type semiconductor. As shown, the transistor may have a floating gate G, the drain D and the source S. The source S may be coupled to the column line C and the drain D may be coupled to the row line R. Lightly doped drain (LDD) regions may create a gate to drain overlap capacitance Cgd and/or a gate to source overlap capacitance Cgs.

[0081] In one embodiment, the 1T ion pixel 100 may work by boot-strapping the row selection line R to the floating gate G while at the same time providing a source of current for the column line bias. In the simplest form, this bootstrapping occurs without adding any extra capacitors. The gate to drain overlap capacitance Cgd, as shown in FIGS. 1 and 2, may naturally form the necessary capacitive coupling. To increase capacitive coupling, if desired, the row selection metal line can form an extra metal capacitor to the floating metal electrode or more significant source and drain extensions can be made with ion implantation.

[0082] FIG. 3 shows the schematic of an array of pixels with column readout switches according to one embodiment of the present invention. For illustrative purposes, four 1T pixels 301, 302, 303 and 304 of an array 300 are shown arranged into two rows and two columns, though the array 300 could extend to an array of any size of 1T pixels. The 1T pixel may be similar to the one shown in FIG. 1. The drains of pixels 301 and 302 are coupled to a row line R0, and the sources of pixels 301 and 302 are coupled to column lines C0 and C1 respectively. The drains of pixels 303 and 304 are coupled to a row line R1, and the sources of pixels 303 and 304 are coupled to column lines C0 and C1 respectively. The pixel array can be loaded with a current source but the simplest implementation makes use of just a single switch that precharges the column line to a low potential such as the substrate potential. A column readout switch 305 is coupled to the column line C0 and a column readout switch 306 is coupled to the column line C1. The column readout switch 305 comprises a switch Sa, a switch Sb, a current source Isource and a capacitor Cw. The switch Sa is used for precharging the column line and to initialize the column line quickly between samples. The switch Sb is used to sample and hold the analog value that is read on the column line. In some cases, neither a sampling capacitor nor a switch Sb are required if the pixel is converted to digital through and analog to digital converter while the pixel is held under bias. The switch Sa is used to ground the column line C0. After the column line switch Sb is open the sample is held in the capacitor, the final value on the column line, as sampled by the capacitor, will be determined almost entirely by the active row because the circuit operates according to “a winner take-all” mode (i.e., the resulting voltage represents the largest voltage of the ISFETs coupled to the readout circuit). The column readout circuit 306 functions similarly.

[0083] The operation of this pixel depends on the fact that the signal range of any given pixel is small compared to the supply voltage or read range of the source follower. For example, the useful signal range may be only 100 mV and the supply voltage may be 3.3V. When a row is selected, the R line is driven to an active high voltage VH, while all other row lines are held at an active low voltage VL. The voltage VL is selected to be approximately equal to the nominal voltage on the column line C during the readout of any given pixel. Because the signal range is small, this voltage is known to within 100 mV in this example. Therefore, the drain to source voltage of all inactive pixels is always held to small values. This point is only critical if the gate to source voltage of inactive pixels is near the threshold of the device. For the row driven to VH, the FG voltages for that row are significantly higher than the other rows because of the bootstrapping that occurs when the row line transitions to VH. After the column line switch Sb is open, the final value on the column line will be determined almost entirely by the active row because the circuit operates according to the winner take-all mode.

[0084] There are two sources of current from other rows that can distort the signal value (one that adds current and one that takes away current) and there must be enough bootstrapping available to successfully read pixels without significant interaction from the other rows that produce these sources. The analysis to determine how much bootstrapping is needed is as follows. By the time the pixel is sampled, the device has entered the subthreshold region of operation which has a transconductance slope, for example, of approximately 100 mV/decade. This means that for every 100 mV of change in gate voltage, the current changes by 10 times. In order to effectively read a single pixel, a criteria is set so that 99% of the current on the column line is attributable to the active row and only 1% is attributable to the inactive rows (distortion current). From here it can be determined how much bootstrapping is necessary. With only 2 rows in the pixel array, a 200 mV difference in the floating gate voltages is needed according to the subthreshold slope. Since a signal range of about 100 mV is also needed to be accounted for, the total requirement is about 300 mV. If there are 10 rows, there may be 10 times more contribution from inactive rows. Therefore an extra 100 mV is needed. If the array is increased to 100 rows, another 100 mV is needed. If the array is increased to 10̂n rows, 300+100*n mV is needed. As an example, a 10000 (10̂4) row pixel array only requires a total of 700 mV (300+100*4) of bootstrapping. This amount of bootstrapping can be achieved from the overlap capacitance of the gate and drain. If more capacitance is needed, extra coupling can be facilitated in the mask layout. The above analysis only applies to pixels contributing to the readout current.

[0085] Pixels can also take current away from the column line and sink it through the deactivated row lines. Since the deactivated row line is set to approximately the level of the column line, this current draw will be minimal but it must still be quantified and controlled. To accomplish this, the final current on the column line should not be allowed to diminish beyond a certain level. This is ensured by loading the column with a small current sink such as 1 uA. For a W/L (width to length) ratio of 1, a transistor biased at its threshold will have a saturation current of about 0.1 uA. This current decreases by a factor of 10 for every 100 mV of reduction in gate to source voltage. If less than 1% contribution of current is required, the VGS of inactive pixels needs to be kept to 100+100*n mV below the threshold voltage where 10̂n is the number of pixels in the row. Thus, for a 10000 row pixel array, VGS needs to be kept to 500 mV below threshold. A typical 3.3V NMOS transistor has a VT of 600 mV. Therefore, VGS should be less than 100 mV for inactive pixels. Assuming that the FG has a nominal voltage of 0V when the row (R) and column (C) lines are at 0V, this condition is met even as R and C couple to the FG. If the FG has a larger nominal voltage than 0V (for example, due to the trapped charge), more bootstrapping is necessary to cause the column line to reach a level within 100 mV of the FG. As long as the nominal FG voltage is sufficiently low, the second criteria for minimizing distortion current is not a limiting factor. Finally, enough bootstrapping is needed to produce a current on the column line that matches the bleeding current so that the pixel can produce a measurable voltage on the column line. If VG is nominally 0 v, then 700 mV is needed for bootstrapping. Therefore, for an NMOS with VT as large as 600 mV, the amount of bootstrapping required is simply limited by the VT. In order to readout the pixel with margin, a good target for bootstrapping is 1V. This leaves 300 mV of range for variation. Achieving 1V of bootstrapping is practical within a 3.3V supply.

[0086] All the current from the column readout is distributed through the row line. This causes significant droop in the voltage of the row line if the column current is also significant. The voltage droop affects the bootstrapping level but is not detrimental to the readout of the source follower because variation in drain voltage has only a second order effect. Since pixels are read out with multiple samples, offsets are canceled such that the droop does not affect the sensitivity of the pixels.

[0087] It should be noted that the same layout can be used for both source follower readout and common source readout as long as optimizations are not made for either. Only accommodations that need to be made are in the column circuits. This makes for a flexible readout architecture and either readout method may be used depending on the necessary signal range. If the signal needs a high gain, the common source mode should be used. Otherwise, the source follower mode may be used.

[0088] FIG. 4 shows the source follower configuration of the 1T pixel according to one embodiment of the present invention. The source follower mode has a buffered readout and operates in a voltage mode, and has a gain less than 1. As shown, the sole transistor 401 may be coupled to an input voltage Vi at its gate G and to a fixed voltage at its drain D. The source S of the transistor 401 may be grounded via a current source Isource. The output voltage Vo may be taken from the source of the transistor 401. A coupling capacitance Cc may exist between the input and the gate of the transistor 401, a parasitic capacitor Cgd may exist between the gate G and the drain D of the transistor 401, and a parasitic capacitor Cgs may exist between the gate and the source S of the transistor 401.

[0089] The following analysis is given for the gain of the source follower readout. Referring to FIG. 4, the gain of the circuit (G) may be defined as Vo/Vi. Using reference pixels the electrode of the system may be swept to measure the gain such that Vo/Vi=G. Using the measured value of a parameter G, which is 0.65 in this example, the ratio of Cc to Cgd may be determined. As will be discussed later, it is this ratio that will determine the gain in the common source mode. The input capacitance of the source follower is Ci=Cgd+Cgs(I−Asf), wherein Asf is the gain of source follower. Due to the body effect, Asf is approximately 0.85. The capacitive divider relating to the input voltage on the FET is Cc/(Ci+Cc) and therefore, Cc/(Ci+Cc)=G/Asf. Since Cgs is about 3-5 times larger than Cgd and Asf is about 0.85, Ci is approximately 2Cgd. Therefore, Cc=2Cgd(G/(Asf−G)). In this example, the ratio of Cc to Cgd is about 6.5.

[0090] In one embodiment, the present invention obtains voltage gain by reading out with the common source configuration. It is desirable to achieve both a reduction in pixel size as well as an increase in signal level. The present invention eliminates the ancillary transistors in other pixel designs (e.g., 2T and 3T discussed below) and uses the source of the ISFET as the selection line to achieve both of these goals. The common source mode is a gain mode and a current mode.

[0091] FIG. 5A shows a 1T common source ion sensitive pixel according to one embodiment of the present invention. As shown, the pixel 500 may have one and only one transistor 501, one and only one row line R and one and only one column line C. The transistor 501 is shown as an n-channel MOSFET (NMOS) transistor in a p-type epitaxial substrate available using standard CMOS processes in this embodiment, although it may be a p-channel MOSFET as well. An NMOS device is typically preferred in use with a P+ epi wafer that requires no front side bulk contacts. Technically a PMOS could be use with a N+ epi wafer, but this configuration is not as commonly produced in standard CMOS processes. The row line R may be coupled to the source S of the transistor 501, and the column line may be coupled to the drain D of the transistor 501. The row selection is facilitated by switching on a path for the source voltage, and the readout of the pixel is through the drain.

[0092] The schematic of an array of pixels with column readout switches according to one embodiment of the present invention is shown in FIG. 6. The array 600 has four 1T common source pixels 601, 602, 603 and 604. The 1T pixel may be similar to the one shown in FIG. 5A. In this example, pixels are arranged into two rows and two columns. The drains of pixels 601 and 602 are coupled to a column line C0, and the sources of pixels 601 and 602 are coupled to row lines R0 and R1 respectively. The drains of pixels 603 and 604 are coupled to a column line C1, and the sources of pixels 603 and 604 are coupled to row lines R0 and R1 respectively. A column readout switch 605 is coupled to the column line C0 and a column readout switch 606 is coupled to the column line C1. The column readout switch 605 comprises a switch Sa, a switch Sb, a resistor R and a capacitor C_w0. The column readout switch 606 comprises a switch Sa, a switch Sb, a resistor R and a capacitor C_w1. The switch Sa may pull the voltage on the column line to a fixed voltage, for example, to a 3.3V supply. When the column line switch Sb is open, the final value on the column line will be determined by the active row since the switch Sb, along with the capacitor C_w0, acts as a sample and hold circuit.

[0093] The pixel array can be loaded with a current source with finite output resistance or another load device such as a resistor. Normally the row selection lines will be held at an active high voltage VH. When a row is selected for readout, its row selection line is pulled low to VL. The value of VL is set such that the nominal current level is about 1 uA. If the FG has a value of 100 mV higher than the norm, 10 times this current will result on the column line. If the value of FG is 100 mV lower than the norm, the current will be 10 times lower. The settling time of the signal on the column line will be signal dependent. The voltage gain is achieved with the selection of the value of R and it can be configurable to achieve programmable gain. For example, if R is 100 k ohms, then the 100 mV, translates to 1V at the output.

[0094] The actual circuit is more complicated than just a simple common source amplifier because of the parasitic capacitance involved. Since the FG node is not driven, but rather capacitively coupled to the output, there is a feedback mechanism that limits the gain. This limit is roughly equal to the total capacitance at the FG node to the gate to drain capacitance. This ratio may be about 3. It could be designed to achieve higher gain such as 10 times with careful mask operations to reduce source and drain extensions.

[0095] FIG. 7A shows the cross section of a 1T common source pixel according to one embodiment of the present invention. The transistor in the 1T pixel may be formed using an n-channel FET device by having a drain D and source S be formed using n-type implants within a p-type semiconductor. As shown, the transistor may have a floating gate G, the drain D and the source S. The source S may be coupled to the row line R and the drain D may be coupled to the column line C. Lightly doped drain (LDD) regions may create a gate to source overlap capacitance Cgs and a gate to drain overlap capacitance Cgd.

[0096] The overlap capacitance created by the LDD regions can be reduced by skipping the LDD implants at the drain for the device. FIG. 7B shows the cross section of a 1T common source pixel according to one embodiment of the present invention. FIG. 7B shows a drain node with a missing LDD region. This missing region reduces the capacitance and increases gain. This can be achieved through masking out the LDD implants and can be implemented in standard CMOS processing.

[0097] In the 1T pixel shown in FIG. 5A, since the source current must be supplied from the row selection line, variations in current due to variations in signal will create variations in voltage. These variations can distort the measurements. Therefore the row selection line should be low resistance and the driver for that line should also supply a steady source voltage independent of the current load. Where this is not possible, the current can be supplied from the column line and a second selection transistor can be added to form a 2T pixel for common source read out, as shown in FIG. 10A described below. Since the gain is limited by the parasitic overlap capacitance, it is expected that the best load to use is a current source implemented with transistors of high output resistance. In this case, relatively constant current will be maintained in all devices since the gain is achieved through capacitor ratios. This makes the 1T configuration feasible since voltage variation at the source is minimal, even with a single row selection line that carries all the current.

[0098] The pixel in common source readout configuration is shown in FIG. 5B. The transistor forms an amplifier with negative voltage gain. This negative voltage gain forms a natural feedback loop with the parasitic capacitors in order to control the gain. The open loop gain of the amplifier is A=gm(ro), wherein gm is a transconductance. The value A is typically larger than 100 for a given bias condition and process technology. As shown in FIG. 5C, the common source equivalent circuit has a feedback capacitance Cgd, a coupling capacitance Cc, and Cgs.

[0099] Since A is large compared to the loop gain, the negative input terminal may be considered as a virtual ground node and the gain of the circuit may be determined as Vo/Vi=−Cc/Cgd. Since this ratio is known from the analysis or measured values of the source follower configuration, the gain may be determined to be about 6.5. However compared to the source follower, the gain is Vo/Vi=2/(Asf−G). In this example, a gain of 10 is realized over the source follower configuration. A lower bound on this gain is given by assuming that the input capacitance of the source follower is solely due to Cgd and that the Asf is equal to 1. In this case the gain is about 3. Since neither of these conditions is realistic, the gain is expected to always exceed this number. Thus, if the gain of the source follower configuration of a pixel is known, the gain of the common source configuration of this pixel is also known. In addition, the higher the gain, the more sensitive the pixel is. This makes the common source configuration preferable.

[0100] Flicker noise can be reduced by using a channel doping of the same type as the minority carrier. For example, an NMOS with a n-type implant produces a buried channel transistor. To shift the workfunction of the device, a P+ gate electrode can be used.

[0101] One derivative of the one-transistor pixel allows for increased programmable gain through a cascoded transistor enabled during readout.

[0102] Since the gain of the common source readout is limited by the Cgd capacitance, as shown in FIG. 5B, lowering this capacitance can increase the gain. FIG. 8 shows a common source pixel with a cascoded row selection device. As shown, a transistor 801 may be added to a common source pixel, e.g., the circuit shown in FIG. 5B. The gate of the transistor 801 may be coupled to a voltage Vb, and the source of the transistor 801 may be coupled to the drain of the transistor 501. The output voltage Vo may be taken from the drain of the transistor 801. The cascode effectively removes the Cgd capacitance from the feedback loop and replaces it with Cds which is much smaller. Gain on the order of the loop gain is then achievable, which may exceed 100.

[0103] Higher gain and variable gain may be produced in the 1T configuration by bringing the cascode device outside the pixel to the column line. FIG. 9 shows a one-transistor pixel array with cascoded column circuit. This allows high gain and yet still allows the pixel pitch to be minimized with only 1 transistor per pixel. The shown pixel array is a column having a number of one-transistor pixels (e.g., 500) connected in series, and has a cascode device at the base of the array. The cascode device may comprise a transistor 901. The gate of the transistor 901 may be coupled to a bias voltage Vb, the source of the transistor 901 may be coupled to the drain of the transistor 501, and the drain of the transistor 901 may be coupled to a fixed voltage via a current source. The output voltage Vo may be taken from the drain of the transistor 901. It should be understood that the array may have a number of columns.

[0104] In this case, the cascode forces the drain of the pixel to remain at a fairly steady voltage over the range of inputs. This causes the pixel to push nearly all of the change in current through the cascode device at the base of the array and into the current load. This reduces the negative feedback from Cds, which would otherwise limit the gain. Given that the current load has infinite output resistance and there is effectively no coupling capacitor to the FG node, the gain of the pixel is now −(gmlrO1+1)gm2rO2, wherein gml is the transconductance of the cascode device at the base of the column line and gm2 is the transconductance of the pixel and rO1 and rO2 are the small signal output resistances as seen at the drain. The value of the output resistance is determined by channel length modulation. Longer gate lengths produce higher output resistance because the effect of channel length modulation is minimized. Since this gain is so large, it can be limited and configured by variation of the current source output resistance, which is shown as Radj in FIG. 9. This allows for programmable gain at the column level while maintaining a simple 1 transistor pixel. The gain of the pixel is then set by −gm2RL, assuming that the load resistance RL is much smaller than the output resistance of the cascode configuration, where R_L is the adjusted value of Radj. The gain is now configurable and programmable within the range of 1 to 100 or larger. For example, if the bias current is about 5 uA, the transconductance of the pixel is about 50 uA/V, and a load resistance of 20K ohms is needed for gain of 1. A gain of 10 is achieved with a 200K ohm load and gain of 100 with a 2M ohm load. There are many was to implement the effect of the cascode device at the column line. The main purpose of the cascode, as shown in FIG. 901 as an NMOS transistor, is that the column line is held to a potential that is largely independent of the current level in the pixel. A differential amplifier with high gain can be applied to maintain this condition more precisely. This approach would be called gain-enhanced cascoding.

[0105] Various layout choices can be made to implement a 1 T and 2T transistor. In order to reduce the size of the pixel the source and drains of adjacent pixels can be shared. In this way a single row selection line enables 2 rows at a time. This reduces the row wiring: two columns are then read out at once for a given column pitch. Such a scheme is shown in FIGS. 10A and 10B. As shown, a pixel array 1000 comprises transistors 1001, 1002, 1003 and 1004 in a column. The source of 1001 is coupled to a row line R2, and the source of 1004 is coupled to a row line R0. Transistors 1001 and 1002 may form a mirror M1, and transistors 1003 and 1004 may form a mirror M2. The drain of 1001 and 1002 are coupled to a column line CA, and the drain of 1003 and 1004 are coupled to a column line CB.

[0106] In one embodiment, the cascoded device is gain-enhanced with a differential amplifier in feedback to control a transistor that maintains a constant voltage on the column line.

[0107] In a pixel array, a row selection device may be used for selection and isolation. When a row selection line is activated, the row selection device (a MOSFET) forms a channel due to the gate voltage exceeding a threshold voltage and acts like a switch. When the row selection is deactivated, the channel is diminished. It is important to note that a row selection device never really completely turns “on” or “off”. It only approximates a switch. When the gate is substantially lower than the source of the row selection transistor, good isolation is achieved and the pixel with the active row selection can be read effectively without input from deactivated pixels. With many rows in an array of pixels, it is necessary to achieve a given level of isolation for each row selection device. That is, the requirements for the row selection device depend on the number of rows.

[0108] FIG. 11 shows a two-transistor (2T) pixel according to one embodiment of the present invention. As shown, the 2T pixel 1100 comprises an ISFET 1101 and a row selection device 1102. In the pixel 1100, the source of the ISFET 1101 is coupled to a column line Cb, the drain of the row selection device 1102 is coupled to a column line Ct, and the drain of the ISFET 1101 is coupled to the source of the row selection device 1102. The gate of the row selection device 1102 is coupled to a row line R.

[0109] Both ISFET 1101 and the row selection device 1102 are shown as NMOS, but other types of transistors may be used as well. The 2T pixel 1100 is configured as the source follower readout mode, although 2T pixels may be configured as the common source readout mode.

[0110] FIG. 12A to 12H illustrate more 2T pixel configurations according to embodiments of the present invention. In these Figures, “BE” stands for “with body effect”, i.e. the ISFET is body-effected because the body terminal is connected to the analog supply voltage or analog ground voltage (depending on whether the ISFET transistor type is p-channel or n-channel MOS). The body effect is eliminated if the body terminal is connected to the source terminal of the transistor. “PR” stands for “PMOS devices in reversed positions”, i.e. the positions of the p-channel ISFET and row selection device in the pixel circuit topology have been reversed (or switched around). “PNR” stands for “PMOS/NMOS devices in reversed positions”, i.e. the positions of the p-channel ISFET and n-channel row selection device in the pixel circuit topology have been reversed (or switched around).

[0111] FIG. 12A illustrates a 2T pixel, according to one embodiment of the present invention. As shown, both the ISFET and the row selection device SEL are p-channel MOS transistors, with the source terminal of the ISFET coupled to the drain terminal of the row selection device. The drain terminal of the ISFET is connected to the analog ground voltage and the source terminal of the row selection device is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the source terminal of the row selection device.

[0112] FIG. 12B illustrates a 2T pixel, according to one embodiment of the present invention. As shown, both the ISFET and the row selection device SEL are p-channel MOS transistors, with the source terminal of the ISFET connected to the body terminal to eliminate the body effect, and also connected to the drain terminal of the row selection device. The drain terminal of the ISFET is connected to the analog ground voltage and the source terminal of the row selection device is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the source terminal of the row selection device.

[0113] FIG. 12C illustrates a 2T pixel, according to one embodiment of the present invention. As shown, both the ISFET and the row selection device SEL are p-channel MOS transistors, with the drain terminal of the ISFET connected to the source terminal of the row selection device. The drain terminal of the row selection device is connected to the analog ground voltage and the source terminal of the ISFET is connected to a current source. The output voltage Vout is read out from the source terminal of the ISFET.

[0114] FIG. 12D illustrates a 2T pixel, according to one embodiment of the present invention. As shown, both the ISFET and the row selection device SEL are p-channel MOS transistors, with the drain terminal of the ISFET connected to the source terminal of the row selection device. The drain of the row selection terminal is connected to the analog ground voltage and the source terminal of the ISFET is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the source terminal of the ISFET. The source terminal of the ISFET is connected to the body terminal to eliminate the body effect.

[0115] FIG. 12E illustrates a 2T pixel, according to one embodiment of the present invention. As shown, the ISFET and the row selection device SEL are p-channel and n-channel MOS transistors respectively, with their source terminals connected together. The drain terminal of the ISFET is connected to the analog ground voltage and the drain of the row selection device is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the drain terminal of the row selection device.

[0116] FIG. 12F illustrates a 2T pixel, according to one embodiment of the present invention. As shown, the ISFET and the row selection device SEL are p-channel and n-channel MOS transistors respectively, with their source terminals connected together. The drain terminal of the ISFET is connected to the analog ground voltage and the drain of the row selection device is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the drain terminal of the row selection device. The source terminal of the ISFET is connected to the body terminal to eliminate the body effect.

[0117] FIG. 12G illustrates a 2T pixel, according to one embodiment of the present invention. As shown, the ISFET and the row selection device SEL are p-channel and n-channel MOS transistors respectively, with their drain terminals coupled together. The source terminal of the row selection device is connected to the analog ground voltage and the source terminal of the ISFET is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the source terminal of the ISFET.

[0118] FIG. 12H illustrates a 2T pixel, according to one embodiment of the present invention. As shown, the ISFET and the row selection device SEL are p-channel and n-channel MOS transistors respectively, with their drain terminals coupled together. The source terminal of the row selection device is connected to the analog ground voltage and the source terminal of the ISFET is connected to a current source, which provides a bias current to the pixel. The output voltage Vout is read out from the source terminal of the ISFET. The source terminal of the ISFET is connected to the body terminal to eliminate the body effect.

[0119] FIGS. 13A to 13D illustrate common source 2T cell configurations according to embodiments of the present invention. In FIGS. 13A and 13B, both the ISFET and the row selection device are n-channel MOS transistors, and in FIGS. 13C and 13D, both the ISFET and the row selection device are p-channel MOS transistors.

[0120] In FIG. 13A, the source terminal of the ISFET is connected to the analog ground supply and the drain terminal of the row selection device is connected to a current source, which provides a bias current to the pixel. The source terminal of the row selection device and the drain terminal of the ISFET are connected together. The output voltage Vout is read out from the drain terminal of the row selection device.

[0121] In FIG. 13B, the source terminal of the row selection device is connected to the analog ground supply and the drain terminal of the ISFET is connected to a current source, which provides a bias current to the pixel. The drain terminal of the row selection device and the source terminal of the ISFET are connected together. The output voltage Vout is read out from the drain terminal of the ISFET.

[0122] In FIG. 13C, the source terminal of the ISFET is connected to the analog supply voltage, and the drain terminal of the row selection device is connected to a current source, which provides a bias current to the pixel. The source terminal of the row selection device and the drain terminal of the ISFET are connected together. The output voltage Vout is read out from the drain terminal of the row selection device.

[0123] In FIG. 13D, the source terminal of the row selection device is connected to the analog supply voltage, and the drain terminal of the ISFET is connected to a current source, which provides a bias current to the pixel. The source terminal of the ISFET and the drain terminal of the row selection terminal are connected together. The output voltage Vout is read out from the drain terminal of the ISFET.

[0124] FIG. 14A shows a 2T pixel array according to one embodiment of the present invention. For illustrative purposes, eight 2T pixels are shown arranged into two columns, though the 2T pixel array 1400 could extend to an array of any size of 2T pixels. Each column pitch contains three column lines cb[0], ct[0] and cb[1], The row lines rs[0], rs[1], rs[2] and rs[3], connect to all columns in parallel. A row selection device 1401RS and an ISFET 1401IS may form one 2T pixel, with the source of 1401IS connected to the drain of 1401RS. The source of 1401RS is connected to the column line cb[0], and the drain of 1401IS is connected to the column line ct[0]. The gate of 1401RS is connected to the row line rs[0). This pixel is mirrored in a pixel comprising 1402IS and 1402RS, with drains of 1401IS and 1402IS connected to the column line ct[0], and the gate of 1402RS connected to the row line rs[1]. The pixel comprising 1402IS and 1402RS is mirrored in a pixel comprising 1403IS and 1403RS, with the source of 1402RS and 1403RS connected to the row line cb[1], and the gate of 1403RS coupled to the row line rs[2]. The pixel comprising 1403IS and 1403RS is mirrored in a pixel comprising 1404IS and 1404RS, with the drains of 1403IS and 1404IS connected to the row line ct[0], the gate of 1404RS coupled to the row line rs[3], and the source of 1404RS coupled to the column line cb[0]. In the embodiment shown in FIG. 14, each of the IS devices is an ISFET and each of the RS devices is a row select device.

[0125] The right column, including a pixel consisting of 1405RS and 1405IS, a pixel consisting of 1406RS and 1406IS, a pixel consisting of 1407RS and 1407IS, and a pixel consisting of 1408RS and 1408IS, is coupled to column traces cb[2], ct[1], and cb[3] in substantially the same manner as described above.

[0126] FIGS. 14B and 14C show a layout for a 2×2 2T pixel array according to an embodiment of the present invention. The 2×2 2T pixel array may be part of the pixel array 1400. FIG. 14B shows that polysilicon gates for 1401RS, 1401IS, 1402RS and 1402IS may be placed on top of a continuous diffusion layer 1410 and polysilicon gates for 1405RS, 1405IS, 1406RS and 1406IS may be placed on top of a continuous diffusion layer 1412. In one embodiment, the continuous diffusion layers 1410 and 1412 may run from the top of the pixel array to the bottom of the pixel array. That is, the diffusion layer may have no discontinuities in the pixel array.

[0127] FIG. 14C shows where microwells for ISFETs 1401IS, 1402IS, 1405IS and 1406IS may be placed. The microwells may be used to hold analyte solutions that may be analyzed by the ISFETs. As shown in FIG. 14C, in one embodiment, the microwells may each have a hexagonal shape and stacked like a honeycomb. Further, in one embodiment, the contact may be placed directly on top of the gate structure. That is, the ISFETs may have a contact landed on polysilicon gate over thin oxide.

[0128] The pixel array 1400 has high density because of continuous diffusion, shared contacts, mirrored pixels, and one ct (column top) line and 2 cb (column bottom) line per physical column. A global bulk contact may be implemented by using a P+ wafer with P− epitaxy region.

[0129] The arrangement of pixel array 1400 provides for high speed operation. Row lines rs[0] and rs[1] are selected together and readout through cb[0] and cb[1]. This leads to a 4 times faster readout due to twice the number of pixels enabled for a single readout and half the parasitic load of a continuous array, allowing each column to settle twice as fast. In an embodiment, the full array is separated into a top half and a bottom half. This leads to another 4 times faster readout time due to twice the number of pixels readout at a time (both out the top and the bottom) and half the parasitic load of a continuous array. Thus, the total increase in speed over a single row selected continuous array is 16 times.

[0130] In an embodiment, both top and bottom halves of the pixel array may be enabled at the same time during readout. This can allow a multiplexing of readout between the top half and the bottom half. For example, one half can be doing a “wash” (e.g., flushing out reactants from the wells over the pixel devices) and the other half can be performing the readout. Once the other half is read, the readout for the two halves is switched.

[0131] In an embodiment, a 2T pixel design can incorporate two chemically-sensitive transistors (e.g., ISFETs) rather than one chemically-sensitive transistor and one row select device as described with respect to FIGS. 11-14. Both chemically-sensitive transistors, or ISFETs, can be NMOS or PMOS device and configured in a source follower or common source readout mode. Possible uses of such a 2T pixel may be where the first chemically-sensitive transistor has a different sensitivity to a particular analyte to that of the second chemically-sensitive transistor, allowing a local and in-pixel differential measurement to be made. Alternatively, both chemically-sensitive transistors may have the same sensitivity to a particular analyte, allowing a local and in-pixel average measurement to be made. These are among two examples of potential uses for this embodiment, and based on the description herein, a person of ordinary skill in the art will recognize other uses for the 2T pixel design that incorporate two chemically-sensitive transistors (e.g., ISFETs).

[0132] In one embodiment, a column circuit allows column lines to be swapped to a sampling circuit such that either source-side or drain-side row selection can be made in either source follower mode or common source mode.

[0133] One or more charge pumps may be used to amplify the output voltage from a chemically-sensitive pixel that comprises one or more transistors, such as those described above.

[0134] One or more charge pumps may be used to amplify the output voltage from a chemically-sensitive pixel that comprises one or more transistors, such as those described above.

[0135] FIG. 15 shows a capacitive charge pump with a two times voltage gain according to one embodiment of the present invention. A charge pump 1500 may comprise φ1 switches 1501, 1502, 1503 and 1504, φ2 switches 1505 and 1506, and capacitors 1507 and 1508. Vref1 and Vref2 are set to obtain the desired DC offset of the output signal, and both are chosen to avoid saturation of the output during the boost phase. The operation of the charge pump may be controlled by timing signals, which may be provided by a timing circuit.

[0136] At time t0, all switches are off.

[0137] At time t1, φ1 switches 1501, 1502, 1503 and 1504 are turned on. The track phase may start. An input voltage Vin, which may be from an ion sensitive pixel, may start to charge capacitors 1507 and 1508.

[0138] At time t2, φ1 switches 1501, 1502, 1503 and 1504 are turned off, and capacitors 1507 and 1508 are charged to Vin−Vref1.

[0139] At time t3, φ2 switches 1505 and 1506 are turned on, while φ1 switches 1501, 1502, 1503 and 1504 remain off. The boost phase may start. The capacitor 1507 may start to discharge through the capacitor 1508. Since the capacitors are in parallel during the track phase and in series during the boost phase, and the total capacitance is halved during the boost phase while the total charge remains fixed, the voltage over the total capacitance must double, making Vout approximately two times Vin.

[0140] A source follower SF may be used to decouple the gain circuit from the following stage.

[0141] The charge pump 1500 may provide a two times gain without a noisy amplifier to provide a virtual ground.

[0142] FIG. 16 shows a charge pump according to an embodiment of the present invention.

[0143] At time t0, all switches are off.

[0144] At time t1, φ1 switches 1501, 1502, 1503, 1504, 1601 and 1602 are turned on. The track phase may start. An input voltage Vin, which may be from an ion sensitive pixel, may start to charge capacitors 1507, 1508 and 1604.

[0145] At time t2, φ1 switches 1501, 1502, 1503, 1504, 1601 and 1602 are turned off, and capacitors 1507, 1508 and 1604 are charged to Vin−Vref1.

[0146] At time t3, φ2 switches 1505, 1606 and 1603 are turned on, while φ1 switches 1501, 1502, 1503, 1504, 1601 and 1602 remain off. The boost phase may start. The capacitor 1507 may start to discharge through the capacitors 1508 and 1604, and the capacitor 1508 may start to discharge through the capacitor 1604. Since the capacitors are in parallel during the track phase and in series during the boost phase, and the total capacitance is divided by three during the boost phase while the total charge remains fixed, the voltage over the total capacitance must triple, making Vout approximately three times Vin.

[0147] FIG. 17 shows an embodiment of a charge pump according to an embodiment of the present invention. Two charge pumps 1500 shown in FIG. 15 are connected in series, enabling gain pipelining and amplifying input voltage Vin by a factor of four.

[0148] Additional series charge pumps can be added to increase the gain further. In a multi-stage charge pump, the capacitor values do not have to be the same size from stage to stage. It can be observed that the total area consumed by capacitors increases with the square of the gain. Although this feature may, in some cases, be undesirable with respect to area usage, power consumption, and throughput, the charge pump can be used without these penalties when the total noise produced by the ion sensitive pixel and associated fluidic noise is larger than the charge pump KT/C noise when a reasonable capacitor size is used.

[0149] FIG. 18 shows an embodiment of a charge pump according to an embodiment of the present invention. A feedback path including a source follower SFP and a switch φfb is added to the charge pump 1500, feeding the output Vout back to the input of the charge pump.

[0150] At time t0, all switches are off.

[0151] At time t1, a switch φsp is on, providing an input voltage Vin to the input of the charge pump 1500.

[0152] From time t2 to time t5, the charge pump 1500 operates to push the output voltage Vout to 2(Vin−Vref1), as described before with reference to FIG. 15.

[0153] From time t6 to t7, the switch φfb is on, feeding the output voltage 2(Vin−Vref1). back to the input of the charge pump 1500, and the first cycle ends.

[0154] During the second cycle, the charge pump 1500 amplifies the output voltage by 2(2(Vin−Vref1)). The process repeats, with the output being amplified during each cycle.

[0155] An ion sensitive MOS electrode is charge coupled to adjacent electrodes to facilitate both confinement and isolation of carriers. Measurements of ion concentration are made by discrete charge packets produced at each pixel and confined by potential barriers and wells. The ion sensitive electrode can act as either a barrier level or as a potential well. Working in the charge domain provides several benefits, including but not limited to: 1) increased signal level and improved signal to noise through the accumulation of multiple charge packets within each pixel, 2) better threshold matching of the MOS sensing and reference structures, 3) reduction in flicker noise, and 4) global-snap shot operation.

[0156] A floating electrode is used to detect ions in close proximity to the electrode. The electrode is charge coupled to other electrodes and to other transistors to form a pixel that can be placed into an array for addressable readout. It is possible to obtain gain by accumulating charge into another electrode or onto a floating diffusion (FD) node or directly onto the column line. It is desirable to achieve both a reduction in pixel size as well as increase in signal level. To reduce pixel size, ancillary transistors may be eliminated and a charge storage node with certain activation and deactivation sequences may be used.

[0157] The ion sensitive (IS) accumulation pixel contains some of the following concepts:

• • [0158] 1. Electrodes are charge coupled to the IS electrode;
  • [0159] 2. A source of carriers (electrons or holes) for charge packets;
  • [0160] 3. A reference electrode to act as a barrier or a well for the charge packets;
  • [0161] 4. A floating diffusion node for charge to voltage conversion;
  • [0162] 5. Ancillary transistors to provide buffering and isolation for addressable readout; and
  • [0163] 6. Sequences to eliminate some or all ancillary transistors depending on the application.

[0164] The basic IS accumulation pixel is shown in FIG. 19. Charge accumulation can occur either locally at the time of readout or globally during a separate integration time. The embodiment shown in FIG. 19 is a three transistor three electrode (3T3E) pixel. The three transistors include a reset transistor RT, a source follower 1901 and a row selection transistor RS, and the three electrodes include an electrode VS, an electrode VR, and an ion sensitive electrode 1902. The pixel also includes a transfer gate TX. It is also possible to configure the IS accumulation pixel with additional elements to allow simultaneous accumulation and readout. This can be done, for example, by adding 2 more electrodes to pipeline the process. In the basic configuration, charge is accumulated onto the floating diffusion node that is connected to the source of the reset (RT) control gate. In a rolling shutter operation, the floating diffusion (FD) is reset to CD=VDD. The row is then selected and readout through the source follower enabled by row selection (RS). Next, charge is accumulated onto the FD node which discharged the parasitic capacitor. A second sample is then taken. The difference between the samples represents the ion concentration. The samples are correlated and taken relatively quickly in time. Therefore, the thermal noise of the readout circuit is eliminated and the 1/f noise is reduced. To operate in a global shutter mode, all FD nodes are simultaneously reset to VDD. Then charge is accumulated on each isolated FD node. After accumulation, each row is selected by enabling the RS gate. The signal value is readout on the column line with a load on the source follower. Next the pixel is reset and sampled again. The difference between the samples represents the ion concentration. The 1/f noise is reduced through the double sampling. However, the thermal reset noise is not eliminated because the reset value is uncorrelated in time. The thermal noise can be reduced by half the power by following the reset operation with a subthreshold reset before sampling. In general, the thermal noise is low compared to the signal due to the charge accumulation. A correlated reset scheme with global shutter is available in other configurations.

[0165] The basic charge accumulation scheme is shown in FIG. 20 using the surface potential diagrams. Only the electrodes are shown since the transistors are only used for readout. In each of these sequences, increasing potential is pointing down as is conventional to show potential wells containing electrons. Four cycles of charge accumulation are shown in FIG. 20 A-Q. First, all charge is removed from the channel under the IS electrode and the channels are fully depleted using a high potential on FD (A). Next, the TX gate transitions to a low potential which creates the confinement barrier (B). A fill and spill operation is used to produce a charge packet proportional to the ion concentration at the IS electrode (C-D). In the next cycle, this charge packet is transferred to the FD node which discharges due to the electrons. The diagram shows electrons accumulating on the FD node, but the voltage is actually decreasing. After many cycles, as shown in FIG. 20 E-Q, the signal to noise ratio is improved and the signal can be read out with gain. Hundreds to millions of cycles can be used to amplify the signal.

[0166] In alternative embodiments, the order of electrodes may be switched, and/or the IS electrode may be used as the barrier rather than the well. Transistors may be added to this accumulation line to enable a large array of pixels. The ancillary transistors are used to increase speed. However, it should be noted that no transistors are necessary to enable a full pixel array of the accumulation line. Instead, an array can be partitioned such that no transistors are needed. In an embodiment, the FD nodes are connected to the column line. Before a pixel is read out, the column line is reset to VDD. Then a row is selected by accumulating charge for that row directly onto the column line. After many cycles, the column discharges to a value directly proportional to the ion concentration. Since the capacitance of the column line depends on the total number of rows, the amount of accumulation required, depends on the number of rows. The array can be partitioned into sub arrays to make timing scalable. For example, every 100 rows can contain a local source follower buffer that is then connected to a global array. This hierarchical approach can be used in general with all readout schemes to make massive arrays of pixels with fast readout.

[0167] Due to the thermal activity of carriers, charge packets cannot be generated without noise. Each fill and spill operation produces charge error proportional to KTC (thermal noise in the floating diffusion capacitor), where C is equal to Cox times the area of the ion sensitive electrode. During the fill operation charge can flow freely between the source of electrons and the confinement well. However, during the spill operation, the device enters the subthreshold mode and carriers move by diffusion, mainly in only one direction, which results in half of the thermal noise of a resistive channel. The total noise in electrons for each charge packet is therefore sqrt(KTC/2)/q where q represents the charge of one electron in coulombs (1.6×10e-19). The signal in electrons is equal to VC/q. The signal to noise ratio after n cycles is equal to V*sqrt(2nC/KT). Note that the signal to noise ratio improves by the square root of the number of cycles of accumulation. For small signal levels, the amount of accumulation will be limited to the threshold mismatch between the VR reference electrode and the ion sensitive electrode. Since there is a reference electrode in every pixel and the electrodes are charge coupled, the relative threshold mismatch between each pair of electrodes is small. Assuming, this difference is about 1 mV, over 1000 accumulation cycles should be feasible, thereby improving the signal to noise by more than 30 times. By way of example, if the signal is 1 mV and the electrode area is 1 square micron with Cox=5fF/um̂2, the signal to noise ratio after 1000 cycles is 50 to 1. Since the signal level then reaches 1 V, it is expected that no other noise source is relevant. For clarity, the dominant noise is simply the charge packet thermal noise which is well known.

[0168] FIGS. 21 and 22 show the IS accumulation pixel with only 2 transistors. The selection transistor is eliminated by using a deactivation sequence after a row is read out. To deactivate, the FD node is discharged, which reduces the potential of the FD node and disables the source follower for that row. The surface potential diagrams for the pixel of FIG. 22 are shown in FIG. 23.

[0169] FIG. 24 shows the IS accumulation pixel with 2 transistors and 4 electrodes. This pixel produces the fill and spill charge packets and readout all at the same FD node. The 4th electrode allows global shutter operation and correlated double sampling. For faster readout, single sampling can be used if charge accumulation sufficiently reduces the 1/f noise contribution. FIG. 25 shows the surface potential diagrams for the basic operation of the pixel of FIG. 24.

[0170] FIG. 26 shows an IS accumulation pixel with 1 transistor and 3 electrodes. The channel can be depleted and supplied from the same node. This pixel depends on charge coupling, and signal range is lower than signal range for the other pixels.

[0171] Several design permutations are available depending on the desired mode of operation. The CCD channels are surface mode and are built in standard CMOS technology preferably below 0.13 um. Extra implants can be added to avoid surface trapping and other defects. A channel stop and channel can be formed from donor and acceptor impurity implants. The channel can be made of multiple implants to produce a potential profile optimal for the mode of operation.

[0172] FIG. 27 shows an embodiment of a three transistor (3T) active pixel sensor. The three transistors are a reset transistor 2701, a source follower 2702 and a row selection switch 2703. The reset transistor 2701 has a gate controlled by a reset signal RST, a source coupled to the floating diffusion (FD) of a pixel, and a drain connected to a fixed voltage. The source follower 2702 has its gate connected to the source of the reset transistor 2701, and its drain connected to a fixed voltage. A row selection transistor 2703 has its gate connected to a row line, its drain connected to a fixed voltage and its source connected to a column. Other electrodes interacting with the pixel includes a transfer gate TG, an ion selective electrode ISE, an input control gate ICG, and an input diffusion ID. These three elements form charge coupled electrodes that are operated in an identical way to VS, VR, and TX in FIG. 19.

[0173] FIG. 28 shows an alternate embodiment of a 3T active pixel sensor. The difference between the sensor in FIG. 28 and the sensor shown in FIG. 27 is that the sensor 2800 has a second input control gate ICG2, which allows more control over the potential barrier near the ion-sensitive electrode.

[0174] FIG. 29 shows an embodiment of a 3T active pixel sensor with a sample and hold circuit, which may be used to eliminate signal variations. As shown, the gate of the row selection transistor 2703 is controlled by a RowSelm signal provided by a row selection shift register. The source of the row selection transistor 2703 is coupled to a current sink ISink 2902 and a column buffer 2903. The current sink ISink 2902 may be biased by a voltage VB1 and the column buffer, which may be an amplifier, may be biased by a voltage VB2.

[0175] The sample and hold circuit 2901 may include a switch SH, a switch CAL, a capacitor Csh, and an amplifier Amp. The switch SH's input is coupled to the output of the column buffer 2903, and its output is coupled to a voltage VREF through the switch CAL, the upper part of the capacitor Csh, and the input of the amplifier Amp. The amplifier is biased by a voltage VB2. The output of the amplifier is coupled to a switch 2904 controlled by a signal ColSeln from a column selection shift register. The output of the switch 2904 is buffered by an output buffer 2905 before reaching the output terminal Vout. The output buffer is biased by a voltage VB3.

[0176] FIG. 30 shows an embodiment of a 3T active pixel sensor with a correlated double sampling circuit. The most significant difference between the sensor in FIG. 30 and that in FIG. 29 is that the former uses a correlated double sampling circuit 3001 to measure the signal from the column buffer 2903. An amplifier in the correlated double sampling circuit 3001 receives at its first input the output of the column buffer 2903 via a switch SH, and a capacitor Cin. The amplifier receives a reference voltage VREF at its second input, and is biased by the voltage VB2. A reset switch RST and a capacitor Cf are coupled in parallel with the amplifier.

[0177] FIG. 31 shows an embodiment of a 2.5T active pixel sensor used for a four pixel array. Each of the pixels has its own transfer transistor TX1, TX2, TX3 and TX4 and its own reset transistor. The drain of each transfer transistor is coupled to the source of the reset transistor in the same pixel, and the source of each transfer transistor is coupled to the gate of the source follower.

[0178] FIG. 32 shows an embodiment of a 1.75T active pixel sensor for a four pixel array. Each of the pixels has its own transfer transistor. The source of each transfer transistor is coupled to the floating diffusion of the same pixel, and the drain of each transfer transistor is coupled to the drain of the reset transistor RST of the sensor.

[0179] The described embodiments may provide a chemical detection circuit with an improved signal-to-noise ratio. The chemical detection circuit may include a current source, a chemical detection pixel, an amplifier and a capacitor. The chemical detection pixel may comprise a chemically-sensitive transistor that may have first and second terminals and a row-select switch connected between the current source and chemically-sensitive transistor. The amplifier may have a first input and a second input, with the first input connected to an output of the chemically-sensitive transistor via a switch and the second input connected to an offset voltage. The capacitor may be connected between an output of the amplifier and the first input of the amplifier. The capacitor and amplifier may form an integrator and may be shared by a column of chemical detection pixels.

[0180] Some embodiments may also provide a chemical detection circuit with an improved signal-to-noise ratio. The chemical detection circuit may include a plurality of columns of chemical detection pixels. Each column of chemical detection pixels may comprise a current source, a plurality of chemical detection pixels, an amplifier and a capacitor. Each chemical detection pixel may comprise a chemical-sensitive transistor that may have first and second terminals and a row-select switch connected between the current source and chemically-sensitive transistor. The amplifier may have a first input and a second input, with the first input connected to an output of each chemically-sensitive transistor via a switch and the second input connected to an offset voltage. The capacitor may be connected between an output of the amplifier and the first input of the amplifier. The capacitor and amplifier may form an integrator that is shared by a column of chemical detection pixels.

[0181] Other embodiments may provide a method to generate an output signal from a chemical detection circuit. The method may comprise selecting a chemical detection pixel from a column of chemical detection pixels for readout, integrating a readout current from the chemical detection pixel to an integrator, and reading out an output voltage of the integrator.

[0182] FIG. 33 illustrates a block diagram of a chemical detection circuit 3300 according to an embodiment of the present invention. The chemical detection circuit 3300 may comprise a plurality of chemical detection pixels 3302.1-3302.N, a current source 3308, an amplifier 3310, an offset voltage V_sci 3314, a capacitor C_int 3312, and three switches 3316, 3318 and 3320. Each chemical detection pixel (e.g., 3302.1, . . . , or 3302.N) may comprise a chemically-sensitive transistor (e.g., 3304.1, . . . , or 3304.N, respectively) and a row-select switch (e.g., 3306.1, . . . , or 3306.N, respectively). The amplifier 3310 may have a first input terminal coupled to an output of the current source 3308 and a second input terminal coupled to the offset voltage V_sd 3314. The capacitor C_int 3312 may have a first side coupled to the first input terminal of the amplifier 3310 and a second side coupled to an output terminal of the amplifier 3310 via the switch 3318. The switch 3316 may be coupled between the first side of the capacitor C_int 3312 and the output terminal of the amplifier 3310. The switch 3320 may be coupled between the second side of the capacitor C_int 3312 and ground. In one embodiment, the plurality of chemical detection pixels 3302.1 to 3302.N may form a column of chemical detection pixels. The capacitor C_int 3312 may be configured as a negative feed back loop for the amplifier 3310 and thus, the capacitor C_int 3312 and amplifier 3310 may form an integrator for the column of chemical detection pixels. In one embodiment, the integrator may be shared by all chemical detection pixels of the column and may be referred to as a column integrator.

[0183] Each chemically-sensitive transistor may have a gate terminal that may be covered by a passivation layer. The gate terminal may have a floating gate structure sandwiched between a gate oxide and a passivation layer (e.g., floating gate G in FIG. 2). During operation, the passivation layer may be exposed to an analyte solution to be analyzed. Each chemically-sensitive transistor 3304.1˜3304.N may further have a first terminal connected to a first side of a respective row-select switch 3306.1˜3306.N and a second terminal connected to ground. For example, as shown in FIG. 33, the transistor 3304.1 may be a PMOS with a first terminal (e.g., the source) connected to a first side of the row-select switch 3306.1 and a second side (e.g., the drain) connected to ground. Each row-select switch (e.g., 3306.1, . . . , or 3306.N) of each chemical detection pixel may have a second side connected to the current source 3308. The second side of each row-select switch may also be coupled to the first input of the amplifier 3310.

[0184] In one embodiment, the chemical detection circuit 3300 may be configured so that each of the chemically-sensitive transistors (e.g., 3304.1, . . . , or 3304.N) may work in a current-mode. That is, each of the chemically-sensitive transistors may work as a transconductance amplifier. The ion-concentration of analyte being measured by the chemically-sensitive transistor may be detected by a current output. In one embodiment, each chemically-sensitive transistor 3304.1 to 3304.N may be an ion-sensitive field effect transistor (ISFET) and each row-select switch 3306.1 to 3306.N may also be a transistor.

[0185] During operation, when one chemical detection pixel is selected, the corresponding row-select switch may be closed. For example, as shown in FIG. 33, the chemical detection pixel 3302.1 may be selected, and thus the row-select switch 3306.1 may be closed. The current source 3308 may provide a DC bias current I_dc to the selected chemically-sensitive transistor 3304.1. The signal current I_sig resulting from gate voltage change of the chemically-sensitive transistor 3304.1 may be integrated onto the capacitor C_int 3312 and an output signal of the amplifier 3310 may be read out as V_out. The offset voltage V_sd 3314 to the second input of the amplifier 3310 may provide the source-to-drain voltage V_sd for the chemically-sensitive transistor to operate.

[0186] Each measurement operation may comprise two phases. The first phase of operation may be an integration phase and the second phase of operation may be a clear phase to clear charges. During the first phase of operation, the switch 3318 may be closed and switches 3316 and 3320 may be left open. After the output signal Vout is read out, the operation may enter the second phase, during which the switches 3316 and 3320 may be closed and the switch 3318 may be left open to clear out the charges accumulated on the capacitor C_int 3312. In one embodiment, a correlated-double-sampling (CDS) scheme may be implemented by closing switch 3320 during the second phase. This may allow the inherent offset voltage of the amplifier 3310 to be stored on the capacitor C_int 3312.

[0187] In one embodiment, the current source 3308 may be a programmable current source attached to each column to provide a DC bias current I_dc, which may be relatively large. In this configuration, the bias current I_dc will not integrate onto the capacitor C_int 3312 and thus the integrator may avoid premature saturation. Amplification level may be derived from the C_int value and duration of integration.

[0188] Further, in one embodiment, the output signal Vout may be converted into a digital signal by an ADC. For example, the charging signal current I_sig may be digitized using a single-slope integration ADC such that a counter may increment (counting a number) until the integrator output voltage crosses some threshold as defined by a comparator. When the single-slope integration ADC is used, calibration may be performed to determine an absolute value of the capacitor C_int 3312. Alternatively, a “dual-slope integrating ADC” may be used that uses a fixed integration period followed by a variable discharge period. In other embodiments, the output signal Vout may be converted into a digital signal by other known analog-to-digital conversion techniques.

[0189] In one embodiment, integration of current response of the chemically-sensitive transistor may provide a better signal-to-noise ratio (SNR) than measurement of instantaneous voltage output of a chemically-sensitive transistor.

[0190] In one or more embodiments, it may be hard to completely cancel the DC current of the chemical sensitive transistor using the current source 3308. Therefore, in one embodiment, the size of the capacitor may be limited to a certain size. In another embodiment, the duration of integration time may be limited to, for example, 1 μs. If the integration time is limited, a dual-slope ADC with a much slower discharge phase may be used to convert the output voltage Vout to a digital output.

[0191] FIG. 34 illustrates a block diagram of another chemical detection circuit 3400 according to an embodiment of the present invention. The chemical detection circuit 3400 may comprise a plurality of chemical detection pixels 3402.1-3402.N, a current source 3408, an amplifier 3410, a resistor 3424, an offset voltage V_set 3414, a capacitor C_int 3412, and three switches 3416, 3418 and 3420. Each chemical detection pixel (e.g., 3402.1, . . . , or 3402.N) may comprise a chemically-sensitive transistor (e.g., 3404.1, . . . , or 3404.N, respectively), a row-select switch (e.g., 3406.1, . . . , or 3406.N, respectively) and an output switch (e.g., 3422.1, . . . , 3422.N). The amplifier 3410 may have a first input terminal coupled to the output switches of the chemical detection pixels so that when a chemical detection pixel is selected, its output switch may be closed to generate an output signal for the first input terminal of the amplifier 3410. The amplifier 3410 may also have a second input terminal coupled to the offset voltage V_set 3414. The capacitor C_int 3412 may have a first side coupled to the first input terminal of the amplifier 3410 and a second side coupled to an output terminal of the amplifier 3410 via the switch 3418. The switch 3416 may be coupled between the first side of the capacitor C_int 3412 and the output terminal of the amplifier 3410. The switch 3420 may be coupled between the second side of the capacitor C_int 3412 and ground. In one embodiment, the plurality of chemical detection pixels 3402.1 to 3402.N may form a column of chemical detection pixels. The capacitor C_int 3412 may be configured as a negative feedback loop for the amplifier 3410 and thus, the capacitor C_int 3412 and amplifier 3410 may form an integrator for the column of chemical detection pixels. The integrator may be shared by all chemical detection pixels of the column and may be referred to as a column integrator.

[0192] Each chemically-sensitive transistor 3404.1˜3404.N may have a gate structure similar to that of the chemically-sensitive transistor 3302.1˜3302.N. Each chemically-sensitive transistor 3404.1˜3404.N may further have a first terminal connected to a first side of a respective row-select switch 3406.1˜3406.N and a first side of a respective output switch 3422.1˜3422.N. Each chemically-sensitive transistor 3404.1˜3404.N may also have a second terminal connected to ground. For example, as shown in FIG. 34, the transistor 3404.1 may be a PMOS with a first terminal (e.g., the source) connected to a first side of the row-select switch 3406.1 and a first side of the output switch 3422.1. Further, the transistor 3404.1 may also have a second side (e.g., the drain) connected to ground. Each row-select switch (e.g., 3406.1, . . . , or 3406.N) of each chemical detection pixel 3402.1˜3402.N may have a second side connected to the current source 3408. The second side of each output switch 3422.1˜3422.N may also be coupled to the first input of the amplifier 3410 via the resistor 3424.

[0193] In one embodiment, the chemical detection circuit 3400 may be configured that each of the chemically-sensitive transistors (e.g., 3404.1, . . . , or 3404.N) may work in a voltage-mode. That is, during operation, each of the chemically-sensitive transistors may work as a voltage amplifier. The ion concentration of analyte being measured by the chemically-sensitive transistor may be detected by a voltage level at the output. When one chemical detection pixel is selected, the corresponding row-select and output switches may be closed. The offset voltage V_set may set an appropriate voltage between the virtual ground (e.g., the first terminal or negative terminal) of the amplifier 3410 and the output of the selected chemically-sensitive transistor. For example, as shown in FIG. 34, the chemical detection pixel 3402.1 may be selected, and thus the row-select switch 3406.1 and output switch 3422.1 may be closed. The current source 3408 may provide a current I_ss to the selected chemically-sensitive transistor 3404.1. The resistor 3424 may be used to convert an output voltage at the selected chemically-sensitive transistor to a charging current I_is to be integrated onto the capacitor C_int 3412. An output signal of the amplifier 3410 may be read out as V_out.

[0194] Similar to operation of the chemical detection circuit 3300, the operation of the chemical detection circuit 3400 may have an integration phase and a clear phase to clear charges. During the integration phase, the switch 3420 may be closed and switches 3416 and 3420 may be left open. After the output signal Vout is read out, the operation may enter the second phase, during which the switches 3416 and 3420 may be closed and the switch 3418 may be left open to clear out the charges accumulated on the capacitor C_int 3412.

[0195] In one embodiment, each chemically-sensitive transistor 3404.1 to 3404.N may be an ion-sensitive field effect transistor (ISFET) and each row-select switch 3406.1 to 3406.N may be a transistor. Each output switch 3422.1 to 3422.N may also be a transistor.

[0196] Further, similar to the chemical detection circuit 3300, the output signal Vout of the chemical detection circuit 3400 may be converted into a digital signal by an ADC. For example, the charging current I_is may be digitized using a single-slope integration ADC or a dual-slope integrating ADC. In other embodiments, the output signal Vout of the chemical detection circuit 3400 may be converted into a digital signal by other known analog-to-digital (A/D) conversion techniques.

[0197] Moreover, in one embodiment, the resistance of the resistor 3424 may dominate over the resistance of the series row-select switch (e.g., 3406.1 to 3406.N) to limit the current to be integrated onto the capacitor C_int 3412.

[0198] FIG. 35 illustrates a block diagram of yet another chemical detection circuit 3500 according to yet another embodiment of the present invention. The chemical detection circuit 3500 may have a pass transistor 3524 that replaces the resistor 3424 of the chemical detection circuit 3400. Other than the transistor 3524, other parts of the chemical detection circuit 3500 may be identical to the chemical detection circuit 3400. The pass transistor 3524 may have a gate voltage V_bias tied to some process-, voltage-, and temperature (PVT) independent bias circuit. The on-resistance of this pass transistor 3524 may be designed to dominate over the resistance of the series row-select switch in the pixel.

[0199] FIG. 36 illustrates a process 3600 for generating an output of a chemical detection circuit according to an embodiment of the present invention. The process 3600 may be performed by the chemical detection circuits 3300, 3400 and 3500 as described above with respect to FIGS. 33-35. The process 3600 may start at step 3602, at which a chemical detection pixel may be selected for readout. As shown in FIG. 33, for example, when the chemical detection pixel 3302.1 is selected, the row-select switch 3306.1 may be closed. Alternatively, as shown in FIGS. 34 and 35, for example, when the chemical detection pixel 3402.1 is selected, the row-select switch 3406.1 and the output switch 3422.1 may be closed.

[0200] Then the process 3600 may proceed to step 3604. At step 3604, the process 3600 may integrate a readout current from the chemical detection pixel to an integrator. As described above, the readout current may be caused by a voltage change at a gate terminal of the selected chemical detection pixel. In one embodiment, as shown in FIG. 33, the chemical detection pixel may work in a current mode, the selected chemical detection pixel may supply a readout current to charge a capacitor of the integrator. In another embodiment, as shown in FIGS. 34 and 35, the chemical detection pixel may work in a voltage mode and an output voltage from the chemical detection pixel may be converted by a resistor or a pass transistor to a current to charge the capacitor of the integrator.

[0201] Then, at step 3506, the process 3600 may read out an output voltage of the integrator. As described above, output voltage of the integrator (Vout at the output of the amplifier 3310 or output of the amplifier 3410) may have a better signal-to-noise ratio (SNR) for detection of ion concentration of the analyte being analyzed by the chemical detection pixel than instantaneous voltage measurement.

[0202] Although in the above description, the chemically-sensitive transistors may be described as PMOS devices, they may also be implemented as NMOS devices in one embodiment. Further, the switches (e.g., row-selected switches, output switches, charge clear switches) may be implemented in either PMOS or NMOS transistors in an embodiment.

[0203] The described embodiments may provide a chemical detection circuit that may comprise a plurality of first output circuits at a first side and a plurality of second output circuits at a second and opposite side of the chemical detection circuit. The chemical detection circuit may further comprise a plurality of tiles of pixels each placed between respective pairs of first and second output circuits. Each tile array may include four quadrants of pixels. Each quadrant may have columns with designated first columns interleaved with second columns. Each first column may be connected to a respective first output circuit in first and second quadrants, and to a respective second output circuit in third and fourth quadrants. Each second column may be connected to a respective second output circuit in first and second quadrants, and to a respective first output circuit in third and fourth quadrants.

[0204] Some embodiments may also provide a chemical detection system that may comprise a motherboard having at least one central processing unit, an output device coupled to the mother board, and a chemical detection reader board connected to the mother board. The chemical detection reader board may have a chemical detection circuit that may comprise a plurality of first output circuits at a first side and a plurality of second output circuits at a second and opposite side of the chemical detection circuit. The chemical detection circuit may further comprise a plurality of tiles of pixels each placed between respective pairs of first and second output circuits. Each tile array may include four quadrants of pixels. Each quadrant may have columns with designated first columns interleaved with second columns. Each first column may be connected to a respective first output circuit in first and second quadrants, and to a respective second output circuit in third and fourth quadrants. Each second column may be connected to a respective second output circuit in first and second quadrants, and to a respective first output circuit in third and fourth quadrants.

[0205] Other embodiments may provide a method to read out data from a chemical detection circuit. The method may comprise selecting a first quadrant of a tile to read out data, selecting one group of first columns and one group of second columns, reading out data of the group of first columns from a first set of output pins located at a first side of the chemical detection circuit, reading out data of the group of second columns from a second set of output pins located at a second side of the chemical detection circuit, and repeating selection and data readouts for next groups of first columns and second columns till all remaining columns of the first quadrant are read out.

[0206] FIG. 37 illustrates a block diagram of a chemical detection circuit 3700 according to an embodiment of the present invention. The chemical detection circuit 3700 may comprise a plurality of tiles of pixels 3702.1-3702.N and 3704.1-3704.N, output circuits 3706 and 3708, control logic and digital interface 3710, and bias circuit and diagnostic output logic 3712. Each tile 3702.1-3702.N and 3704.1-3704.N may include pixels formed in columns with each column containing many rows. For example, each tile may contain 6848 columns×11136 rows of pixels. The tiles 3702.1-3702.N may form a slice (e.g., a top slice) and the tiles 3704.1-3704.N may form another slice (e.g., a bottom slice). The plurality of tiles 3702.1-3702.N and 3704.1-3704.N may form a conglomerate pixel array. The output circuits 3706 and 3708 may be placed at two opposite sides of the tiles (e.g., top and bottom). The output circuits 3706 and 3708, control logic and digital interface 3710, and bias circuit and diagnostic output logic 3712 may each contain a plurality of pins for input and output data for the chemical detection circuit 3700. In one embodiment, the chemical detection circuit 3700 may be formed on an integrated circuit chip. Further, in one embodiment, the output circuits 3706 and 3708 may include analog-to-digital converters (ADCs) to generate digital outputs. Moreover, in one embodiment, two slices may be operated independently and exposed to a different analyte. For example, while data is being read out for the top tile, the bottom tile may be flushed out of fluid for another round of test. This may be used in conjunction with a dual-channel flow cell (e.g., mounted on top of the chemical detection circuit 3700) that two different flow channels may carry out different tasks at the same time.

[0207] Pixels of each tile 3702.1-3702.N and 3704.1-3704.N may be divided into four quadrants and data generated at each pixel may be read out from either the top or the bottom. An exemplary configuration of the pixels within a pair of tiles is shown in FIG. 38A.

[0208] FIG. 38A illustrates a block diagram 3800 of components of the chemical detection circuit 3700 (of FIG. 37) according to an embodiment of the present invention. As shown in FIG. 38A, the tile 3702.1 may comprise four quadrants: top left (TL) quadrant 3802, top right (TR) quadrant 3804, bottom left (BL) quadrant 3806, bottom right (BR) quadrant 3808; and four row select registers: top left row select register 3828, top right row select register 3836, bottom left row select register 3830, bottom right row select register 3838. The tile 3704.1 may comprise four quadrants: top left quadrant 3816, top right quadrant 3818, bottom left quadrant 3812, bottom right quadrant 3814; and four row select registers each for a respective quadrant: top left row select register 3832, top right row select register 3844, bottom left row select register 3834, bottom right row select register 3846. The tiles 3702.1 and 3704.1 may share a current sources and swizzles block 3810. The current sources and swizzles block 3810 may be sandwiched between the pair of tiles. Further, the tiles 3702.1 and 3704.1 may share top and bottom output circuits including channel circuits 3820 and 3822, column multiplexers 3848 and 3850, output multiplexers 3824 and 3826, and output buffers 3840 and 3842. The channel circuits 3820 and 3822 may include sample and hold (S/H) circuits. In one embodiment, each quadrant of the tile may comprise a plurality of columns that each may include a plurality of rows. For example, a quadrant may have 1712 columns that each may contain 2784 rows of pixels. In one embodiment, each tile may include reference pixels. For example, a predetermined number (e.g., 4) of columns and or rows of pixels at outer peripheral of each tile may be designated as reference pixels. The reference pixels may be used to generate signals representing the background and are not exposed to the analyte.

[0209] Each column may generate an output signal when one row of pixels is selected according to the respective row select register for the quadrant. In one embodiment, each column of a quadrant may be designated as a first or second column (e.g., an odd column or even column), and the output signal may be read from either the top or the bottom output circuits. The columns may be grouped for parallel read out operation. That is, a group of first columns or a group of second columns (n columns, n being an integer larger than one) may be read out together simultaneously in parallel. For example, if n is equal to 8, odd column groups may be columns [1:8], [17:24], [33:40], etc., and even column groups may be [9:16], [25:32], [41:48], etc. The column groups may be connected according to quadrant they are in. For a top left quadrant (e.g., 3802, 3812) odd column groups may be connected to the output circuit at a first side (e.g., top output circuits including the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and output buffer 3840) and even column groups may be connected to the output circuit at a second side (e.g., bottom output circuits including the channel circuit 3822, column multiplexer 3850, output multiplexer 3826 and output buffer 3842). For a top right quadrant (e.g., 3804, 3814) odd column groups may be connected to the output circuit at the first side and even column groups may be connected to the output circuit at the second side. For a bottom left quadrant (e.g., 3806, 3816) odd column groups may be connected to the output circuit at the second side and even column groups may be connected to output circuit at the first side. For a bottom right quadrant (e.g., 3808, 3818) odd column groups may be connected to the output circuit at the second side and even column groups may be connected to the output circuit at the first side.

[0210] In one embodiment, a group of first columns and a group of second columns may form a data channel to be read out together simultaneously from either the first or the second side of the output circuits. Each data channel may comprise one group of first columns and one group of second columns located in each quadrant between a first side of output circuits and a second side of output circuits (e.g., n first columns and n second columns from each of TL 3802, BL 3806, BL 3816, TL 3812).

[0211] The readout operation may use the row select shift registers (e.g., a vertical shift register) to select rows and column shift registers to select columns (e.g., a horizontal shift register). When the operation starts, switches inside the current sources and swizzles block 3810 may be enabled to provide driving currents to the signal lines, any pixel select lines of an unused flow cell may be disabled, all row and column shift registers may be reset. Then, the vertical shift register may start counting by increments of 1 and the horizontal shift register may start counting by 16. Data for the frame may start with the vertical shift registers selecting row 1 of TL, TR, BL, and BR since reset. In one embodiment, the word “swizzle” may refer to the configuration that a metal line which passes through one column in the top circuitry may be routed in the space between the top and bottom circuitry in such a way that it passes through a different column in the bottom circuitry, as shown below in FIG. 40.

[0212] The readout operation may start with a TL quadrant (e.g., the TL quadrant 3802). The first group of odd columns (e.g., n first columns [1:8]) in TL 3802 and first group of even columns (e.g., n second columns [9:16]) in TL 3802 may be selected by the horizontal shift registers. Selected odd column pixels in row 1 of TL 3802 may be routed to top outputs through the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and then they may be read out through the top output buffer 3840. At the same time, the selected even column pixels in row 1 of TL may be routed to bottom outputs through the channel circuit 3822, column multiplexer 3850, output multiplexer 3826 and then they may read out through the bottom output buffer 3842.

[0213] During the above readout time, the next group of odd pixels (e.g., columns [17:24]) in row 1 of TL 3802 may be connected to the top column multiplexer 3848 and output multiplexer 3824 via the channel circuit 3820. Similarly, the next group of even pixels (e.g., columns [25:32]) in row 1 of TL 3802 may be connected to the bottom column multiplexer 3850 and output multiplexer 3826 via the channel circuit 3822. Then, the top and bottom output multiplexers 3824 and 3826 may switch their respective multiplexers, and subsequently, the next group of odd pixels in row 1 of TL 3802 may be read out through the top outputs, and the next group of even pixels in row 1 of TL 3802 may be read out through the bottom outputs. This may continue until all pixels from row 1 of TL 3802 have been read out. At the end of the read out of row 1, the TL 3802's vertical shift register may shift to the next row, and the outputs may begin to settle.

[0214] After row 1 of the TL quadrant 3802 is finished, the readout operation may continue to TR quadrant 3804. A first group of odd columns (e.g., columns [1:8]) in TR quadrant 3804 and a first group of even columns (e.g., columns [9:16]) in TR quadrant 3804 may be selected by the horizontal shift registers. Then the selected first group of odd column pixels in row 1 of TR quadrant 3804 may be routed to top outputs through the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and then they may be read out through the top output buffer 3840. At the same time, the selected first group of even column pixels [9:16] in row 1 of TR 3804 may be routed to bottom outputs through the channel circuit 3822, column multiplexer 3850, output multiplexer 3826 and then they may read out through the bottom output buffer 3842.

[0215] During the above readout time, the next group of odd pixels (e.g., columns [17:24]) in row 1 of TR 3804 may be connected to the top column multiplexer 3848 and output multiplexer 3824 via the channel circuit 3820. Similarly, the next group of even pixels (e.g., columns [25:32]) in row 1 of TR 3804 may be connected to the bottom column multiplexer 3850 and output multiplexer 3826 via the channel circuit 3822. Then, the top and bottom output multiplexers 3824 and 3826 may switch their respective multiplexers, and subsequently, the next group of odd pixels in row 1 of TR 3804 may be read out through the top outputs, and the next group of even pixels in row 1 of TR 3804 may be read out through the bottom outputs. This may continue until all pixels from row 1 of TR 3804 have been read out. At the end of the read out of row 1, the TR 3804's vertical shift register may shift to the next row, and the outputs may begin to settle.

[0216] After row 1 of the TR quadrant 3804 is finished, the readout operation may continue to BL quadrant 3806. A first group of odd columns (e.g., columns [1:8]) in BL quadrant 3806 and a first group of even columns (e.g., columns [9:16]) in BL quadrant 3806 may be selected by the horizontal shift registers. Then the selected first group of odd column pixels in row 1 of BL quadrant 3806 may be routed to bottom outputs through the channel circuit 3822, column multiplexer 3850, output multiplexer 3826 and then they may read out through the bottom output buffer 3842. At the same time, the selected first group of even column pixels [9:16] in row 1 of BL 3806 may be routed to top outputs through the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and then they may be read out through the top output buffer 3840.

[0217] During the above readout time, the next group of odd pixels (e.g., columns [17:24]) in row 1 of BL 3806 may be connected to the bottom column multiplexer 3850 and output multiplexer 3826 via the channel circuit 3822. Similarly, the next group of even pixels (e.g., columns [25:32]) in row 1 of BL 3806 may be connected to the top column multiplexer 3848 and output multiplexer 3824 via the channel circuit 3820. Then, the top and bottom output multiplexers 3824 and 3826 may switch their respective multiplexers, and subsequently, the next group of odd pixels in row 1 of BL 3806 may be read out through the bottom outputs, and the next group of even pixels in row 1 of BL 3806 may be read out through the top outputs. This may continue until all pixels from row 1 of BL 3806 have been read out. At the end of the read out of row 1, the BL 3806's vertical shift register may shift to the next row, and the outputs may begin to settle.

[0218] After row 1 of the BL quadrant 3806 is finished, the readout operation may continue to BR quadrant 3808. A first group of odd columns (e.g., columns [1:8]) in BR quadrant 3808 and a first group of even columns (e.g., columns [9:16]) in BR quadrant 3808 may be selected by the horizontal shift registers. Then the selected first group of odd column pixels in row 1 of BR quadrant 3808 may be routed to bottom outputs through the channel circuit 3822, column multiplexer 3850, output multiplexer 3826 and then they may read out through the bottom output buffer 3842. At the same time, the selected first group of even column pixels [9:16] in row 1 of BR quadrant 3808 may be routed to top outputs through the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and then they may be read out through the top output buffer 3840.

[0219] During the above readout time, the next group of odd pixels (e.g., columns [17:24]) in row 1 of BR quadrant 3808 may be connected to the bottom column multiplexer 3850 and output multiplexer 3826 via the channel circuit 3822. Similarly, the next group of even pixels (e.g., columns [25:32]) in row 1 of BR quadrant 3808 may be connected to the top column multiplexer 3848 and output multiplexer 3824 via the channel circuit 3820. Then, the top and bottom output multiplexers 3824 and 3826 may switch their respective multiplexers, and subsequently, the next group of odd pixels in row 1 of BR quadrant 3808 may be read out through the bottom outputs, and the next group of even pixels in row 1 of BR quadrant 3808 may be read out through the top outputs. This may continue until all pixels from row 1 of BR quadrant 3808 have been read out. At the end of the read out of row 1, the BR quadrant 3808's vertical shift register may shift to the next row, and the outputs may begin to settle.

[0220] After row 1 of all four quadrants are read out, the operation may return to TL, and the pattern may be repeated until all rows in TL, TR, BL, and BR are read out to complete one frame for a tile (e.g., 3702.1). And then, the operation may be carried on in a next tile (e.g., 3702.2). This scheme may allow row n in a quadrant to settle for ¾ of the time that it takes to read out row n from all four quadrants.

[0221] In one embodiment, the readout operation may be performed to complete one quadrant at a time. That is, after one row for a quadrant has finished, move on to the next row of the same quadrant; and continue to a next quadrant only after all rows of same the quadrant are finished.

[0222] In one embodiment, the tiles at the top slice (e.g., 3702.1˜3702.N) may operate concurrently, and the tiles at the bottom slice may operate alternately with corresponding tiles of the top slice (e.g., 3702.1 and 3704.1 would operate alternately.)

[0223] FIG. 38B illustrates shift directions in different quadrants of a tile according to an embodiment of the present invention. As shown in FIG. 38B, in one embodiment, the readout operation may start from the center of a tile and move outward (e.g., increment the row and column select registers).

[0224] FIG. 39 illustrates a block diagram of part of a data channel 3900 of the chemical detection circuit 3700 according to an embodiment of the present invention. The data channel 3900 may comprise n first columns 3902 and n second columns 3904 of TL quadrant 3802, and n first columns 3908 and n second columns 3906 of BL quadrant 3806. FIG. 39 only shows the data channel 3900 in the top slice (e.g., TL quadrant 3802 and BL quadrant 3806 of the tile 3702.1). Although not shown, the data channel 3900 may further comprise n first columns and n second columns in each of TL quadrant 3812 and BL quadrant 3816 of the tile 3704.1. As shown in FIG. 39, the data channel 3900 may have two 2n signal lines with top 2n signal lines connected to the output channel circuit 3820 at the right side. Although not shown, the bottom 2n signal lines may be connected to the output channel circuit 3822 to the left side. The pixels of the n first columns 3902 and the n second columns 3906 may each be connected to a respective top 2n signal lines. The pixels of the n second columns 3904 and the n first columns 3908 may each be connected to a respective bottom 2n signal lines. The current sources and swizzles block 3810 at the left side of the 2n signal lines may provide 2n current sources that each may drive a respective signal line. Further, the 2n signal lines may be swizzled in the current sources and swizzles block 3810 (details of one exemplary embodiment of the swizzle will be described later with respect to FIG. 40).

[0225] During operation, data from the top 2n signal lines may be read out from the channel circuit 3820, column multiplexer 3848, output multiplexer 3824 and output buffer 3840 and the bottom 2n signal lines may be readout from corresponding circuits at the left.

[0226] FIG. 40 illustrates a swizzle configuration of signal lines of the chemical detection circuit 3700 (of FIG. 37) according to an embodiment of the present invention. In one embodiment of the chemical detection circuit 3700, there may be two output lines running through each column so that a column of pixels may be connected to the column circuitry at the top of the chemical detection circuit 3700 (e.g., the IC chip) or to the column circuitry at the bottom of the chemical detection circuit 3700. The column output lines may run the full height of the die and may be very long, and therefore may be susceptible to crosstalk. To reduce crosstalk, the column output lines may be swizzled in the middle of the chemical detection circuit 3700 (e.g., in the current sources and swizzles block 3810 of FIG. 38A). As shown in FIG. 40, the four columns may have 8 wires (e.g., each column may contain two wires). Each wire may be connected to either the top column circuitry 4002 or the bottom column circuitry 4004. For example, the wires A, D, E and H may be connected to the top column circuitry 4002 and wires B, C, F and G may be connected to the bottom column circuitry 4004. The sequence of the 8 wires may be swizzled in the middle. For example, top half of wire A may run through pixels of column 1 and bottom half of wire A may run through pixels of column 2, top half of wire B may run through pixels of column 1 and bottom half of wire B may run through pixels of column 3, top half of wire C may run through pixels of column 2 and bottom half of wire C may run through pixels of column 1, top half of wire D may run through pixels of column 2 and bottom half of wire D may run through pixels of column 4, top half of wire E may run through pixels of column 3 and bottom half of wire E may run through pixels of column 1, top half of wire F may run through pixels of column 3 and bottom half of wire F may run through pixels of column 4, top half of wire G may run through pixels of column 4 and bottom half of wire G may run through pixels of column 2, top half of wire H may run through pixels of column 4 and bottom half of wire H may run through pixels of column 3. In one embodiment, the swizzle according to pattern shown in FIG. 40 may be repeated for every four columns. As a result, the crosstalk may be reduced by as much as 50%.

[0227] FIG. 41 illustrates a process 4100 for outputting data from a chemical detection circuit according to an embodiment of the present invention. The process 4100 may be performed by the chemical detection circuit 3700. The process 4100 may start at step 4102, at which a first quadrant of a tile may be selected to read out data. As described above with respect to FIG. 38A, for example, a readout operation may be performed for a tile (e.g., 3702.1) by starting at the top left quadrant 3802. Then the process 4100 may proceed to step 4104. At step 4104, the process 4100 may select one group of first columns and one group of second columns. As described above, the readout operation may be performed in groups of first columns and second columns (e.g., odd columns [1:8] and even columns [9:16]). Then, at step 4106, the process 4100 may read out data for the group of first columns from a first set of output pins (e.g., top output buffers 3840) and for the group of second columns from a second set of output pins (e.g., bottom output buffers 3842).

[0228] Then the process 4100 may proceed to step 4108. At step 4108, the process 4100 may repeat selection and data readouts for a next group of first columns and a next group of second columns until all remaining columns of the first quadrant are read out. For example, the chemical detection circuit 3700 may repeat the readout operation for odd column groups [17:24], [33:40], etc. and even column groups [25:32], [41:48], etc. for all remaining columns of the first quadrant (e.g., TL quadrant 3802).

[0229] FIG. 42 illustrates a system architecture 4200 for chemical detection according to an embodiment of the present invention. The system architecture 4200 may comprise a motherboard 4202, an output device 4208, a reader board 4210 and a valve board 4212. The motherboard 4202 may include CPUs 4204 and storage 4206 (e.g., a Double Date Rate (DDR) memory device). The CPUs 4204 may scale from 2 cores to 6 cores. The memory DDRs 4206 may be 1 GB to 96 GB DDR3 (double data rate type 3). The motherboard 4202 may also support on board RAID (6 SATA ports) and graphics processor board (GPU). The output device 4208 may be a color display with high brightness (e.g., an XGA multi-touch input independent and display independent 8 wire analog resistive). The reader board 4210 may include a sensor 4218 (e.g., the chemical detection circuit 3700) and other peripheral circuits (details shown in FIGS. 43 and 44 for analog and digital chemical sensors respectively). The valve board 4212 may include a FPGA 4214 and valve controls 4216. During operation, the FPGA 4214 may be loaded with control logic to control the operation of the valve board. The valve controls 4216 may include a plurality of valves (e.g., 30 valves) that controls flow of fluid containing analyte to be analyzed by the sensor 4218. The valve board 4212 may further include thermistor inputs, pressure sensor inputs and may further include heater/cooler controls (not shown) that may control heaters/coolers for the sensor 4218 and analytes (e.g., to assist in controlling reactions during the testing of samples). In one embodiment, the motherboard 4202 and the reader board 4210 may be connected according to the PCI express (PCIe) standard, the reader board 4210 and the valve board 4212 may be connected by serial link over LVDS (low-voltage differential signaling).

[0230] FIG. 43 illustrates an analog reader board 4300 for a chemical detection circuit according to an embodiment of the present invention. The analog reader board 4300 may include an analog chemical sensor 4302, a clock 4304, a power supply 4306, a serial link for LVDS 4308, a reader FPGA 4310, a memory 4312, ADCs 4314, a PCIe switch 4316, a PCIe connector 4318, two satellite FPGA blocks 4320 and 4322, and a voltage reference and DACs block 4324. The analog chemical sensor 4302 may be an IC chip embodiment of chemical detection circuit 3700. The analog data read out from the analog chemical sensor 4302 may be digitized by the ADCs 4314, which may use voltage references and DACs 4324. The digitized data may be sent to the satellite FPGAs 4320 or 4322, which may perform settling correction, and then sent to the reader FPGA 4310. The reader FPGA 4310 may buffer data in the memory 4312, which may include a plurality of DDR memory blocks. The reader FPGA 4310 may also perform frame averaging (e.g., average a pixel's data value among multiple frames; this is possible because the analog chemical sensor may read out data at a frame rate higher than required (e.g., 30 FPS) or variable rate frame averaging (e.g., average different portions of a pixels time history at a different rate) and then send data out to a server motherboard (e.g., motherboard 4202 of FIG. 42) via PCIe switch 4316 and PCIe connector 4318. The PCIe switch 4316 may include multiplexers that multiplex links to a PCIex 16 link of the PCIe connector 4318. The LVDS 4308 may provide serial links to a valve board (e.g., the valve board 4212 of FIG. 42). The power of the reader board 4300 may be provided by the power supply 4306 and the timing signals may be provided by the clock 4304. In one embodiment, the ADCs 4314 may be placed close to the analog chemical sensor 4302.

[0231] FIG. 44 illustrates a digital reader board 4400 for a chemical detection circuit according to an embodiment of the present invention. The digital reader board 4400 may include a digital chemical sensor 4402, a clock 4404, a power supply 4406, a serial link for LVDS 4408, a reader FPGA 4410, a memory 4412, a PCIe switch 4416, and a PCIe connector 4418. The digital chemical sensor 4402 may be an IC chip embodiment of chemical detection circuit 3700 (of FIG. 37) with ADCs incorporated on the chip that digitize the output data signals on-chip. The clock 4404, power supply 4406, serial link for LVDS 4408, reader FPGA 4410, memory 4412, PCIe switch 4416, and PCIe connector 4418 may perform functions similar to their counterparts on the analog digital reader board 4300. In one embodiment, the digital chemical sensor 4402 may be placed on a replaceable board separate from the digital reader board 4400.

[0232] FIG. 45 illustrates a block diagram 4500 of an output configuration for a chemical detection circuit according to an embodiment of the present invention. The block diagram 4500 may show an analog front end and noise calculations for analog data output from an analog chemical detector 4502. The DAC 4504 may generate analog signals according to digital reference values, and analog signals from the DAC 4504 may be buffered by the buffers 4506.1˜4506.4. The output from analog chemical detector 4502 may be amplified by the amplifiers 4508.1˜4508.4 and the amplified signals may be filtered by the low pass filters 4510.1˜4510.4. The filtered signals may be input to the ADC module 4512, which may contain a plurality of differential amplifiers 4514.1˜4514.4. The amplified signals may pass another round of low pass filters 4516.1˜4516.4 and then finally the signals may be converted by the Quad ADC 4518 into digital data and sent to FPGAs. The Quad ADC 4518 may receive clock signals from a clock fanout 4524. The clock signals may be generated by a PLL 4522 based on signals from an oscillator 4520. In one embodiment, the analog chemical detector 4502 may be an IC chip embodiment of the chemical detection circuit 3700.

[0233] FIG. 46 illustrates a block diagram 4600 of bandwidth utilization for a chemical detection circuit according to an embodiment of the present invention. An analog chemical detector 4602 may send its data to a plurality of ADCs 4604. The ADCs 4604 may send digital data to the FPGA(s) 4606, which may in turn send data to CPU(s) 4608 and storage units 4616 and 4618 (e.g., DDR3 memory). The CPU(s) 4608 may cache data in a memory cache 4610 (e.g., DDR3 memory) and hard drives 4612 and 4614. In one embodiment, the analog chemical detector 4602 may be an IC chip embodiment of the chemical detection circuit 3700. The numbers given in FIG. 46 may be theoretical maximums. The FPGA(s) 4606 may perform a 3:1 compression of samples (e.g., settling correction) and a 2:1 or greater compression of frames (e.g. frame averaging).

[0234] FIG. 47 illustrates a block diagram 4700 for clock distribution for an analog reader board (e.g., the analog reader board 4300 of FIG. 43) according to an embodiment of the present invention. The clock signals for various components of an analog reader board may be generated based on a 100 MHz oscillator 4702. The clock generator 4704 may receive the signals from the 100 MHz oscillator 4702 and generate various clock signals. For example, the clock generator 4704 may generate 120 MHz clock signals to be sent to two zero delay buffers 4706.1 and 4706.2, a flip flop 4716 and a first PLL of the FPGA 4714 (e.g., the reader FPGA 4310 of the analog reader board 4300). The zero delay buffers 4706.1 and 4706.2 may provide the 120 MHz clock signals to ADC sets 4718.1 and 4718.2 for the ADCs to send digitized data to FPGAs 4720.1 and 4720.2 (e.g., the satellite FPGAs 4320 and 4322 of the analog reader board 4300) at a frequency of, for example, 840 MHz. The first PLL of the FPGA 4714 may send clock signals to first PLLs in respective FPGAs 4720.1 and 4720.2, and may also send a clock signal internally to a flip flop port of the FPGA 4714. The flip flop 4716 may generate channel increment/decrement signals for data read out based on an output of the flip flop port of the FPGA 4714 and the 120 MHz clock signal from the clock generator 4704. The oscillator 4702 may also generate a 33 MHz clock signal for a clock driver 4708, which may provide clock signals for second PLLs in the FPGAs 4720.1 and 4720.2 respectively and for a second PLL in the FPGA 4714. The second PLL in the FPGA 4714 may generate an internal clock (e.g., 267 MHz). In one embodiment, basing all of these clocks may allow for the synchronization of the channel outputs of the sensor and the sampling by the ADCs.

[0235] Communication external to the analog board reader may be based on a 100 MHz clock signals from a PCIe connector. The 100 MHz clock signals from PCIe connector may be buffered by a PCIe clock buffer 4710. The buffered 100 MHz clock signals may be sent to first and second SerDes PLL (serialization/deserialization phase locked loop) of the FPGA 4714 and may also be sent to a PCIe switch 4712 (e.g., PCIe switch 4316).

[0236] In one embodiment, the zero delay buffers 4706.1 and 4706.2 may allow for skew adjustment between ADC sample clocks and data channels of a chemical detector (not shown). The clocks may be differential LVDS where possible, but the clocks for channels of the chemical detector may be differential low-voltage positive emitter-coupled logic (LVPECL). Further, in one embodiment, the combination of ADC and data channel 120 MHz clocks need to be low jitter (e.g., <15 ps—as drawn ˜4.5 rms). In one embodiment, in the case of the analog chemical sensor, the “clock” provided to the sensor may be the “channel increment/decrement signal”—allowing for synchronization of channel switch and sampling by the ADCs.

[0237] FIG. 48 illustrates a block diagram 4800 for power distribution of system components according to an embodiment of the present invention. A PC power supply 4802 may be coupled to an AC input. The valve board 4804 (e.g., the valve board 4212 of the system architecture 4200) may receive power from the PC power supply 4802 by two 4-pin connectors. The motherboard 4806 (e.g., the motherboard 4202 of the system architecture 4200) may receive power from the PC power supply 4802 by two 8-pin connectors and a 24-pin connector. The 24-pin cable to provide power to the motherboard 4806 may be “Y-cabled” to also provide power to the reader board 4808 (e.g., the reader board 4210 of system architecture 4200, which may be an analog reader board (e.g., 4300) or digital reader board (e.g., 4400)). The reader board 4808 may include an onboard power supply 4810 that may include a plurality of power regulators (e.g., low dropout linear regulator, programmable output low dropout regular) and/or DC/DC power supplies (e.g., high voltage high current DC/DC power supply). In one embodiment, all of the DC/DC switching power supplies may be synchronized with the main reader board clock. This may keep any switching noise from the power supplies from “beating” against the clocks used elsewhere. Further, the clocks for the switching power supplies may be arranged in time such that the instantaneous current load on the PC power supply is minimized.

[0238] FIG. 49 illustrates a block diagram 4900 for DACs of an analog reader board according to an embodiment of the present invention. The DAC configuration 4900 may include a voltage reference 4902, a DAC 4904, a low pass filter including a resistor 4906 and a capacitor 4908. The filtered signal may be amplified by an operational amplifier 4910. The output from the operational amplifier 4910 may be filtered by a bead 4912 and a plurality of capacitors 4914 (e.g., a bulk LPF/LF charge). The filtered signal then may be sent to first inputs of the operational amplifiers 4916.1˜4916.4 via local decoupling circuits (e.g., the capacitors 4918 and corresponding resistors 4920). The second inputs to the operational amplifiers 4916.1˜4916.4 may be respective channel inputs 4924.1˜4924.4. The output of the operational amplifiers 4916.1˜4916.4 may be coupled back to the first inputs via respective feedback resistors 4922.1˜4922.4. In one embodiment, a plurality of DACs may be provided for channel offset, references voltages, electrode drive, built-in self-test (BIST) drive and ISFET Bias.

[0239] FIG. 50 illustrates a block diagram of FPGA configuration for an analog reader board 5000 according to an embodiment of the present invention. The analog reader board 5000 may be an embodiment of the analog reader board 4300. The analog reader board 5000 may comprise a plurality of ADC modules 5002.1 and 5002.2, two satellite FPGAs 5004.1 and 5004.2 (e.g., satellite FPGAs 4320 and 4322), a reader FPGA 5006 (e.g., reader FPGA 4310) and its memory modules 5008.1 and 5008.2, a PCIe switch 5010 (e.g., PCIe switch 4316), an analog chemical sensor 5012 (e.g., analog chemical sensor 4302), LVDS drivers and receivers 5014 (e.g., LVDS 4308) and a plurality of DAC modules 5016 and 5018. In one embodiment, each ADC of the ADC modules 5002.1 and 5002.2 may include a PLL. The satellite FPGAs 5004.1 and 5004.2 may perform sample averaging, which may be controlled by Dstrobe signal from the reader FPGA 5006 and software setup in the satellite FPGAs 5004.1 and 5004.2. Moreover, in one embodiment, the control logic for the reader FPGA 5006 may be loaded from the memory modules 5008.1 and/or 5008.2, which may be SPI flash devices (e.g., EEPROM) that hold two images: (1) a default “loader” image, and (2) a current run-time image. The control logic (e.g., images) for the satellite FPGAs 5004.1 and 5004.2 may be loaded by the reader FPGA 5006 from a motherboard (e.g., motherboard 4202) over PCIe. Further, in one embodiment, the FPGAs (including the valve FPGA 4214) may be de-configured and reloaded by a PCIe reset. Also, PCI enumerations may be trigged once PCIe FPGA may be programmed.

[0240] FIG. 51 illustrates a block diagram of FPGA power monitoring for a reader board 5100 according to an embodiment of the present invention. The reader board 5100 may be an embodiment of the reader board 4808. The reader board 5100 may comprise a power supply module 5102 (e.g., power supply 4810) that receives 5V and 12 V inputs from a 24 pin connector 5106. The power supply module 5102 may provide power to the rest of reader 5104. The output voltages from the power supply module 5102 may be monitored by a plurality of voltage monitors 5108 and 5112. The first voltage monitor 5108 may receive its power supply VCC from the 24 pin connector 5106 and generate a reset RST signal if one of the monitored voltage (including VCC) deviates more than a threshold from a predetermined voltage level (e.g., 1.5% deviation). The RST signal from the voltage monitor 5108 may be “OR”-ed with a reset signal from a PCIe connector 5110 by an OR gate 5116. The output of the OR gate 5116 may be input as a monitored voltage for the second voltage monitor 5112, which may also receive its power supply VCC from the 24 pin connector 5106 and generate a reset RST signal if one of the monitored voltage (including VCC) deviates more than a threshold from a predetermined voltage level (e.g., 1.5% deviation). The RST signal from the voltage monitor 5112 may be sent to a clock driver 5114, which may generate a reset signal and nCONFIG signal to the rest of reader 5104. The reset signal from the clock driver 5114 may be sent to a PCIe switch of the reader board. The nCONFIG signal from the clock driver 5114 may be sent to a FPGA of the reader board to cause the FPGA to reload. In one embodiment, “OR”-ing the RST signal from the voltage monitor 5108 with the reset signal from the PCIe connector 5110 may guarantee the nCONFIG pulse width requirements being met.

[0241] The described embodiments may provide a chemical detection circuit that may comprise a column of chemically-sensitive pixels. Each chemically-sensitive pixel may comprise a chemically-sensitive transistor and a row selection device. The chemical detection circuit may further comprise a column interface circuit coupled to the column of chemically-sensitive pixels and an analog-to-digital converter (ADC) coupled to the column interface circuit.

[0242] Some embodiments may also provide a chemical sensor that may comprise a plurality of columns of chemically-sensitive pixels. Each column may comprise a plurality of chemically-sensitive pixels formed in rows. Each chemically-sensitive pixel may comprise a chemically-sensitive transistor and a row selection device. The chemical sensor may further comprise a column interface circuit coupled to the column of chemically-sensitive pixels and an analog-to-digital converter (ADC) coupled to the column interface circuit.

[0243] Other embodiments may provide a method of generating an output signal for a chemical detection circuit. The method may comprise generating a row selection signal by a row decoder of the chemical detection circuit. The chemical detection circuit may have a pixel array that includes a column of chemical detection pixels. Each chemical detection pixel may include a chemically-sensitive transistor and a row selection device. The method may further comprise applying the row selection signal to a respective row selection device of a selected chemical detection pixel, converting an analog signal at a readout signal line of the column of chemical detection pixels to a digital signal by an Analog-to-Digital converter (ADC) and outputting the converted digital signal as the output signal for the chemical detection circuit.

[0244] The described embodiments may further provide a chemical detection circuit that may comprise a pixel array comprising a plurality of chemically-sensitive pixels formed in columns and rows. Each chemically-sensitive pixel may comprise a chemically-sensitive transistor and a row selection device. The chemical detection circuit may further comprise a pair of analog-to-digital converter (ADC) circuit blocks, a pair of input/output (I/O) circuit blocks coupled to the pair of ADC circuit blocks respectively and a plurality of serial link terminals coupled to the pair of IO circuit blocks.

[0245] The described embodiments may further provide a method to read out data from a chemical detection device. The method may comprise reading data from a plurality of columns of chemically-sensitive pixels on the chemical detection device in parallel. Each chemically-sensitive pixel may comprise a chemically-sensitive transistor and a row selection device. The method may further comprise digitizing data read from the plurality of columns of chemically-sensitive pixels in parallel, serializing the digitized data in parallel for each column of chemically-sensitive pixels respectively, and transmitting the buffered digitized data on a plurality of serial links in parallel.

[0246] FIG. 52 illustrates a digital chemical detection circuit 5200 according to an embodiment of the present invention. The digital chemical detection circuit 5200 may be an IC chip comprising a pixel array 5202, row decoders 5204.1 and 5204.2, column ADCs 5206.1 and 5206.2, I/O circuits 5208.1, a plurality of bias circuits 5210.1˜5210.3, a timing sequencer 5212 and 5208.2 and a plurality of output terminals D[0]˜D[N−1]. The pixel array 5202 may comprise chemical detection pixels formed in columns with each column including a plurality of rows of pixels. In one embodiment, the pixel array 5202 may include many tiles of pixels and the tiles may be placed in slices as described above with respect to FIG. 37. The row decoders 5204.1 and 5204.2 may generate row selection signals for rows of pixels based on control logic.

[0247] In one embodiment, the column ADCs 5206.1 and 5206.2 may include a plurality of ADCs each corresponding to one column. In another embodiment, the column ADCs 5206.1 and 5206.2 may include a plurality of ADCs and each ADC may be shared between several columns (e.g., using one or more multiplexers). Moreover, in one embodiment, the column ADCs 5206.1 and 5206.2 may perform offset cancellation.

[0248] The bias circuits 5210.1˜5210.3 may generate all bias and reference voltages needed for the chemical detection circuit 5200 on chip. That is, the chemical detection circuit 5200 does not have any external analog references. The bias circuits 5210.1˜5210.3 may need an external power supply, such as VDDA to function. The timing sequencer 5212 may provide the internal timing signals for the chemical detection circuit 5200.

[0249] The plurality of output terminals D[0]˜D[N−1] may provide serial links to one or more devices external of the IC chip. The serial links may use printed circuit board transmission lines. The signaling over the transmission lines may use differential signaling or CMOS signaling. The differential signaling may be any differential signaling scheme, for example, Low-voltage differential signaling (LVDS) or current mode logic (CML). In one embodiment, half of the output pins may be placed on one side of the chip 5200 and another half on an opposite side. The number N may be an even number (e.g., 24, 32). Further, in one embodiment, the serial interface may be programmable. For example, the strength of the drivers may be programmed and tuned for a given system. The driver type (LVDS, CML, CMOS) may also be programmed. Moreover, on-chip termination may be enabled or disabled. Various aspects of the protocol may be configured such as run-length control and format.

[0250] FIG. 53 illustrates a more detailed block diagram 5300 of the output circuits of the digital chemical detection circuit of FIG. 52 according to an embodiment of the present invention. As shown in FIG. 53, column interface 5302 may be connected to the pixel array (e.g., pixel array 5202) to read data out of the pixel array. The column comparators 5304 may include a plurality of comparators (e.g., ADCs). In one embodiment, the column comparators 5304 may perform offset cancellation.

[0251] The DACs 5312.1 and 5312.2 may provide reference voltages for the column comparators 5304 to perform analog-to-digital conversion. A pair of latches blocks 5306 and 5308 may provide buffer for the output data. The digitized data may be sent from the column comparators first to the A latches 5306 and then sent from the A latches 5306 to the B latches 5308 according to the control provided by the gray code blocks 5314.1 and 5314.2. A plurality of output serializer 5310.1˜5310.n may be coupled between the buffer and the output terminals D[0]˜D[n−1] (e.g., n=N/2). The gray code circuits 5314.1˜5314.2 may distribute a digital count to all of the latches that are controlled by the comparators 5304 such that when a given comparator transitions, the gray code may be latched into memory. This gray code may be set to count synchronously with a DAC ramp circuit which establishes a global reference for all comparators of the comparators 5304. When the global reference falls below the pixel value held at given column, a corresponding comparator may fire. Since the comparator can transition asynchronous to the clock, the count may be distributed with a gray code where only one bit transition is made at any time is used to avoid invalid codes.

[0252] In one embodiment, each output terminal may comprise two pins for differential signaling (e.g., low-voltage differential signaling). In one embodiment, the A latches 5306 may be the master latches while the B latches 5308 may be the slave latches. The latches may allow the analog-to-digital conversion to run in parallel to the readout of a previously converted row.

[0253] The on-chip bias and reference voltages may be provided by the bias and reference circuit block 5316. As shown in FIG. 53, the bias and reference circuit block 5316 may provide bias and reference voltages to the column interface circuit 5302, the DACs 5312.1 and 5312.2, and gray code circuits 5314.1˜5314.2. Further, as shown in FIG. 53, the timing sequencer circuit 5320 may provide timing signals to the gray code circuits 5314.1˜5314.2, DACs 5312.1˜5312.2, row decoders 5318.1˜5318.2, pixel controls and column controls. The timing sequencer 5320 may be connected to pins TMODE, TEN, RST for reset and test modes. The timing sequencer 5320 may also be connected to pins for SDA, SCL, CLK for programming registers with a serial protocol such as SPI. The timing sequencer 5320 may control all the timing on the chip by advancing each row during a frame time and providing stimulus to the circuits during each row time. Because the pixel can operate in many different ways, the timing sequencer5320 may be reprogrammed to change the operation of the control signals.

[0254] FIG. 54 illustrates a serializer circuit 5400 according to an embodiment of the present invention. The serializer circuit 5400 may comprise a plurality of shift registers 5402, bit alignment logic 5406, a pair of ping-pong registers 5406, a multiplexer 5408 to select one of the ping-pong registers, a multiplexer 5410 to multiplex the output from the multiplexer 5408 and built-in self test (BIST), an encoder 5412, a serializer 5416 and driver 5418.

[0255] The shift registers 5402 may be part of the I/O buffer. In one embodiment, as shown in FIG. 54, 14-bit shift registers may be used. The data shifted out of the shift registers may be sent to the bit alignment logic 5406, where the data may be aligned. For example, the 14-bits data may be aligned to 8-bits data according to align control. The aligned data from the alignment logic 5406 may be sent to the pair of ping-pong registers 5406 that latch each parallel word with timing overlap to prevent glitches in the data.

[0256] Then, the multiplexer 5408 may select one of the pair of ping-pong registers 5406 to output its data to the multiplexer 5410. The multiplexer 5410 may select either the output data from the multiplexer 5408 or the BIST data to be sent to the encoder 5412. To achieve DC balance, an encoding scheme such as 8b/10b may be used. Thus, in one embodiment, the encoder 5412 may be an 8B10B encoder. It should be noted that other encoding schemes may also be applied. The encoded data from the encoder 5412 may be serialized in the serializer 5413 and sent out by the driver 5418. In one embodiment, the serializer 5413 may be driven by a PLL clock signal. In one embodiment, the driver 5418 may be configured to transmit data signals by low-voltage differential signaling. In one embodiment, the driver 5418 may work in a differential mode, in which one bit may be transmitted at each transmit clock signal and the pair of output pins may carry the differential data pair. In another embodiment, the driver 5418 may work in a dual-channel mode, in which two bits may be transmitted per transmit clock cycle in parallel by the pair of output pins of an output terminal.

[0257] FIG. 55 illustrates a more detailed block diagram 5500 As shown in FIG. 55, the serializer 5413 may comprise a plurality of registers 5502, a pair of multiplexers 5504, a pair of registers 5506, a multiplexer 5508 and a buffer register 5510. The registers 5502 may each hold one bit of the encoded data. The registers 5502 may work at a clock speed at one tenth of the PLL clock signal. The output from the registers 5502 may be sent to the pair of multiplexers 5504. Each of the multiplexers 5504 may generate an output to be sent to one of the pair of registers 5506. The pair of registers 5506 may operate at one half of the clock speed of the PLL clock signal. The output from the pair of registers 5506 may be input to the multiplexer 5508, which may select one of the outputs from the pair of registers 5506 to be the output. The output from the multiplexer 5508 may be sent to the buffer register 5510. The buffer register 5510 may operate at the clock speed of the PLL clock signal to send out its content to the driver 5418.

[0258] FIG. 56 illustrates a block diagram of a digital chemical detection circuit according to an embodiment of the present invention. FIG. 56 shows a layout of a digital chemical detection circuit according to an embodiment of the present invention.

[0259] FIG. 57 illustrates a block diagram of another digital chemical detection circuit according to an embodiment of the present invention. FIG. 56 shows another layout of a digital chemical detection circuit according to an embodiment of the present invention.

[0260] FIG. 58 illustrates a block diagram of another digital chemical detection circuit 5800 according to an embodiment of the present invention. The digital chemical detection circuit 5800 may comprise a pixel array 5802, a plurality of output circuits 5804.1˜5804.2, a plurality of serial output circuits 5806.1˜5806.2, a plurality of row select circuits 5808.1˜5808.2, a clock tree 5810 and a plurality of thermometers 5812. The pixel array 5802 may be a 2T pixel array 1400 shown in FIG. 14A and comprise a plurality of 2T pixels configured according to FIG. 14A. The row select 5808.1˜5808.2 may be the row decoders as described above with respect to FIGS. 52-53. Also, the output circuits 5804.1˜5804.2 may include column interface, offset cancellation and column ADC as described above with respect to FIGS. 52-53. The serial output circuits 5806.1˜5806.2 may include the serializer circuits described above with respect to FIGS. 52-55. The clock tree 5810 may be an embodiment of the timing sequencer described above with respect to FIGS. 52-53. In one embodiment, the digital chemical detection circuit 5800 may include four thermometers placed on chip. Further, in one embodiment, the four thermometers may be placed at or near the four corners of the pixel array 5802.

[0261] Several embodiments of the present invention are specifically illustrated and described herein. However, it will be appreciated that modifications and variations of the present invention are covered by the above teachings. In other instances, well-known operations, components and circuits have not been described in detail so as not to obscure the embodiments. It can be appreciated that the specific structural and functional details disclosed herein may be representative and do not necessarily limit the scope of the embodiments. For example, some embodiments are described with an NMOS. A skilled artisan would appreciate that a PMOS may be used as well.

[0262] Those skilled in the art may appreciate from the foregoing description that the present invention may be implemented in a variety of forms, and that the various embodiments may be implemented alone or in combination. Therefore, while the embodiments of the present invention have been described in connection with particular examples thereof, the true scope of the embodiments and/or methods of the present invention should not be so limited since other modifications will become apparent to the skilled practitioner upon a study of the drawings, specification, and following claims.

[0263] Various embodiments may be implemented using hardware elements, software elements, or a combination of both. Examples of hardware elements may include processors, microprocessors, circuits, circuit elements (e.g., transistors, resistors, capacitors, inductors, and so forth), integrated circuits, application specific integrated circuits (ASIC), programmable logic devices (PLD), digital signal processors (DSP), field programmable gate array (FPGA), logic gates, registers, semiconductor device, chips, microchips, chip sets, and so forth. Examples of software may include software components, programs, applications, computer programs, application programs, system programs, machine programs, operating system software, middleware, firmware, software modules, routines, subroutines, functions, methods, procedures, software interfaces, application program interfaces (API), instruction sets, computing code, computer code, code segments, computer code segments, words, values, symbols, or any combination thereof. Determining whether an embodiment is implemented using hardware elements and/or software elements may vary in accordance with any number of factors, such as desired computational rate, power levels, heat tolerances, processing cycle budget, input data rates, output data rates, memory resources, data bus speeds and other design or performance constraints.

[0264] Some embodiments may be implemented, for example, using a computer-readable medium or article which may store an instruction or a set of instructions that, if executed by a machine, may cause the machine to perform a method and/or operations in accordance with the embodiments. Such a machine may include, for example, any suitable processing platform, computing platform, computing device, processing device, computing system, processing system, computer, processor, or the like, and may be implemented using any suitable combination of hardware and/or software. The computer-readable medium or article may include, for example, any suitable type of memory unit, memory device, memory article, memory medium, storage device, storage article, storage medium and/or storage unit, for example, memory, removable or non-removable media, erasable or non-erasable media, writeable or re-writeable media, digital or analog media, hard disk, floppy disk, Compact Disc Read Only Memory (CD-ROM), Compact Disc Recordable (CD-R), Compact Disc Rewriteable (CD-RW), optical disk, magnetic media, magneto-optical media, removable memory cards or disks, various types of Digital Versatile Disc (DVD), a tape, a cassette, or the like. The instructions may include any suitable type of code, such as source code, compiled code, interpreted code, executable code, static code, dynamic code, encrypted code, and the like, implemented using any suitable high-level, low-level, object-oriented, visual, compiled and/or interpreted programming language.