High-performance microbolometer using VOx, CNT and graphene for longwave infrared (LWIR) applications

Abstract

A high-performance Microbolometer that incorporates vanadium oxide (VOx) along with carbon nanotubes (CNTs) or graphene. This Microbolometer, which uses a microbridge comprising Si3N4 and VOx, provides low noise and high dynamic range longwave infrared (LWIR) band detection. Addition of CNTs/graphene provides a high level of performance [low 1/f noise, noise equivalent temperature difference (NETD), and thermal time constant] due to the high temperature coefficient of resistance (TCR) of these materials.

Claims

1. A microbolometer structure, consisting essentially of: a Si substrate; a Si3N4 bridge layer; a VOx layer; and a layer of CNTs or a layer of graphene.

2. The microbolometer structure of claim 1, wherein the Si substrate has a thickness of approximately 500 micrometers.

3. The microbolometer structure of claim 1, wherein the Si substrate has a (100) crystal orientation according to a Miller index notation.

4. The microbolometer structure of claim 1, wherein the Si substrate includes a top layer of dry oxide SiO2 having a thickness between 200 nm to 300 nm.

5. The microbolometer structure of claim 1, wherein the Si3N4 bridge layer has a thickness of approximately 100 nm.

6. The microbolometer structure of claim 1, wherein the VOx layer has a thickness of approximately 200 nm.

7. The microbolometer structure of claim 1, wherein the VOx layer has a ratio of oxygen to vanadium (O/V) in the range of 1.8 and 1.9.

8. The microbolometer structure of claim 1, wherein the VOx layer is deposited using a DC sputtering technique having a 18:2 argon to oxygen gas flow ratio.

9. The microbolometer structure of claim 1, wherein the layer of CNTs or the layer of graphene comprises the layer of CNTs.

10. A microbolometer structure, comprising: a Si substrate; a Si3N4 bridge layer; a VOx layer; and a layer of CNTs or a layer of graphene, wherein the Si substrate has a resistivity in the range of 0.001 Ohm-cm to 0.005 Ohm-cm.

11. The microbolometer structure of claim 10, wherein the Si substrate has a thickness of approximately 500 micrometers.

12. The microbolometer structure of claim 10, wherein the Si substrate has a (100) crystal orientation according to a Miller index notation.

13. The microbolometer structure of claim 10, wherein the Si3N4 bridge layer has a thickness of approximately 100 nm.

14. The microbolometer structure of claim 10, wherein the VOx layer has a thickness of approximately 200 nm.

15. The microbolometer structure of claim 10, wherein the VOx layer has a ratio of oxygen to vanadium (O/V) in the range of 1.8 and 1.9.

16. The microbolometer structure of claim 10, wherein the VOx layer is deposited using a DC sputtering technique having a 18:2 argon to oxygen gas flow ratio.

17. The microbolometer structure of claim 10, wherein the layer of CNTs or the layer of graphene comprises the layer of CNTs.

18. A microbolometer structure, comprising: a Si substrate; a Si3N4 bridge layer; a VOx layer; and a layer of CNTs or a layer of graphene, wherein the top layer of dry oxide SiO2 has a thickness of 285 nm.

19. A microbolometer structure, comprising: a Si substrate; a Si3N4 bridge layer; a VOx layer; and a layer of CNTs or a layer of graphene, wherein the layer of CNTs are unaligned and has a thickness of 2 nm.

20. A microbolometer structure, comprising: a Si substrate; a Si3N4 bridge layer; a VOx layer; and a layer of CNTs or a layer of graphene, wherein the layer of CNTs or the layer of graphene comprises the layer of graphene, and wherein the layer of graphene is a single layer having a thickness of 0.34 nm.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

(1) The invention description below refers to the accompanying drawings, of which:

(2) FIG. 1A is an illustration of uncooled microbolometer structure with individual pixel and the array configuration

(3) FIG. 1B is a graph depicting responsivity as a function of the bias voltage;

(4) FIG. 2A depicts growth of multiwall CNTs with good length/diameter uniformity;

(5) FIG. 2B is a graph depicting reflectivity measurements for multiwalled CNTs along with preliminary results on TCR measurements;

(6) FIG. 3A is a three-dimensional temperature map of a bolometer pixel when the net absorbed power is 1 nW, and the CNT thermal resistance is 5×108 K/W. It is assumed that the pixel is tightly packed with the CNTs. The temperature gradient of the contacts is connected to the ROIC;

(7) FIG. 3B is a three-dimensional temperature map of a bolometer pixel when the net absorbed power is 1 nW, and the CNT thermal resistance is 1×109 K/W. It is assumed that the pixel is tightly packed with the CNTs. The temperature gradient of the contacts is connected to the ROIC;

(8) FIG. 4A is a top view schematic of a high-performance 17 μm pitch or smaller CNT/graphene-based microbolometer array on Si wafer substrate. The two side contacts connections adjoining the graphene/CNT microbridge sensing element provide electrical addressing of each microbolometer pixel featuring VOx and Si.sub.3N.sub.4 layers;

(9) FIG. 4B is a cross-sectional view of a CNT-based microbolometer, featuring layer of CNTs grown on layer of VOx, deposited in turn on Si.sub.3N.sub.4 bridge layer on a Si substrate;

(10) FIG. 4C is a cross-sectional view of a graphene-based microbolometer, featuring layer of graphene on layer of VOx, grown on Si.sub.3N.sub.4 bridge layer on a Si substrate;

(11) FIG. 5 depicts a test structure to evaluate the CNT and graphene film characterization;

(12) FIG. 6 is a test structure to evaluate the VOx, CNT, and graphene film characterization;

(13) FIG. 7 is a plot depicting RCDA analysis of conductivity temperature dependence;

(14) FIG. 8 is a Schematic of apparatus used for noise measurements;

(15) FIG. 9 is a graph depicting normalized waveforms analyzed for trapping at various temperatures; and

(16) FIG. 10 is a graph depicting normalized NETD/S dependence on TCR and 1/f noise changes.

DETAILED DESCRIPTION

(17) FIG. 4A is a top view schematic of a high-performance 17 μm pitch or smaller CNT/graphene-based microbolometer array 400 on Si wafer substrate 410. The two side contacts connections 420 adjoining the graphene/CNT microbridge sensing element (e.g., CNT or graphene layer on a VOx layer on a Si3N4 channel 430) provide electrical addressing of each microbolometer pixel featuring VOx and Si.sub.3N.sub.4 layers. As shown, the microbolometer 400 can include either a CNT or graphene layer on a VOx layer on a Si3N4 channel 430, as will be described with respect to FIGS. 4B-C. The microbolometer array 400 can also include a plurality of insulated supports/connections 420 for readout of the pixel graphene/sensing elements in the array. The configuration, as described below, allows a thermal time constant for 17 μm pitch arrays of below 10 ms, less than half that of current VOx microbolometer pixels.

(18) FIG. 4B is a cross-sectional view of a CNT-based microbolometer, featuring layer of CNTs grown on layer of VOx, deposited in turn on Si.sub.3N.sub.4 bridge layer on a Si substrate. As shown, a Si substrate 430 is provided. On top of the substrate 430 is a Si3N4 bridge layer 440 having a thickness of approximately 100 nm. On top of the Si3N4 bridge layer 440 is a VOx layer 450 having a thickness of approximately 250-500 nm. On top of the VOx layer 450 is a layer of CNTs (460). Each of the CNTs forming layer (460) can be randomly situated (e.g., unaligned) and can have a thickness of 2 nm.

(19) In some examples, the VOx layer 450 can have a thickness of approximately 200 nm, i.e., 200 nm+/−5% depending on manufacturing tolerances. The VOx layer 450 can also have a ratio of oxygen to vanadium (O/V) in the range of 1.8 and 1.9. This O/V ratio can be achieved according to a DC sputtering technique described below.

(20) The thin film layer of VOx can be deposited according to a DC (direct current) sputtering technique. The sputtering technique can include providing a vanadium target. Next, the vanadium target can be bombarded with ionized gas molecules. In this example, ionized gas particles can be provided at an 18:2 argon to oxygen gas flow ratio to allow control and variability of oxygen to vanadium (O/V) concentration in order to achieve the range of 1.8 to 1.9 O/V described above. When the vanadium target is bombarded with the ionized gas molecules, the vaporized atoms are subsequently condensed onto the Si substrate as the VOx thin film layer.

(21) In some examples, the Si substrate 430 can have a (100) crystal orientation, according to Miller index notation, and can have a singled polished side (the side opposite the side on which the VOx is deposited). The Si substrate can have a low resistivity in the range of 0.001 Ohm-cm to 0.005 Ohm-cm. The Si substrate can have a diameter of at least 100 mm and a thickness of approximately 500 micrometers. The substrate can include a top layer (e.g., on which the VOx is deposited) of deposited dry oxide SiO2 having thickness between 200 to 300 nm, and in one particular example a thickness of 285 nm.

(22) FIG. 4C is a cross-sectional view of a graphene-based microbolometer, featuring layer of graphene on layer of VOx, grown on Si.sub.3N.sub.4 bridge layer on a Si substrate. As shown, a Si substrate 465 is provided. On top of the substrate 465 is a Si3N4 bridge layer 470 having a thickness of approximately 100 nm. On top of the Si3N4 bridge layer 470 is a VOx layer 475 having a thickness of approximately 250-500 nm. On top of the VOx layer 475 is a layer of graphene 480. The layer of graphene 480 can be a single layer (e.g. monolayer) having a thickness of 0.34 nm.

(23) In some examples, the VOx layer 475 can have a thickness of approximately 200 nm, i.e., 200 nm+/−5% depending on manufacturing tolerances. The VOx layer 475 can also have a ratio of oxygen to vanadium (O/V) in the range of 1.8 and 1.9. This O/V ratio can be achieved according to a DC sputtering technique described below.

(24) The thin film layer of VOx can be deposited according to a DC (direct current) sputtering technique. The sputtering technique can include providing a vanadium target. Next, the vanadium target can be bombarded with ionized gas molecules. In this example, ionized gas particles can be provided at an 18:2 argon to oxygen gas flow ratio to allow control and variability of oxygen to vanadium (O/V) concentration in order to achieve the range of 1.8 to 1.9 O/V described above. When the vanadium target is bombarded with the ionized gas molecules, the vaporized atoms are subsequently condensed onto the Si substrate as the VOx thin film layer.

(25) In some examples, the Si substrate 465 can have a (100) crystal orientation, according to Miller index notation, and can have a singled polished side (the side opposite the side on which the VOx is deposited). The Si substrate can have a low resistivity in the range of 0.001 Ohm-cm to 0.005 Ohm-cm. The Si substrate can have a diameter of at least 100 mm and a thickness of approximately 500 micrometers. The substrate can include a top layer (e.g., on which the VOx is deposited) of deposited dry oxide SiO2 having thickness between 200 to 300 nm, and in one particular example a thickness of 285 nm.

(26) The examples of FIGS. 4A-C advantageously provide a microbolometer structure/array that achieve a TCR in the range of 3%/K to 5%/K TCR and 1/f noise Hooge parameter α.sub.H/p≈1×10-24 cm3, enabling a noise equivalent temperature difference (NETD) below 25 mK for 17 μm pitch and 12 μm LWIR focal plane arrays. In one example, this can be achieved in the 1-10 kHz range.

(27) In order to measure devices over a broad resistance range and to evaluate surface effects and contact resistance, resistor structures are characterized with varying aspect ratios as shown in 500 and 600 in FIG. 5 and FIG. 6. Resistance, TCR, and noise measurements are performed with the samples placed in a dark, cold-shielded cryostat to ensure isothermal measurements with the sample in thermal equilibrium.

(28) Sample resistance R(T) at various temperatures is obtained by recording the current-voltage (I-V) characteristics, determining the Ohmic region, and extracting the resistance. Contact resistance, drift, and thermal hysteresis effects are measured and analyzed. The TCR at a specific temperature T is experimentally obtained by numerical differentiation of the temperature dependence of the sample resistance R(T):

(29) $\begin{array}{cc}{\left(TCR\right)}_{T_i}=\frac{1}{R_i}{\left\{\frac{\mathrm{dR}}{\mathrm{dT}}\right\}}_i=\frac{1}{R_i}\left\{\frac{1}{2}\left(\frac{R_{i+1}-R_i}{T_{i+1}-T_i}+\frac{R_i-R_{i-1}}{{\tau }_{i-T_{i-1}}}\right)\right\}& \left(1\right)\end{array}$

(30) Electrical transport in semiconductors is via band states or hopping in localized states. The temperature dependent conductivity, σ(T), generally follows the model:
(TCR).sub.T.sub.i=σ.sub.0e.sup.−(T.sup.0.sup./T).sup.p  (2)

(31) where T0 is the characteristic temperature, T is the material temperature, and σo is the conductivity prefactor. The value of the exponent p depends on the details of the density of states (DOS) as affected by long range Coulomb interaction among the charge carriers. Strong Coulomb interaction results in a gap in the density of states, generally referred to as the Coulomb gap.

(32) In such a case, p=½ and the conductivity is described by the Efros-Shklovskii Variable Range Hopping (VRH) model. If the Coulomb interaction is negligible, the conductivity follows the Mott VRH model and p=¼. In most materials the Coulomb gap is small, and Efros-Shklovskii type conductivity is observed at low temperature. For intermediate temperatures, there is a crossover from Efros-Shklovskii to Mott conduction, and, at high temperatures, both the Mott and Efros-Shklovskii transition to Nearest Neighbor Hopping (NNH) conduction where p=1.

(33) For conduction via extended states the exponent is also p=1 and T0=E/kB, where E is the thermal activation energy and kB the Boltzmann constant.

(34) In order to determine which conductivity mechanism dominates, a technique known as Resistance Curve Derivative Analysis (RCDA) method is used. In the RCDA method, the logarithmic conductivity data with temperature is numerically differentiated, where the quantity w is determined:

(35) $\begin{array}{cc}w\left(T\right)=\frac{d\left(\log \sigma \right)}{d\left(\log T\right)}={p\left(\frac{T_0}{T}\right)}^p& \left(3\right)\end{array}$

(36) The value p is obtained by the slope and the characteristic temperature TO from the y-intercept when plotting:
log w=−p log T+log p(T.sub.0).sup.p  (4)

(37) The conductivity prefactor σ0 can then be calculated from the conductivity expression. An example of the RCDA conductivity analysis is shown in FIG. 7. The exponent p=0.58 is indicative of Efros-Shklovskii VRH conduction, where p=½. The characteristic temperature is T0=8.24×104 K, and the conductivity prefactor is σ0=1.06×106 (Ω.Math.cm)−1.

(38) The TCR depends on the conductivity regime described by Eq. (3) and specified by p and the characteristic temperature TO in that regime, as shown by Eq. (5):

(39) $\begin{array}{cc}\left|\mathrm{TCR}\right|=\left|\frac{1}{R}\frac{\mathrm{dR}}{\mathrm{dT}}\right|=\left|\frac{1}{\sigma }\frac{d\sigma }{\mathrm{dT}}\right|=p\frac{T_0^p}{T^{p+1}}& \left(5\right)\end{array}$

(40) The temperature dependence of conductivity is measured and the TCR and hopping parameters p, σ0, and T0 are determined and tracked as a function of film structure, deposition conditions, and postdeposition processing.

(41) A schematic of the noise measurements is shown in FIG. 8. The measured noise Vm is:
V.sub.m.sup.2=V.sub.R.sub.Feedback.sup.2+(GV.sub.n).sup.2  (6)

(42) where VRFeedback is the Johnson noise of the amplifier feedback resistor, Vn is the sample noise, and G is the amplifier gain. The gain is the ratio of the amplifier output signal voltage Vout to the input signal voltage Vin, which is equal to the ratio of the feedback resistance to the sample resistance R:

(43) $\begin{array}{cc}G=\frac{V_{out}}{V_{in}}=\frac{R_{\mathrm{Feedback}}}{R}& \left(7\right)\end{array}$

(44) The system noise is also characterized by using metal film resistors whose resistances are comparable to the devices tested. Using the metal film resistor noise results, the low frequency noise of the cross-linked metal particles film noise and system noise can be quantified.

(45) Experimental room temperature noise measurements are acquired in a shielded and light-tight enclosure. For temperature-dependent noise measurements, the film is mounted within a cryostat. All measurements are made with the films in dark conditions. The gain is adjusted by use of a variable feedback resistor within the preamplifier.

(46) Generally, in semiconductors, the total noise measured is comprised of 1/f noise, white noise, and potentially generation-recombination (G-R) noise, which can be expressed quantitatively as:

(47) $\begin{array}{cc}\frac{V_{nois{e}^2}}{\Delta f}=\frac{V_{1/f}^2}{\Delta f}+\frac{{V_{}}_{\mathrm{white}}^2}{\Delta f}+\frac{V_{G-R}^2}{\Delta f}& \left(8\right)\end{array}$

(48) The 1/f noise can be described using the Hooge model:

(49) $\begin{array}{cc}\frac{V_{1/f}^2}{\Delta f}=\frac{{\alpha }_H}{P}\frac{{V_{}}_{\mathrm{Bias}}^2}{f^{\gamma }}=\frac{{\alpha }_H}{pV}\frac{{V_{}}_{\mathrm{Bias}}^2}{f^{\gamma }}=\frac{B}{f^{\gamma }}& \left(9\right)\end{array}$

(50) where

(51) $\begin{array}{cc}B=\frac{{\alpha }_H}{pV}{V}_{Bias}^2& \left(10\right)\end{array}$

(52) and P is the total number of carriers, p is the carrier density, V is the device volume, VBias is the device bias, and γ is the frequency exponent typically near a value of 1 (in the range of 0.9 to 1.1). The normalized Hooge parameter is α_H/p, where is the Hooge coefficient. For the Hooge model:
√{square root over (B)}∝V.sub.Bias and B/V.sub.Bias.sup.2∝1/V for fixed bias voltages  (11)

(53) The G-R noise for a single trap is described by:

(54) $\begin{array}{cc}\frac{V_{G-R}^2}{\Delta f}=\frac{A\tau V_{\mathrm{Bias}}^2}{1+{\left(2\pi \tau f\right)}^2}& \left(12\right)\end{array}$

(55) where τ is the trap time constant and A=n_t/(Vp{circumflex over ( )}2). Here, n.sub.t is the trap density, V is the device volume, and p is the carrier density. The noise waveform can be analyzed for trapping by:

(56) $\begin{array}{cc}W\equiv f\frac{\left(V_{n\mathrm{oise}}^2-V_{w\mathrm{hite}}^2\right)}{V_{\mathrm{Bias}}^2}=\frac{\alpha }{P}+\frac{f{A}_t\tau }{1+{\left(2\pi \tau f\right)}^2}& \left(13\right)\end{array}$

(57) Here it is assumed that the exponent of the 1/f noise frequency is equal to 1. For the case of a single trap, a plot of [f(V.sub.noise.sup.2−V.sub.white.sup.2)/V.sub.Bias.sup.2]V (determined from the measured noise data) yields a horizontal line (of value α_H/P) superimposed with a Lorentzian (trap).

(58) FIG. 9 shows an example of W plot as a function of frequency for noise data taken at different temperatures. For temperatures of 360 K and 380 K, the curves have no slope, indicative that the measured low frequency noise is 1/f noise only. The values of these curves are equal to α_HT. For temperatures 400 K and above, the curves at low frequency (less than 10 Hz) increase with increasing temperature.

(59) The temperature dependence of VOx/CNT films with varying packing densities of CNTs into the VOx is measured and analyzed to determine the conduction mechanism and to optimize and control resistivity, TCR, and noise.

(60) A baseline of microbolometer performance for VOx, CNT and Graphene based devices to model their expected performance is the Figure of Merit (FOM) of microbolometer imaging devices, which at the pixel level is:
FOM=NETD×τ.sub.th  (14)

(61) where the Noise Equivalent Temperature Difference (NETD) and thermal time constant τ_th are described by Eq. (15) and Eq. (16), respectively:

(62) 0$\begin{array}{cc}NE\mathrm{TD}=\frac{1}{\frac{1}{4F^2+1}\frac{{\mathrm{dP}}_s}{{\mathrm{dT}}_s}}\frac{G_{th}}{\eta A_D}\frac{V_{\mathrm{noise}}}{\mathrm{TCR}.Math.V_{B\mathrm{ias}.}}& \left(15\right)\end{array}$$\begin{array}{cc}{\tau }_{th}=\frac{C_{p}m}{G_{th}}& \left(16\right)\end{array}$

(63) (The thermal time constant τ_th is measured by applying pulsed bias and measuring the pixel temperature rise and decay and monitoring the emitted infrared radiation with a fast LWIR detector.) The TCR, Vnoise, and heat capacity Cp are material parameters, whereas pixel mass m, pixel area AD, and thermal conductance Gth are device parameters. The absorptance η is a material parameter dependent on the material optical constants n+ik, and a device parameter dependent on the pixel resonant cavity design.

(64) For a VOx-CNT or VOx-Graphene microbolometer design, the pixel material parameters to be modified by the addition of CNTs or Graphene are Vnoise and TCR. In the case that the dominant noise is 1/f, Eq. (15) becomes:

(65) $\begin{array}{cc}\mathrm{NETD}=\frac{1}{\frac{1}{4F^2+1}\frac{{\mathrm{dP}}_s}{{\mathrm{dT}}_s}}\frac{G_{th}}{\eta A_D}\frac{1}{\sqrt{\mathrm{Vol}f}}\frac{\sqrt{\frac{{\alpha }_H}{n}}}{\mathrm{TCR}}\equiv S\frac{\sqrt{\frac{{\alpha }_H}{n}}}{\mathrm{TCR}}& \left(17\right)\end{array}$

(66) FIG. 10 shows that the NETD changes linearly with the square root of the normalized Hooge parameter αH/n, and inversely proportionally with TCR. The NETD/S ratio is shown for the expected range of αH/n, and TCR of a composite VOx pixel modified by the addition of CNTs or graphene.

(67) TCR values reported for CNT based microbolometers with different architectures vary in the range of 0.14%/K to 2.94%/K. These values are comparable to the TCRs of VOx films. CNTs and/or graphene can form composite films with VOx by deposition on the surface of a VOx film, or by incorporation into the VOx film.

(68) Placing CNTs on the surface of VOx films will modify the conductivity and TCR of the composite structure according to the conductivities of the VOx and CNT layers:
σ.sub.tot=σ.sub.VOx+σ.sub.CNT  (18)

(69) $\begin{array}{cc}{.Math.\mathrm{TCR}.Math.}_{tot}=\frac{{\sigma }_{\mathrm{VOx}}}{{\sigma }_{\mathrm{VOx}}+{\sigma }_{ES}}{.Math.\mathrm{TCR}.Math.}_{\mathrm{VOx}}+\frac{{\sigma }_{CNT}}{{\sigma }_{\mathrm{VOx}}+{\sigma }_{CNT}}{.Math.\mathrm{TCR}.Math.}_{CNT}& \left(19\right)\end{array}$

(70) where σVOx is the initial VOx film conductivity and σCNT is the initial CNT overlayer conductivity.

(71) The VOx and CNT films are assumed to be noninteracting and conducting in parallel. The CNT conductivity and TCR vary with the CNT packing density of the CNT film structure, and consequently the conductivity and TCR of the composite VOx-CNT film will vary with the CNT packing density. When CNTs are incorporated in a VOx film, the composite film conductivity and TCR depends on the various conduction paths, i.e., VOx conduction, VOx to CNT conduction, or CNT to CNT conduction. The predominant conduction pathway depends on the CNT packing density in the VOx film.

(72) The foregoing has been a detailed description of illustrative embodiments of the invention. Various modifications and additions can be made without departing from the spirit and scope of this invention. Features of each of the various embodiments described above may be combined with features of other described embodiments as appropriate in order to provide a multiplicity of feature combinations in associated new embodiments. Furthermore, while the foregoing describes a number of separate embodiments of the apparatus and method of the present invention, what has been described herein is merely illustrative of the application of the principles of the present invention. Accordingly, this description is meant to be taken only by way of example, and not to otherwise limit the scope of this invention.