IMPERIAL COLLEGE

RAPID DNA DETECTION USING TRANSISTOR SUBSTRATES

FINAL YEAR PROJECT INTERIM REPORT
Fayemi, Oluwaseun 12/15/2011

Imperial College London Department of Electrical and Electronic Engineering Of07@imperial.ac.uk

Supervisors: Professor C. Toumazou, Dr C. Mehring

ABSTRACT
Chemical sensors are micro devices which are used to transduce chemical information into electrical signals; these devices are essential to many of the biochemical sensors in use today and are therefore being heavily researched upon to produce more advanced and accurate devices for the future. Significant advancements made in the field of technology have led to devices such as the ISFET (ion sensitive field effect transistor) which has evolved from being a pH sensitive device to one which is capable of sensing both chemical and biological samples. Devices like this have enabled healthcare services to be greatly improved upon in various ways. One area of healthcare which has benefited greatly from improving technology has been diagnosis. In recent years there has been a keen interest in chemical sensors which have paved the way for a new era in biomedical related applications such as the monitoring and assessment of complex chemical based body reactions. Since the invention of the ISFET (ion sensitive field effect transistor) in the 1970s there has been great interest in low power devices which can detect and assess biochemical inputs such as DNA (Deoxyribonucleic acid) so as to diagnosis diseases such as diabetes. The ISFET is not the only chemical sensor being researched on, but it is the most popular and prevalent design out of all of them. This is due to some advantages it has over the other sensor designs such as their small size, fast response time and compatibility with recent semiconductor fabrication techniques. The latter point is arguably the most important advantage as it allows for mass fabrication and therefore low cost, also it makes it possible for the ISFET to be monolithically integrated onto a single chip. However the ISFET also suffers from some abnormalities, of which threshold voltage and drift are most important.

Rapid DNA detection using transistor substrates

Page 1

TABLE OF CONTENTS
ABSTRACT ........................................................................................................................................................... 1 1. INTRODUCTION .............................................................................................................................................. 3 INTRO .................................................................................................................................................................. 3 PROJECT SPECIFICATION ..................................................................................................................................... 4 OBJECTIVES ......................................................................................................................................................... 4 2. BACKGROUND ................................................................................................................................................ 4 THEORY ............................................................................................................................................................... 4 MOSFET .......................................................................................................................................................... 5 ISFETS .............................................................................................................................................................. 6 BULK DRIVEN MIXER ....................................................................................................................................... 9 3. IMPLEMENTATION ....................................................................................................................................... 11 STAGE GATE PROCESS....................................................................................................................................... 11 PRIMARY RESEARCH ......................................................................................................................................... 12 FURTHER RESEARCH ......................................................................................................................................... 12 DESIGN AND SIMULATION ................................................................................................................................ 12 FINAL DESIGN EVALUATION ............................................................................................................................. 12 GANTT CHART ................................................................................................................................................... 12 4. PROJECT EVALUATION ................................................................................................................................. 13 CONCLUSION ................................................................................................................................................ 13 5. REFERENCES ................................................................................................................................................. 14

Rapid DNA detection using transistor substrates

Page 2

1. INTRODUCTION
INTRO
As we enter a new wave of technology inspired by lifestyle and healthcare, precise and efficient point of care testing is becoming increasingly important. The possibility to accurately detect a gene sequence in real-time using a standalone, fully portable, low power unit would provide end users with technology as yet unavailable outside a laboratory. This presents opportunities in areas where a real-time result can save time, money and lives. The ISFET in the past gained popularity for its use as a pH sensor, however due to the success of the identification of SNPs (single nucleotide polymorphisms) in DNA there has been a growing interest in them being used as a device to diagnose diseases leading to personalized drug therapy and medicine. A SNP is defined as a mutation of a single base in the DNA sequence affecting at least 1% of a defined population, for example two DNA sequences like the one shown below contain a difference in a single nucleotide (AAGCCTA to AAGCTTA). Although they do not generally cause diseases, their association with disease and with effects on pharmacokinetics of many drugs provides information for diagnosis and pharmalogical treatment options for many different diseases.

FIGURE 1: SNP FORMATION

DNA is a twisted double-helix structure made up of four bases called nucleotides; these bases are called adenine (A), guanine (G) which are the purines and cytosine (C), thymine (T) which are the pyrimidines. Due to the double stranded structure of DNA, it can copy itself by the process known as DNA replication. In this process the DNA double strand unwinds into two whereby each strand acts as a template to which nucleotides are added in the correct sequence through complementary base pairing; where A can only pair with T and C can only pair with G. Therefore each single strand of DNA grows into another double stranded DNA sequence resulting in DNA replication.

Rapid DNA detection using transistor substrates

Page 3

PROJECT SPECIFICATION
RAPID DNA DETECTION USING TRANSISTOR SUBSTRATES: The project involves the design of Novel CMOS circuits whereby the substrate is used to detect DNA in real time. This will involve design, simulation and laboratory testing of various fabricated devices. At the end of this project I intend to have a working prototype which would have been thoroughly investigated and tested through simulations on cadence beforehand. Due to the research nature of my project I shall be simulating various circuits and concluding the best route to take from the results.

OBJECTIVES
The project involves two major parts namely the design and simulation process and the fabrication process. The former requires good background knowledge about the ISFET operation and mixer technologies as it involves designing circuits where the bulk of the ISFET is to be used as contact. It also involves simulation of circuits and assessing the results obtained. The second part which is the fabrication process involves testing the finalised circuit to see if it gives similar results to what was obtained from the simulations.

2. BACKGROUND
THEORY
The ISFET is basically a MOSFET with the gate connection removed and replaced with a reference electrode inserted into an aqueous solution (electrolyte) which is in contact with the gate oxide. The differences and similarities between the two are shown in the figure below. Therefore the ISFET and MOSFET have a lot of things in common with regards to equations and modes of operations. So an understanding of the MOSFET operation will give a basic idea to the fundamentals of the ISFET.

FIGURE 2: SCHEMATIC DIAGRAM OF A MOSFET (LEFT) AND AN ISFET (RIGHT).

Rapid DNA detection using transistor substrates

Page 4

MOSFET
STRUCTURE
The MOSFET (Metal Oxide Semiconductor Field Effect Transistor) is a type of FET which measures the conductance of a semiconductor as a function of the electric field perpendicular to the gate oxide surface. It consists of four major parts namely the gate, source, drain and bulk (substrate) which are shown in the figure below.

FIGURE 3: SCHEMATIC OF A MOSFET STRUCTURE.

FIGURE 4: ELECTRONIC DIAGRAM

The substrate is made with a semiconductor (usually silicon) and so are the source and drain but with metal contacts, while the gate is a metal underneath which is an oxide (usually silicon dioxide).There are two design structures of the MOSFET which depend on the type of carriers present in the conducting channel, which can either be p-type or n-type. The channel is formed by the gate source voltage (V GS) being greater than the threshold voltage (V th). For an n-type MOSFET when this happens the carriers that the substrate is made up of near the gate (holes) are repelled, therefore leaving the region just below the gate negatively charged (electron carriers) and hence a channel between the source and drain is formed. In addition to this the MOSFET can either be a depletion or enhancement mode device. In the enhancement case Vth > 0 while in the depletion mode Vth < 0, therefore the latter always has a channel formed even in the absence of VGS.

MODES OF OPERATION
Looking at the voltages in an n-channel enhancement mode MOSFET there are three major modes of operation which are discussed below: Weak inversion/ sub threshold mode: This occurs when VGS < Vth, and as discussed above this means no channel is formed between the source and the drain. Therefore in an ideal MOSFET there should be no conduction between the drain and source which means no current, however there is a weak inversion current which is detected called the subthreshold leakage. Triode mode/ linear region: This occurs when VGS > Vth and VDS < (VGS - Vth). In this mode the MOSFET behaves like a resistor and the gate-source voltage is greater than the threshold voltage, therefore there is a channel for current to flow between the source and drain. This current is called the drain current and it increases linearly with the drain-source voltage, the equation for the drain current is (1)

Rapid DNA detection using transistor substrates

Page 5

With Cox being the oxide capacitance per unit area, W and L are the width and length of the channel respectively and is the electron mobility of the carriers in the channel. Saturation Region: This occurs when VGS > Vth and VDS > (VGS - Vth). In this mode the drain voltage is higher than the gate voltage, and a channel for current flow between the drain and source has also been created due to the gate voltage being higher than the threshold voltage. Also the MOSFET is said to be switched on in this region, and this leads to an increased channel resistance which leads to the flattening out of ID in the saturation region (green region in the plot shown below).

FIGURE 5: ID VS. VDS CURVE FOR A MOSFET WITH VGS AS A PARAMETER.

ISFETS
STRUCTURE
As mentioned above the ISFET is basically a MOSFET with the gate connection replaced with a reference electrode inserted into an electrolyte whose concentration of certain ions is wished to be determined and is in contact with the gate oxide. In an ISFET the polysilicon gate and metal are replaced with an insulating membrane for the hydrogen ions to bind onto. The reference electrode which is usually made up of Ag/AgCl is used to apply the gate bias for the transistor. Insulating membranes are predominantly made up of materials such as aluminium oxide (Al 2O3), silicon dioxide (SiO2) and silicon nitride Si3N4 as they are all very sensitive to hydrogen ions. Also advancements in technology allowed for the ISFET to be fabricated in unmodified CMOS technology therefore allowing for further circuitry to be integrated with it therefore improving the transistor, below is a schematic of this.

FIGURE 6: SCHEMATIC OF AN ISFET IN UMODIFIED CMOS

Rapid DNA detection using transistor substrates

Page 6

OPERATION
Due to the major similarities with the MOSFET the ISFET therefore also shares the same drain current equation for the MOSFET in its non-saturated mode equation (1). For both cases the drain current ID is a unique function of the gate voltage VGS only when the geometric sensitivity , the threshold voltage and the drain-source voltage are constant. The parameter is a design constant and the drainsource voltage can be kept constant by an electric circuit. Also the threshold voltage is also constant due to the fabrication process being so well under control, therefore leaving VGS as the only input variable. However in the case of the ISFET the effect of the ion concentration from the electrolyte had to be taken into account, this was done by modifying the threshold equation for the ISFET as the reference electrode is defined as a remote gate. This means that the ISFETs threshold voltage is dependent to the pH. The original MOSFET threshold equation and the modified ISFET threshold equation are show below. (2) (3) From equation (2) the first term is the difference between the workfunction of the gate metal (M) and that of the silicon (Si), the second term is the accumulated charge in the oxide (Q ox), at the oxide silicon interface (Qss) and the depletion charge in the silicon (Q B) and the last term being the onset of inversion depending on the doping level of the silicon. In the ISFET due to the same fabrication process used for the MOSFET, the physical variables for the threshold voltage are also the same. However some additional terms need to be added to compensate for some new contributions. These are Eref which is the constant potential of the electrode and which is the interface potential at the electrolyte/oxide interface, where is the chemical input parameter and is the surface dipole potential of the solvent which has a constant value. The three extra terms added in the ISFET equation can be grouped together and called Vchem (5), which is the additional threshold voltage to that of the intrinsic MOSFET from which the ISFET is fabricated from (4). A more simplified version for Vchem is also shown below (6), where Ut is the thermal voltage of the device, is a grouping of non chemically related potentials and is a constant determining the reduction in sensitivity from the Nernstian response, typically 59mV/pH. (4) (5) (6)

PASSIVATION CAPACITANCE
When the ISFET is implemented using CMOS technology, a passivation capacitance is introduced between the top metal and the solution which scales the voltage established on the insulating membrane. The top metal in contact with the insulator which defines the active area of the device which is seen by the solution charge can be modelled as a capacitor by assuming the two passivation dielectrics (SiO 2, Si3N4) used for the chip insulation as shown in figure 5 form a series capacitive network. This passivation capacitance can be calculated using equation (7) where (W x L)chem is the area of the top metal in contact with the passivation, 0, and are the permitivities of free space, silicon nitride and silicon dioxide respectively, and , are the respective passivation thicknesses. (7)

Rapid DNA detection using transistor substrates

Page 7

The effect of the passivation capacitance is a division of the surface potential. A schematic of the CMOS ISFET with the capacitances added is shown in figure 6 below. This shows , which is the effective gate voltage now seen by the device and can be calculated as follows: (8) Where Cpass , Cox and Cd are the passivation, oxide and depletion capacitances respectively.

FIGURE 7: CAPACITANCES OF A CMOS ISFET

MEASUREMENT AND RESULTS

In an ISFET a reference potential V ref is applied to the aqueous solution via a reference electrode. Measuring the response to changes in ion concentration can be done in a number of ways; however the most important factor is how the potential drop at the electrolyte/oxide layer varies with regards to ion concentration. This potential drop is due to a number of interactions taking place namely: 1. 2. 3. 4. 5. The drop across the solution The drop across the electrolyte/oxide interface The drop across the oxide layer The drop across the oxide/semiconductor interface The drop across the bulk semiconductor

In a good approximation the interactions 2 to 5 are independent of a change in ion concentration, therefore leaving as the main influence on the potential drop across the electrolyte/oxide interface. So with a constant Vref (formerly VG) the potential drop across the oxide/semiconductor system is changing with , which leads to a relationship between VG and ion concentration. This in turn leads to a relationship between drain current (ID), also ID can be kept constant while a change in V ref is measured. The results of this relationship are show in the pl ots below. The influence of the ion concentration on can be explained by the site binding model which will be explained below.

Rapid DNA detection using transistor substrates

Page 8

FIGURE 8: ID/VDS GRAPH WITH VGS AS A PARAMETER

FIGURE 9: ID/VDS GRAPH WITH PH AS A PARAMETER

SITE BONDING MODEL

This model proposes that atoms at the surface layer of semiconductor oxides are amphoteric, meaning that they can act as a base or an acid. This allows them to release protons into the electrolyte as donors and thus become negatively charged, or form neutral OH sites and bind protons from the electrolyte as acceptors which results in a positive surface charge. M OH M O + H
+ +

DONOR
+

M OH + H M + OH2

ACCEPTOR

These reactions depend on the acidity or basicity constant of the oxide groups and also the concentration of H+ in the electrolyte. In the case of a high concentration of H + (low pH), the M OH groups will rather accept a proton than release one, therefore most of them act as acceptors and the oxide becomes positively charged. Otherwise if the concentration of H+ is low (high pH) then most of the M- OH groups will release a proton therefore making the surface charge negative. The changes in surface charge due to change in pH in the electrolyte directly affect the surface potential , thus leading to a relationship between Vref and pH shown below. (pH) Vref() = Vref(pH)

BULK DRIVEN MIXER

In the case of monitoring the chemical reaction between two substances, it is desirable to be able to compare say the pH of the two solutions under test. Such a comparison can be done by applying mixer technology which can multiply the two inputs electronically. Although there are various mixer technologies out there, I shall be focusing mainly on the bulk driven mixer architecture as it reduces supply voltage and power dissipation while maintaining reasonable gain and linearity. The MOSFET is usually used as a four terminal device (drain, bulk, source and gate) where the signal is input through the gate and the bulk used to isolate the MOSFET from other devices on the substrate. However in a bulk driven mixer a twin well technology is required and the MOSFET becomes a five terminal device, where both the gate and the bulk can be used as an input while the isolation is provided by a deep n/p well for an nMOS and pMOS respectively. Below is a schematic of an N-channel MOSFET in a p substrate twin well.

Rapid DNA detection using transistor substrates

Page 9

FIGURE 10: N-CHANNEL MOSFET IN A PSUBSTRATE TWIN WELLTECHNOLOGY, WITH BULK CHANNEL PARASITIC P-WELL RESISTANCE

Using a twin-well technology provides 2 reverse biased p-n junctions worth of isolation between the MOSFET and the substrate, which is important at high frequencies. The topology of the proposed bulk driven mixer core is illustrated in Fig 10. In the circuit, the switching action is provided via the gate, while the transconductance is obtained through the bulk. LO+ and LO- are two antiphase signals from the local oscillator that are used to switch M1-4 ON and OFF. When switched ON, M1-4 are held in the saturation region by a sufficiently large VDS. This ensures that reasonable gain can be obtained. The mixer core is used as a common source pair in any one phase of the local oscillator. As long as the RF input is differential, the circuit will work differentially. In the positive phase of the local oscillator, LO+ is sufficiently larger than the threshold voltage of M1 and M4 to switch them ON (in the saturation region). At the same time, LO- is less than the threshold voltage of M2 and M3 to switch them OFF (in cut-off). The RF input can now pass to the IF output via the back gate transconductance (gmbs) of M1 and M4. In the negative phase of the local oscillator, M2 and M3 are ON, while M1 and M4 are OFF. The RF input is now inverted when it appears at the IF output. Thus the RF input is commutated by the action of the local oscillator on the gates of M1-4, and hence converted to the desired IF frequency.

FIGURE 11: A BULK DRIVEN MIXER CORE

Using the bulk-driven mixer configuration I intend to either modulate the pH signal with a frequency signal, or take a differential/comparative pH sensing approach where the gate and substrate are connected to the sensing membranes of different pHs. Driving the bulk node can be used to modulate the threshold voltage via the body effect as shown below. (9) Where VT0 is the threshold voltage for zero substrate bias, F stands for the surface potential, the body effect factor and VBS the bulk-source voltage. In the design of the circuit, the reference solution (pH1) is Rapid DNA detection using transistor substrates Page 10

connected to the sensing membrane of the gate while the second solution (pH2) is attached to the sensing membrane of the bulk. Taking all these into account we can formulate an expression for the drain current in various modes with the bulk as an input. In the weak inversion for example: (10) Where ID0 represents current at VGS = Vth, VG the effective gate voltage seen by the device, VS the source voltage, Vchem the additional threshold voltage to that of the intrinsic MOSFET from which the ISFET is made, VT0 the threshold voltage of the MOSFET without the bulk connected, n is the slope factor while U T is the thermal voltage. For strong inversion: (11) For Saturation: (12)

3. IMPLEMENTATION
STAGE GATE PROCESS
Due to my project being more open-ended and research based I have decided to use the stage gate project management tool which separates all the major stages of my project by gates. The gates represent meetings with my superiors such as supervisors and PhD students as to how best to proceed with the project and also relevant deliverables. I also plan to schedule more meetings with my supervisor than shown in the diagram below, currently the project is at stage 2.

Rapid DNA detection using transistor substrates

Page 11

PRIMARY RESEARCH
This stage involved reading background material such as published papers and past student thesis so as to get familiar with the core material and principles which were relevant to the project. After this stage I was familiar with the basic concepts and theory of the ISFET, also I read up on the devices course notes from previous years so as to familiarise myself again with the basic MOSFET theory.

FURTHER RESEARCH
After gaining a confident understanding of the background knowledge needed to understand the basic theory of the ISFET, I then proceeded to research into how the ISFET was specifically needed in my project. The main concept of my project is to investigate how the bulk of the ISFET can be utilised as a fourth contact for the rapid detection of DNA, therefore understanding mixer technologies are of great importance. As a result various mixer architectures are being researched on by me such as the bulk driven ISFET based chemical mixer.

DESIGN AND SIMULATION

Due to the research based nature of my project I will have to propose different circuit designs and then test them by conducting various simulations. I shall be using Cadence Spectre in the spring term for most of my simulations as it has all the relevant libraries in regards to the ISFET model I am going to be utilising, it is also the industry standard for the designing of circuits. This stage will require simulations of circuits which will either modulate the pH signal with a frequency signal or be of a differential nature comparing the pH of two substances applied on the sensing membranes of the gate and bulk. This will also involve important design decisions such as can the same reference voltage be applied to both solutions which will be ideal case or do they require different voltages. Also this will be the longest and most important of all the stages as it will provide me with tangible results which will help to reinforce my understanding of the concepts. During this stage I shall schedule several meetings with my supervisor and PhD mentor so as to get feedback on the circuits and simulation result.

FINAL DESIGN EVALUATION

At this stage I will have already discussed the various circuit designs and the results obtained from them with my supervisor and would have come to a conclusion as to what the best design is, that is either going the frequency modulation or differential route. After deciding on the design to use I will perform further tests and simulations and record the results for use in a possible publication paper. In addition to this I plan to have the circuit design fabricated and tested in the labs to see if the desired results from the simulations can be duplicated.

GANTT CHART

Rapid DNA detection using transistor substrates

Page 12

4. PROJECT EVALUATION
The project description states that I will be involved in the design of Novel CMOS circuits whereby the substrate is used to detect DNA in real time. This will also involve design, simulation and laboratory testing of various fabricated devices. Therefore the success of the project will be based on firstly designing circuits where the bulk of the ISFET can be used as an input contact. Secondly testing the circuits validity by performing relevant simulations on them and discussing the results gotten with my supervisor. Finally fabrication of the final circuit design so that it can be tested in real life situations in the lab.

CONCLUSION
In this paper I have presented the concept of using an ISFET for rapid DNA detection. Using the concept of the bulk driven mixer architecture I have proposed a way in which to use the ISFET as some sort of differential device where two substances of different pH are applied to the bulk and gate, or to modulate the pH signal with a frequency signal. Also due to the bulk being connected the ISFET will experience some sort of body effect as the MOSFET does, therefore I have proposed some equations for weak and strong inversion in the ISFET where the bulk-source voltage features. In the coming term I plan to learn the simulation platform Cadence and use it to test these assumptions.

Rapid DNA detection using transistor substrates

Page 13

5. REFERENCES
http://csrg.ch.pw.edu.pl/tutorials/isfet/ ISFET characteristics in CMOS and their application to weak inversion operation. Pantelis Georgiou and Christofer Toumazou A bulk driven ISFET based chemical mixer. W.S Wan Zain, T. Prodromakis and Christofer Toumazou Thirty years of ISFETOLOGY. P. Bergveld Ion sensitive field effect transistors basics and applications. Stephan Schmid Weak inversion ISFETs for ultra-low power biochemical sensing and real-time analysis. Leila Shepherd and Christofer Toumazou http://www-g.eng.cam.ac.uk/mmg/teaching/linearcircuits/mosfet.html http://www.ieee-sensors.org/files/2011/08/ISFET-Bergveld.pdf Toward ISFET based DNA logic for rapid nucleic acid detection. Winston Wong, Leila Shepherd, Pantelis Georgiou and Chris Toumazou An ISFET based translinear sensor for DNA methylation detection. Melpomeni Kalofonou, Pantelis Georgiou, Chung-Pei Ou and Christofer Toumazou Designing ISFETs which are insensitive to passivation capacitance variation. Mohammadreza Sohbati, Yan Liu, Pantelis Georgiou and Christofer Toumazou A low bulk driven downconversion mixer core. Ganesh Kathiresan and Christofer Toumazou

Rapid DNA detection using transistor substrates

Page 14