+ All Categories
Home > Documents > Multiple State Electrostatically Formed Nanowire Transistors

Multiple State Electrostatically Formed Nanowire Transistors

Date post: 22-Nov-2023
Category:
Upload: independent
View: 0 times
Download: 0 times
Share this document with a friend
14
Multiple State EFN Transistors Gideon Segev, Iddo Amit, Andrey Godkin, Alex Henning and Yossi Rosenwaks School of Electrical Engineering, Tel Aviv University, Israel Abstract Electrostatically Formed Nanowire (EFN) based transistors have been suggested in the past as gas sensing devices. These transistors are multiple gate transistors in which the source to drain conduction path is determined by the bias applied to the back gate, and two junction gates. If a specific bias is applied to the side gates, the conduction band electrons between them are confined to a well-defined area forming a narrow channel- the Electrostatically Formed Nanowire. Recent work has shown that by applying non-symmetric bias on the side gates, the lateral position of the EFN can be controlled. We propose a novel Multiple State EFN Transistor (MSET) that utilizes this degree of freedom for the implementation of complete multiplexer functionality in a single transistor like device. The multiplexer functionality allows a very simple implementation of binary and multiple valued logic functions. 1. Introduction Electrostatically Formed Nanowires (EFN) based transistors have been recently suggested as robust sensing[1] and memory devices [2]. The EFN device is based on the silicon-on- insulator (SOI) four gate field-effect transistor (G 4 -FET) developed in 2002 [3], [4] which emerged from the volume inversion SOI MOSFET[5]. The G 4 -FET combines MOSFET and junction gate field-effect transistor (JFET) principles as it consists of a top MOS gate (V TG ), a bottom substrate gate (V BG ) and is enclaved between by two lateral junction gates (V JG1 , V JG2 ). The four gate transistor can be naturally adapted to CMOS technology scaling and manufactured in conventional silicon-on-insulator (SOI) processes with a low cost and high volume manufacturing. In EFN devices the top gate is removed in order to allow
Transcript

Multiple State EFN Transistors

Gideon Segev, Iddo Amit, Andrey Godkin, Alex Henning and Yossi Rosenwaks

School of Electrical Engineering, Tel Aviv University, Israel

Abstract

Electrostatically Formed Nanowire (EFN) based transistors have been suggested in the

past as gas sensing devices. These transistors are multiple gate transistors in which the source

to drain conduction path is determined by the bias applied to the back gate, and two junction

gates. If a specific bias is applied to the side gates, the conduction band electrons between

them are confined to a well-defined area forming a narrow channel- the Electrostatically

Formed Nanowire. Recent work has shown that by applying non-symmetric bias on the side

gates, the lateral position of the EFN can be controlled. We propose a novel Multiple State

EFN Transistor (MSET) that utilizes this degree of freedom for the implementation of

complete multiplexer functionality in a single transistor like device. The multiplexer

functionality allows a very simple implementation of binary and multiple valued logic

functions.

1. Introduction

Electrostatically Formed Nanowires (EFN) based transistors have been recently suggested

as robust sensing[1] and memory devices [2]. The EFN device is based on the silicon-on-

insulator (SOI) four gate field-effect transistor (G4-FET) developed in 2002 [3], [4] which

emerged from the volume inversion SOI MOSFET[5]. The G4-FET combines MOSFET and

junction gate field-effect transistor (JFET) principles as it consists of a top MOS gate (VTG), a

bottom substrate gate (VBG) and is enclaved between by two lateral junction gates (VJG1,

VJG2). The four gate transistor can be naturally adapted to CMOS technology scaling and

manufactured in conventional silicon-on-insulator (SOI) processes with a low cost and high

volume manufacturing. In EFN devices the top gate is removed in order to allow

functionalization of the top surface, for example for gas sensing as suggested in [6]. If a

specific bias is applied to the side gates, the conduction band electrons between them are

confined to a well-defined area forming a narrow channel, the electrostatically-formed

nanowire[7]. Figure 1 (a) shows a schematic illustration of such a transistor and the formation

of the EFN. Figure 1 (b) illustrates the formation of the EFN in a cross section view.

Glazman et al. proposed the lateral position control of an electron channel with a split-

gate FET[8]. Shalev et al.[1] have shown in 3D electrostatic simulations that by applying

non-symmetric bias on the side gates, the position of the EFN can be moved towards one of

the gates.

We propose here novel Multiple State EFN Transistors (MSET) that exploits the EFN

lateral movement in order to form a single transistor multiplexer. In this device the drain is

split into several isolated individual drains and the MSET output is defined by a thin

conduction channel between a specific drain and the source. The multiplexer functionality

allows implementation of any logical operation within its inputs and outputs range.

Furthermore, since it supports multiple well defined conduction states, it can perform multiple

valued logic (MVL) operations as well. Since this device is based on simple SOI concepts, it

can be integrated into current technology relatively easily. To the best of our knowledge, this

is the first CMOS compatible transistor for MVL large-scale circuits to be suggested.

Although this work focuses on utilizing the MSET for logic applications, it is clear that it can

be used in many other fields as well.

Figure 1, 3D view of the 3 gate EFN transistor (a), cross section and formation of the EFN (b).

2. Operation

2.1. Two Gates MSET

The basic MSET configuration closely resembles the transistor illustrated in Figure 1 with

one important difference- the drain is split into several isolated drain inputs, labeled d1..dn,

where each drain is connected to a specific voltage, labeled Vd,1..Vd,n, according to the

required functionality. A known combination of side gate voltages brings the EFN next to one

of the drains, defining a single conduction path between a single drain and the source. For

logic applications as discussed below, the bottom gate can be removed. Figure 2 shows an

illustration of a two-drain MSET device embedded in a circuit which allows definition of

voltage based states. When V1 is applied to one of the gates and 0 is applied to the other, the

conduction channel will be next to the drain close to the latter gate. In Figure 2, the applied

biases are VG,1=V1 and VG,2=0 such that the EFN is next to d2. Since the resistance of the

conduction path between d2 and the source is significantly lower than the resistance of the

serial resistor R, the output voltage is Vs=Vd,2. In a similar fashion, applying opposite

voltages, VG,1=0 and VG,2=V1 yields Vs=Vd,1. When both gates are at voltage V2 the entire

region between the two gates is completely depleted. As a result, the resistance between the

source and the two drains is higher than R and Vout is 0. When both gates are at 0 there is

VD

VG,1

VG,2

(a) (b)

EFN

conduction between both drains and the state is undefined. When one gate is at V1 and the

other is at V2 the output voltage can be between 0 and the corresponding drain voltage

depending on the specific device engineering. These intermediate states can increase the

device functionality considerably but may result in high currents between the drains and high

power consumption. Table 1 shows the truth table for the MSET based circuit described in

Figure 2. It should be noted that as in CMOS configurations, the proposed configuration

possesses very low current flow through the device once the state is determined. In order to

further reduce the source current, the resistor in Figure 2 can be replaced with transistors that

will conduct or isolate the path between Vout and the ground according to the desired

functionality.

Figure 2, two drains MSET device in device in an exemplary circuit which allows definition of voltage

based states.

Vg1 Vg2 Vout

0 0 --

0 V1 Vd,1

V1 0 Vd,2

V2 V2 0 Table 1, 2 drains MSET truth table.

The addition of a third drain to the MSET will greatly increase its functionality. Having

three possible conduction states, this device can perform ternary based logical operations. We

assume a circuit similar to that shown in Figure 2 with a 3 drain MSET replacing the 2 drain

R

VO

VD1VG,1

VG,2

VD2

MSET. In this case, conduction through drain 1 is possible when Vg,1=0 and Vg,2=V2.

Conduction through the middle drain is possible when both gates are at V1 and conduction

through drain 3 is possible when Vg,2=0 and Vg,1=V2. When both gates are at V2 there is no

conduction through any of the drains and the output is 0. V1 and V2 in this example do not

necessarily have the same values as V1 and V2 in the previous example. Table 2 shows the

truth table for this configuration.

Vg,1 Vg,2 Vout

0 V2 Vd,1

V1 V1 Vd,2

V2 0 Vd,3

V2 V2 0 Table 2, basic 3 drain MSET truth table.

2.2. Four Gates, Three Dimensional Architecture

In order to add degrees of freedom to the MSET a three dimensional architecture is

proposed. The basic mode of operation is two dimensional movement of the channel, and the

conduction path is determined by 4 independent gates. The drains are isolated islands between

the gates and the source is located beneath the drains. In this case, instead of 2 gates which

determine the chosen drain, two side gates voltages (Vg,E and Vg,W respectively) determine

the chosen drain column and the other two gates voltages (Vg,N and Vg,S respectively)

determine the chosen row. We denote Vd,i,j the drain in the ith row and j

th column. Figure 3

shows a schematic illustration of such a device (a) and a cross sectional view along with

voltage based state circuit (b). In this example VJG,s>VJG,n and VJG,e>VJG,w and the EFN forms

a conduction path between d1 and the source. The device shown has 4 input channels VJG,s,

VJG,n, VJG,e, VJG,w. If the device output is determined by the source voltage and V1 and V2 are

defined as the 2 drain MSET above, 4 outputs Vd,1,1, Vd,1,2, ,Vd,2,1 and Vd,2,2 are possible.

Table 3 shows a truth table for a 3D MSET with 4 drains. As above, in order to allow a

voltage based output in which little power is dissipated in the device, only specific

combinations of inputs are allowed. Choosing combinations other than those in Table 3 can

add to the number of possible states but may cause high currents to pass between the drains

leading to high power consumption. As in the 2 gates configuration, adding more drains will

increase the device functionality significantly and allow realization of non binary logic.

Assuming voltages V1 and V2 as in the 3 drain MSET example above, the truth table of a 4

gates, 9 drain MSET is given in Table 4.

Figure 3, a three dimensional, 4 drain MSET device, 3D view (a) and cross section in a voltage based

states circuit (b)

Vg,N Vg,S Vg,W Vg,E Vout

0 0 0 0 --

0 V1 0 V1 Vd,1,1

0 V1 V1 0 V d,1,2

V1 0 0 V1 V d,2,1

V1 0 V1 0 V d,2,2

V2 V2 V2 V2 0 Table 3, 4 gates 4 drain MSET truth table.

Vg,E Vg,W Vg,N Vg,S Vout

0 V2 0 V2 V1,1

V1 V1 0 V2 V1,2

V2 0 0 V2 V1,3

(a)

R

V12

V11

VO

VG,E

VG,W

(b)

0 V2 V1 V1 V2,1

V1 V1 V1 V1 V2,2

V2 0 V1 V1 V2,3

0 V2 V2 0 V3,1

V1 V1 V2 0 V3,2

V2 0 V2 0 V3,3 Table 4, 4 gates, 9 drain MSET truth table.

3. 2-4 Multiplexer realization

The advantages of the MSET can be easily demonstrated by realizing a 2 control 4 inputs

multiplexer. The 4 gates, 4 drains multiplexer as shown in section 2.2 can be used to select

one of its 4 inputs. However, the need for 4 different gates voltages is somewhat

cumbersome. Hence, there is a need for a circuit that translates the multiplexer control inputs

to the appropriate gate voltages. We refer to this circuit as the State to Gate (S2G) circuit. As

seen in Table 3, conduction through the first column requires voltages of 0 and V1 on JGE and

JGW respectively. On the other hand, in order to conduct from the second column, JGE and

JGW should be V1 and 0 respectively. Hence, the voltage on JGW is always the inverse of JGE

and the S2G circuit can be realized with a simple NOT gate. Figure 4 shows an illustration of

a 2 control 4 inputs multiplexer realized with a 4 gates 4 drains MSET and two not gates. It

should be noted that a similar nonrestoring multiplexer realized in CMOS consists of 16

transistors[9].

Figure 4, an illustration of a two control 4 inputs multiplexer realized with a single MSET and two not

gates.

a1

a0

Vout

In order to realize complex logical several devices must be connected in series. Hence,

the MSET output voltage (source voltage) must be of the same sign and magnitude as the

input voltage (gates and drain voltage). However, in order to avoid forward biasing between

the gates-drains p-n junction the drain voltages must be non negative and the gate voltages

must be non positive. This contradiction can be removed in several ways: addition of a

voltage shifting circuit, addition of a voltage inverting circuit or alternating p-n design. The

first two circuits can be realized together with the S2G component in order to reduce the

number of transistors used in the circuit. Using an alternating p-n design requires that the

drain voltages of the n type MSET will be exactly the required gate input of p type MSETs

and vies-versa. Obviously, this layer is not required when the output of one MSET is

connected to the drain of another.

4. Simulation

In order to demonstrate the basic MSET concept of operation, a proof-of-concept two-

drains MSET was simulated with Sentaurus TCAD. The MSET device in these simulations is

the same as in the previous section coupled with a 10MΩ resistor which is placed between the

source and the ground. The drain voltages Vd,1 and Vd,2 are 1.5V and 1V respectively, the rest

of the MSET parameters are listed in Table 5. The simulated circuit including the MSET

device, resistor and voltage sources is as in Figure 2. The source voltage as a function of the

two gate voltages is shown in Figure 6. When there is a high voltage on one of the side gates

the conduction path is pushed away from it forming a single connection between the opposite

drain and the source. As a result, when Vg,2 is more negative than -1.5V and Vg,1 is close to

0V, the output voltage is exactly Vd,1. Similarly, when Vg,1 is below -1.5 and Vg,2 is close to

zero, the output voltage is exactly Vd,2. Last, when both gates are close to -3V, there is no

conduction through the device and the output is close to 0V. Hence, three different states can

be defined for this device: S0- where Vo=0, S1 where Vo=Vd,1 and S2 where Vo=Vd,2. From

Figure 6 it can be seen that input voltages of -3V and 0V are sufficient to switch between the

three states. Furthermore, since the output and input are of the same order of magnitude,

device concatenation is possible simply by shifting the voltage output by -3V. Table 6 shows

a truth table of the simulated circuit summarizing the different MSET states.

Table 5, two gates, 2 drains MSET simulation parameters.

Parameter Symbol Size (nm)

SOI width W 600[nm]

SOI length L 1000[nm]

SOI thickness T 145[nm]

Buried Oxide thickness TB 5[nm]

Drains length and width LD,WD 100[nm]

Drains thickness TD 145[nm]

Source width WS 400[nm]

Source length LS 100[nm]

Source thickness TS 145[nm]

Gates length LG 1000[nm]

Gates width WG 5[nm]

Gates thickness TG 145[nm]

Oxide buffer length LB 300[nm]

Oxide buffer width WB 160[nm]

Oxide buffer thickness TB 145[nm]

Drain 1 voltage VD,1 100[mV]

Drain 2 voltage VD,2 200[mV]

Bulk Doping Nd,0 1017

[cm-3

], Arsenic

Drains Doping Nd,D 1019

[cm-3

], Arsenic

Source Doping Nd,S 1019

[cm-3

], Arsenic

Gates Doping Na,G 1019

[cm-3

], Boron

Figure 5, 2 gates MSET simulation parameters

WD

WS

LD

LB

WG

LG

WB

LS

W

L

Top ViewFront View

Side View

TG

TDTB

T

TBOX

Back View

TS

Figure 6, two gates, two drains MSET source voltage as a function of the Vg,1 and Vg,2. The circuit is as

in Figure 2 and the MSET parameters are listed in Table 5. The drain voltages Vd,1 and Vd,2 are 1.5V and 1V

respectively and R=10MΩ.

Vg,1 Vg,2 Vo State

0 -1.5 Vd,1 S1

-1.5 0 Vd,2 S2

-3 -3 0 S0

Table 6, 2 drain MSET truth table.

As discussed above increasing the number of drains can increase the device functionality

significantly. However, since these devices are larger than the two drain devices presented

above, larger areas must be depleted leading to higher required gates voltages. In order to

reduce the area that has to be depleted and the gate voltages, the silicon oxide drain separators

can be replaced with highly doped p type gates, each connected to the near side gate. Figure

7(a) shows a schematic design for a two gates, three drains MSET with such active drain

separators. Table 7 lists all the device parameters. The source voltage as a function of the two

side gates voltages is shown in Figure 7(b). As in the previous simulation, the different states

can be easily distinguished.

Table 7, two gates, three drains MSET simulation parameters.

Vg,1

[V]

Vg,2

[V

]

Vs [V]

-3 -2.5 -2 -1.5 -1 -0.5 0

-3

-2.5

-2

-1.5

-1

-0.5

0

0.2

0.4

0.6

0.8

1

1.2

1.4S0

S2

S1

Parameter Symbol Value

Device width L,W 500 nm

Device length L 800 nm

Drains length and width LD,WD 50 nm

Source width WS 400 nm

Source length LS 100 nm

Gates length LG 400 nm

Gates width WG 50 nm

Oxide buffer length LB 300 nm

Oxide buffer width WB 50 nm

Drain 1 voltage VD,1 600mV

Drain 2 voltage VD,2 400mV

Drain 3 voltage VD,3 200mV

Resistor R 1TΩV

Bulk Doping Nd,0 2∙1017[cm

-3], Arsenic

Drains Doping Nd,D 1019

[cm-3

], Arsenic

Source Doping Nd,S 1019

[cm-3

], Arsenic

Gates Doping Na,G 1019

[cm-3

], Boron

Figure 7, (a) the simulated two gates, three drains MSET. (b) Source voltage as a function of the Vg,1

and Vg,2. The circuit is as is Figure 2 and the MSET parameters are as in Table 7. The drain voltages Vd,1,

Vd,2 and Vd,3 are 0.6V, 0.4V and 0.2V respectively

5. Conclusion

A new device capable of performing multiple valued operation with an inherent

multiplexer functionality is suggested. Wide spread implementation of MSET based circuits

in ASIC and FPGA chips relies on the ability to fabricate them in appropriate magnitudes

with low power consumption at sufficient frequency. As discussed above, an MSET based 4

Vg,1

[V]

Vg,2

[V]

-2.5 -2 -1.5 -1 -0.5 0

-2.5

-2

-1.5

-1

-0.5

00.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

1.1S0 S1

S3

S2

s

d1 d2 d3

g2

g2g1

g1

(a) (b)

inputs multiplexer can be realized with a single MSET and 4 MOSFET transistors, far less

than any current technology. Hence, even if a 4 gates, 4 drains MSET is 12 times less

efficient than current transistors in terms of magnitude, power consumption and speed, the

suggested MSET based circuit may still be significantly more attractive. Although we have

discussed using MSETs for implementation of logic operations, this transistor may prove

beneficial in many analog and digital applications where simple multiplexing circuits are

required.

Since the MSET operation is based on depletion, the gates and drains voltages must differ

in sign which may lead to device concatenation difficulties. This issue can be solved in the

suggested 2 drain MSETs by using a level shifter. However, in more complex devices such as

the 4 gates MSET or the 3 drains MSET, further optimization is required in order to allow

concatenation. Furthermore, in these devices the conduction channel has to have a larger

range of motion and higher gate voltages are required. Reducing the magnitude of the drains

and buffers along with detailed optimization of the device geometry and doping profiles can

reduce these issues considerably.

In order to fully exploit the functionality of MSET devices the number of states (drains)

should be maximized. This number is limited by the fabrication process and the range of

motion of the conduction channel for a specific range of gates voltages. Since the number of

possible inputs of a 4 Gates MSET is the square of the number of drains, there is great

motivation to increase the amount of drains to larger numbers than the numbers that were

discussed in this work.

References

[1] G. Shalev, G. Landman, I. Amit, Y. Rosenwaks, and I. Levy, “Specific and label-free

femtomolar biomarker detection with an electrostatically formed nanowire biosensor,” NPG Asia Mater, vol. 5, p. e41, Mar. 2013.

[2] G. Shalev, A. Doron, and A. Cohen, “Virtual semiconductor nanowire, and methods of using same,” US Pat. 8,318,505, vol. 2, no. 12, 2012.

[3] B. J. Blalock, S. Cristoloveanu, B. M. Dufrene, F. Allibert, and M. M. Mojarradi, “The multiple-gate MOS-JFET transistor,” Int. J. high speed Electron. Syst., vol. 12, no. 02,

pp. 511–520, 2002.

[4] S. Cristoloveanu, B. Blalock, F. Allibert, B. M. Dufrene, and M. M. Mojarradi, “The four-gate transistor,” in Solid-State Device Research Conference, 2002, pp. 323–326.

[5] F. Balestra, S. Cristoloveanu, M. Benachir, J. Brini, and T. Elewa, “Double-gate

silicon-on-insulator transistor with volume inversion: A new device with greatly

enhanced performance,” Electron Device Lett. IEEE, vol. 8, no. 9, pp. 410–412, 1987.

[6] A. Henning, N. Swaminathan, A. Godkin, G. Shalev, and I. Amit, “Tunable Diameter Electrostatically-formed Nanowire for High Sensitivity Gas Sensing,” pp. 1–17.

[7] B. Dufrene, K. Akarvardar, S. Cristoloveanu, B. J. Blalock, R. Gentil, E. Kolawa, and

M. M. Mojarradi, “Investigation of the four-gate action in G< sup> 4</sup>-FETs,” Electron Devices, IEEE Trans., vol. 51, no. 11, pp. 1931–1935, 2004.

[8] L. I. Glazman and I. A. Larkin, “Lateral position control of an electron channel in a

split-gate device,” Semicond. Sci. Technol., vol. 6, no. 1, p. 32, 1991.

[9] “CMOS VLSI Design: A Circuits and Systems Perspective (4th Edition): Neil Weste, David Harris: 9780321547743: Amazon.com: Books.” [Online]. Available:

http://www.amazon.com/CMOS-VLSI-Design-Circuits-Perspective/dp/0321547748.

[Accessed: 28-Jan-2015].


Recommended