Random telegraph noise native device for true random number generator and noise injection

Abstract

A method (and system) for generating random numbers includes setting a drain voltage Vd on an MOSFET (metal oxide semiconductor field effect transistor) device and a gate voltage Vg of the MOSFET device so that the MOSFET device comprises a noise source configured in a manner such as to tune as desired a random number statistical distribution of an output of the MOSFET device. An output voltage of the MOSFET is provided as an input signal into a low noise amplifier and an output voltage of the low noise amplifier provides values for a random number generator.

Description

BACKGROUND OF THE INVENTION

Field of the Invention

The present invention generally relates to a device for generating true random numbers. More particularly, artificial traps are created in a native device for different random number generations and voltage is used to tune the tunneling time of trapping/de-trapping, and the native device is biased to maximize the trapping/detrapping impacts so as to amplify the random noise.

Description of the Related Art

In recent years, stochastic circuits have become important because they are used in many applications, including for example, security encryption and accelerators for neural networks. Additionally, probability representations made a lot of random numbers. And, noise injection is an important nub for Bayesian learning. However, the PDF of random numbers should be controlled. Probability some and product should be implemented efficiently. Conditional probabilities should be generated efficiently. Existing solutions for random noise generation often uses noise sources from thermal noise, RTM, or other entropy sources. The system is tuned to statistics and then generate random numbers. FIG. 1A shows an exemplary conventional random number generator circuit 100.

This conventional circuit 100 has several problems. First, there is no direct control in the entropy source. Second, because of the small noise generator 102, a relatively large and complicated system is required, including, for example, a relatively large analog-to-digital (AC) circuit 104 comprising an amplifier, filter, and comparator, which increases area and cost.

SUMMARY OF THE INVENTION

In view of the foregoing and other problems and concerns, the present invention provides a device and circuit for generating random numbers, in which artificial traps are engineered and which have a controlled tuning process to generate the random numbers. The native devices of the present invention can be customized MOS (Metal Oxide Semiconductor). The traps can be generated through STM (scanning tunneling microscopy), quantum dots, or nanocrystalline deposition.

In a first exemplary aspect of the present invention, described herein is a random telegraph noise native device including a source region and a drain region and a gate structure formed between the source and drain regions, wherein the gate structure includes an oxide layer comprising at least one artificial trapping layer in which carrier traps are designed to occupy a predetermined distance from conduction and valance bands of material of the artificial trapping layer.

In a second exemplary aspect, the present invention provides a method for a random number generator that can be tuned for specific random number statistical characteristics.

These and many other advantages may be achieved with the present invention.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing and other exemplary purposes, aspects and advantages will be better understood from the following detailed description of an exemplary embodiment of the invention with reference to the drawings, in which:

FIG. 1A illustrates an exemplary conventional random number generator that uses a small noise source that requires a relatively large amplifier and filter circuits;

FIG. 1B illustrates a comparable circuit of an exemplary embodiment of the present invention, using a relatively smaller amplifier and filter circuits;

FIG. 2 shows exemplarily the concept of controlling tunneling time 200;

FIG. 3 shows exemplarily the concept of artificial traps 300 for noise engineering;

FIG. 4 shows exemplarily a MOS-type device 200 having at least one trapping layer 206 in the gate oxide structure 204;

FIG. 5A illustrates an exemplary embodiment 500 of a noise generator circuit of the present invention using Ig-RTN;

FIG. 5B illustrates an exemplary embodiment 550 of a noise generator circuit of the present invention using Id-RTN; and

FIG. 6 illustrates how the random noise generator circuits of the present invention can be selectively biased and tuned.

DETAILED DESCRIPTION OF EXEMPLARY EMBODIMENTS OF THE INVENTION

Referring now to the drawings, and more particularly to FIGS. 1B-6, exemplary embodiments of the method and structures of the present invention will now be described.

As shown exemplarily in FIG. 1B, an exemplary embodiment of the present invention uses a large noise source device 152, which generates large noise at high frequency, which, in turn, permits a smaller amplifier and filter for the A/D converter 154. These smaller dimensions for the A/D converter downsize the circuit area of the present invention to 1200 mm², which provides a savings in circuit area of the conventional circuit 100 shown in FIG. 1A.

The large noise source 152 of the present invention uses native device trapping, which will be explained shortly.

FIG. 2 shows exemplarily the concept of controlling of tunneling time. On the left side is shown how different trap depths change tunneling times, which aspect provides different noise signals shown on the right side of FIG. 2. There are several different parameters known in the art the control tunneling time, including temperature, voltage, and oxide layer composition (e.g., high K or other material). Prior art methods relied upon temperature for changing tunnel times. In contrast, the present invention uses voltage to change tunneling time and engineer the random variable shown on the right side of FIG. 2.

FIG. 3 shows exemplarily how artificial traps can be engineered to control tunneling time by change location of the traps relative to the conduction and valence bands; the closer the traps are located to the conduction and valence bands, the more often emissions occur.

FIG. 4 shows exemplarily an MOS-type RTN native device 400 having source/drain 402, 404 and gate structure 406 with at least one artificial trapping layer 408 in the gate oxide.

Such MOS-type structure 400 permits small capacitance for trapping, ΔQ=CΔV. In most implementations of the present invention, the traps will be intentionally created close to Ef. There can be multiple artificial trapping layers 408, or only a single artificial trapping layer, in the gate oxide. With single trap layer, the output can only have two different states. With multiple layers, the output can have more than two states, which feature enhances the information (data) delivered from the same amount of devices. The MOS structure permits small width and short channel, which is ideal for high density circuit integration.

The artificial trapping layer(s) 408 can exemplarily be fabricated by such any method that permits traps to be selectively located relative to the conduction and valence bands. For example, high density Ru nanocrystals can be deposited, much as currently done for non-volatile memory applications. Other methods such as STM (scanning tunnel microscopy), quantum dots, other nanocrystal deposition can be used for generating the artificial trapping layer(s).

FIG. 5A shows exemplarily how the MOS RTN native device 502 is connected to provide Ig as an output signal for low noise amplifier (LNA) 504 to provide output signal Ig*R, where R is the feedback resistance of the LNA. The control parameters 506 are voltages Vd and Vg.

FIG. 5B shows how the RTN native device 552 is connected to provide Id as input into LNA 554, and control parameters 556 are again Vd and Vg. The output signal in this configuration is Id*R.

FIG. 6 shows details related to the control parameters Vd and Vg shown in FIGS. 5A and 5B. Thus, as shown in FIG. 6, the circuit of FIG. 5A can be controlled by biasing the noise-source transistor to be in the maximum gm regime, meaning that Vd is chosen to maximize gm. To modify the random number behavior, the statistical distribution, Vg is chosen to tune the tunneling time and thereby control the random variable's probability function.

While the invention has been described in terms of several exemplary embodiments, those skilled in the art will recognize that the invention can be practiced with modification.

Further, it is noted that, Applicants' intent is to encompass equivalents of all claim elements, even if amended later during prosecution.

Claims

1. A semiconductor device, comprising: a substrate;a MOS (metal oxide semiconductor) device comprising a source region and a drain region embedded in the substrate and a gate structure formed between the source and drain regions;an operational amplifier formed on the substrate; anda resistance R interconnected between a first input terminal and an output terminal of the operational amplifier, such that an input terminal of the operational amplifier directly connects to one terminal of the resistance R and is also directly connected to a contact of one of the drain region and the gate structure of the MOS device, the operational amplifier and resistor thereby forming a low noise amplifier for the MOS device,wherein the MOS device is configured as a noise source transistor in such a manner that a random number tuned to a random number statistical distribution of voltage values is provided at the output terminal of the operational amplifier by setting a drain voltage Vd on the MOS device and a gate voltage Vg of the MOS device so as to tune as desired a random number statistical distribution of an output of the MOS device.

2. The semiconductor device of claim 1, wherein the gate structure comprises an oxide layer comprising at least one artificial trapping layer in which carrier traps are designed to occupy a predetermined distance from conduction and valance bands of material of the artificial trapping layer, and wherein each of the at least one trapping layer is fabricated using one of nanocrystal deposition, scanning tunnel microscopy, E-beam, and quantum dots formation.

3. The semiconductor device of claim 1, wherein the MOS device comprises a noise source that is a high level noise source relative to a noise level of the low noise amplifier.

4. The semiconductor device of claim 1, as configured so that an amount of a gate current Ig of the MOS device passes through the resistance R, so that an output voltage of the low noise amplifier V=Ig*R, and wherein a voltage at a second input terminal of the operational amplifier sets a voltage Vg for the gate of the MOS structure.

5. The semiconductor device of claim 4, wherein a voltage at the drain of the MOS is set at a voltage value predetermined to maximize a transconductance gm of the MOS device and wherein Vg is selectively set to tune a random number statistical distribution of voltage values of the output voltage of the operational amplifier.

6. The semiconductor device of claim 1, as configured so that an amount of a drain current Id of the MOS device passes through the resistance R, so that an output voltage of the low noise amplifier V=Id*R, and wherein a voltage at a second input terminal of the operational amplifier sets a voltage Vd for the drain of the MOS structure.

7. The semiconductor device of claim 6, wherein the voltage Vd at the drain of the MOS is set at a voltage value predetermined to maximize a transconductance gm of the MOS device and wherein Vg is selectively set to tune a random number statistical distribution of voltage values of the output voltage of the operational amplifier.

8. A random number generator, comprising: a random telegraph noise native device;a low noise amplifier connected to an output of the random telegraph noise native device;a filter; anda comparator,wherein the low noise amplifier, the filter, and the comparator are configured to form an analog-to-digital (A/D) converter for an output signal from the random telegraph noise native device,wherein the random telegraph noise native device is configured as a noise source transistor in such a manner that a random number tuned to a random number statistical distribution of voltage values is provided at an output terminal of the low noise amplifier, andwherein an input terminal of the operational amplifier directly connected to one terminal of the resistance R to form the low noise amplifier is also directly connected to a contact of one of the drain region and the gate structure of the MOS device.

9. The random number generator of claim 8, wherein the random telegraph noise native device comprises one of an MOSFET (metal oxide semiconductor field effect transistor) structure and an RRAM (resistive random access memory) structure.

10. The random number generator of claim 8, wherein the random telegraph noise native device comprises a semiconductor device comprising: a substrate; a source region and a drain region embedded in the substrate; and a gate structure formed between the source and drain regions, wherein the gate structure comprises an oxide layer comprising at least one artificial trapping layer in which carrier traps are designed to occupy a predetermined distance from conduction and valance bands of material of the artificial trapping layer.

11. The random number generator of claim 10, wherein each of the at least one trapping layer is fabricated using one of nanocrystal deposition, scanning tunnel microscopy, E-beam, and quantum dots formation.

12. The random number generator of claim 10, wherein the source and drain regions and the gate structure thereby form a MOS (metal oxide semiconductor) device, and wherein the low noise amplifier comprises: an operational amplifier formed on the substrate; and a resistance R interconnected between a first input terminal and an output terminal of the operational amplifier, and wherein the MOS comprises a large noise source relative to the low noise amplifier.

13. The random number generator of claim 12, as configured so that an amount of a gate current Ig of the MOS device passes through the resistance R, so that an output voltage of the low noise amplifier V=Ig*R, and wherein a voltage at a second input terminal of the operational amplifier sets a voltage Vg for the gate of the MOS structure.

14. The random number generator of claim 13, wherein a voltage at the drain of the MOS is set at a voltage value predetermined to maximize a transconductance gm of the MOS device and wherein Vg is selectively set to tune a random number statistical distribution of voltage values of the output voltage of the operational amplifier.

15. The random number generator of claim 12, as configured so that an amount of a drain current Id of the MOS device passes through the resistance R, so that an output voltage of the low noise amplifier V=Id*R, and wherein a voltage at a second input terminal of the operational amplifier sets a voltage Vd for the drain of the MOS structure.

16. The random number generator of claim 15, wherein the voltage Vd at the drain of the MOS is set at a voltage value predetermined to maximize a transconductance gm of the MOS device and wherein Vg is selectively set to tune a random number statistical distribution of voltage values of the output voltage of the operational amplifier.

17. A method of generating random numbers using a semiconductor device comprising a metal oxide semiconductor field effect transistor (MOSFET) device configured to be a noise source transistor into a low noise amplifier formed having an input terminal of the operational amplifier directly connected to one terminal of a resistance R that forms the low noise amplifier and also directly connected to a contact of one of the drain region and the gate structure of the MOSFET device, the method comprising: setting a drain voltage Vd on an MOSFET (metal oxide semiconductor field effect transistor) device and a gate voltage Vg of the MOSFET device so that the MOSFET device comprises a noise source configured in a manner such as to tune as desired a random number statistical distribution of an output of the MOSFET device; andproviding an output voltage of the MOSFET as an input signal into a low noise amplifier, an output voltage of the low noise amplifier providing values for a random number generator.

18. The method of claim 17, wherein the MOSFET device comprises a gate structure comprising an oxide layer including at least one artificial trapping layer in which carrier traps are designed to occupy a predetermined distance from conduction and valance bands of material of the artificial trapping layer.

19. The method of claim 18, wherein each of the at least one trapping layer is fabricated using one of nanocrystal deposition, scanning tunnel microscopy, E-beam, and quantum dots formation.

20. The method of claim 17, wherein the low noise amplifier comprises a component of an analog-to-digital (A/D) converter.