Design and characterization of Adaptive microbolometers

7/28/2019 Design and characterization of Adaptive microbolometers

1/7

INSTITUTE OF PHYSICS PUBLISHING JOURNAL OF MICROMECHANICS AND MICROENGINEERING

J. Micromech. Microeng. 16 (2006) 10731079 doi:10.1088/0960-1317/16/5/028

Design and characterization of adaptivemicrobolometers

Woo-Bin Song1 and Joseph J Talghader2

1 School of Electrical and Computer Engineering, Purdue University, West Lafayette,

IN 47907, USA2 Department of Electrical and Computer Engineering, University of Minnesota,

Minneapolis, MN 55455, USA

E-mail: [email protected] and [email protected]

Received 16 October 2005, in final form 5 January 2006Published 19 April 2006Online at stacks.iop.org/JMM/16/1073

AbstractWe report the first practical results on design and characterization ofadaptive microbolometers with a thermally tuned responsivity. Such devicesare needed to simultaneously image scenes that contain objects at ambientand extremely hot temperatures. In the high detectivity state, themicrobolometers are operated similar to standard commercial devices. Inthe low detectivity state, portions of the support beams are brought incontact with the substrate, which thermally shorts the devices on apixel-by-pixel basis. This is the first time this concept has been practicallyimplemented in a microbolometer device architecture as opposed to simplecantilever beams and plates. The maximum actuation voltage is set to 17 V,and the thermal conductances, responsivities and detectivities of a typicaldevice can be switched more than an order of magnitude between 1.65

105 W K1 and 2.99 104 W K1, between 1.5 V W1 and 0.2 V W1and between 1.8 106 cm Hz1/2 W1 and 1.5 105 cm Hz1/2 W1,respectively. This extends the dynamic range of the device more than anorder of magnitude. The device pixel size is 100 m 100 m.

1. Introduction

Photon detectors based on small bandgap materials suchas Hg0.8Cd0.2Te (MCT), InSb and others have been studied

and commercialized primarily for military and space scienceapplications [1]. These photon detectors have very high

performance but require cryogenic cooling systems to reduceparasitic dark currents.

With the advent of micromachining technology, it wasrealized that sufficiently high performance could be achievedwith uncooled detectors that operated based on the heating

of a small mass. Several classes of devices for readingthe rise in temperature were developed. First and most

prominent of these was the bolometer [2, 3] butalso developedwere pyroelectric detectors [4], thermoelectric detectors [5]

and others [6]. The cost and weight of these devices aredramatically reduced compared to infrared photon detectors

because cooling is no longer needed and because the materialsinvolved are much easier to work with. For example, Wood

et al [2] reported high-quality infrared imaging in an infraredfocal plane array using a monolithic silicon-based process.

Recent studies have attempted to improve the performance ofthe basic microbolometer by developing high temperature-coefficient-of-resistance (TCR) materials such as high-Tcsuperconductor (YBaCuO [7]), vanadium oxide [8] and so

on [9, 10]. Also new device structures have been proposed toenhance sensitivity [11], thermal isolation [12] and array fill

factor [13]. Furthermore, new devices using built-in thermalchopping have been proposed [1416] including a focal plane

array with a built-in thermal chopper and a two-color thermaldetector with a built-in thermal chopper.

Typical thermal detectors are designed for high thermalisolation, and this makes them extremely receptive to smalldifferences in background temperature. However, high

sensitivity potentially has drawbacks when a scene containsvery bright areas that may result from fires, explosions orengine operation. For example, microbolometer arrays thatare tracked across the sun develop a temporary afterimage

or blindness that diminishes over a period of many minutesto a few hours. One of the causes for this is that the mostcommon TCR material in microbolometers, vanadium oxide,is thermochromic, meaning that its absorption and resistance

0960-1317/06/051073+07$30.00 2006 IOP Publishing Ltd Printed in the UK 1073
http://dx.doi.org/10.1088/0960-1317/16/5/028mailto:[email protected]:[email protected]://stacks.iop.org/JMM/16/1073http://stacks.iop.org/JMM/16/1073mailto:[email protected]:[email protected]://dx.doi.org/10.1088/0960-1317/16/5/028


2/7

W-B Song and J J Talghader

properties change if the microbolometer is heated higher than

a semiconductormetal phase transition temperature of about

68 C [17]. Since many microbolometers heat up by about

100 C when imaging a 3000 K signal, which is typical of

many fires, one can see that thermochromic transitions can

be frequent. In addition, special electronics must be used

to amplify the signals at the edges of the array in order

to handle such a large dynamic range. Intensity-dependent

noise and device damage are other possible issues. Previous

research in our group has shown that the thermal conductance

[18] of micromachined beams and plates can be tuned both

continuously and discretely. Continuous tuning of the thermal

conductance was demonstrated over a factor of 3 and discrete

tuning over almost two orders of magnitude. While these

initial studies were suggestive, no electronic devices were built

to take advantage of the effect.

In this paper, we report a microbolometer device

architecture that incorporates an adaptive responsivity [18].

These are the first array-compatible microbolometer devices

to use microactuation to tune device thermal conductance on

a pixel-by-pixel basis. This is a significant innovation in both

concept and function because it prevents thermal runaway in

microbolometers due to intense signals and allows one to see

in detail both hot and cold regions of a scene in the same

focal plane array. The required actuation voltage to change

responsivity is only 17 V, making the devices compatible with

common change pumping circuits. For simplicity, the

bolometers are operated to have a binary responsivity: high

and low for cool and hot signals, respectively. However, if

one supplied multiple voltages to the devices, a variety of

responsivities would be possible. The devices do not require

major changes to standard commercial architectures but rather

use an already existing electrostatic actuator between the

ground electrode and detector plate to bring a small area of

the supports into contact with the substrate, thereby increasing

the device thermal conductance.

First, we discuss the basics of an adaptive microbolometer

in the background. Then, the fabrication of the device

is explained in detail. To fully characterize this adaptive

microbolometer, measurements of the thermal conductance,

responsivity, detectivity and noise equivalent power have been

made with and without electrostatic actuation.

2. Principles of bolometer operation

The responsivity of a thermal detector is expressed as

= IR0T

Pi=

IR0

G(1 +22)12

(VW1), (1)

where is the TCR, Iis the bias current, R0 is the resistance of

the detector at ambient temperature, T is the temperature

increase of the bolometer, Pi is the power of the incident

radiation, is the fraction of the incident radiation absorbed,

G is the thermal conductance of the microbolometer, is the

frequencyof the sinusoidal excitation and is the thermal time

constant. Since the responsivity as shown in (1) is inversely

proportional to its thermal conductance, the detectors must

be thermally isolated from their surroundings. As implied in

the previous section, a good thermal isolation can be achieved

by having a vacuum gap between the sensing element and

the substrate with only thin support beams connecting them.

In vacuum, absorbed heat on the pixel plate from the incident

radiation dissipates to the substrate through the support beams.

With this, the thermal conductance for the device in the

radiation limit can be written as

G = 2k A

L

, (2)

where G is the thermal conduction through twosupport beams,

kis the thermal conductivity of a support beam,A is thevertical

cross-sectional area of a support beam and L is the length of

a support beam. The factor of 2 comes from the two support

beams of the devices. Based on (2), we can reduce the thermal

conductance by using materials with low thermal conductivity

and having long support beams with a small cross-sectional

area. To ensure that the thermal conductance is determined

by the support design, detector arrays are packaged under low

vacuum conditions in hermetically sealed packages.

Since the thermal conductance of a typical

microbolometer is limited by the solid-state conduction

of the supports, conductances are on the order of 10

7

WK1 for commercial systems. In theory, the conductance

of a microbolometer can be reduced so that it is limited by

radiation, obtaining a value on the order of 108 W K1.

Since it has been shown that the thermal conductance across

a microactuated interface is on the order of 104 W K1, the

fundamental limit of thermal conductance tuning is almost

four orders of magnitude. In practice, stiction in the high

conductance state is likely to limit the maximum thermal

conductance tuning that can be achieved to one or two orders

of magnitude.

The performance of a detector can be also determined by

its detectivity, which indicates the relative magnitudes of the

signal and noise at operating frequencies. The detectivity can

be calculated from the expression

D =(A)1/2(fbandwidth)

1/2V

Vnoise2

, (3)

where RV is the responsivity, A is the area of a pixel plate,

fbandwidth is the electrical bandwidth and V2noise is the mean

square value of the fluctuating noise voltage.

As the last important figure of merit, there is the noise

equivalent power (NEP) defined as the absorbed power change

which corresponds to a signal-to-noise ratio of unity. TheNEP

is obtained by measuring thenoise voltage andtheresponsivity

as shown in (4)

NEP =

V2noise

V, (4)

where V2noise is the mean square value of the fluctuating noise

voltage and RV is the responsivity.

The devices constructed for this study were

microbolometers that used an inherent electrostatic

actuator to bring its support beams into partial contact

with the substrate as shown in figure 1. This induced a large

change in the thermal conductance of the devices which

directly led to similar changes in responsivity and detectivity.

The thermal contact conductances of the material interfaces

used in micromachining are high enough [19] so that touching

a small portion is sufficient to cause dramatic changes in

1074


3/7

Design and characterization of adaptive microbolometers

(a)

(b)

Figure 1. Conceptual diagram of a tunable responsivitymicrobolometer. In (a), the device stands in a state of low thermalconductance and high responsivity. In (b), the device supports havebeen actuated into contact with the substrate, creating a high thermalconductivity and low responsivity state. In practice, only the edge

farthest from the ground support contacts the substrate to preventstiction.

detectivity. While total thermal contact could change the

potential dynamic range by four orders of magnitude [20],

most imaging requires only a fraction of this range. Partial

contact also helps to eliminate stiction during operation.

Relatively low actuation voltages can be used due to the low

rigidity of the supports.

3. Fabrication

3.1. Building the foundation layer and the first sacrificial

layer

As the first step, a 500 nm LPCVD silicon nitride layer is

deposited on the top of a highly doped silicon wafer for

electrical insulation as shown in figure 2(a). Then, PI2610

polyimide (HD Microelectronics) is spin coated as the first

sacrificial layer. Polyimide is used in lieu of a typical silicon

dioxide because HF used in etch release also etches PECVD

nitride, which is used as the structural layer later. To improve

the adhesion between the PI2610 polyimide and LPCVD

nitride, 0.1% VM-651 diluted in DI water is spin coated at

5000 rpm for 60 s, and baked at 125 C before applying

polyimide. For uniform spin coating of polyimide, polyimide

is spin coated at 500 rpm for 20 s, and then the spin rate

is ramped to 4000 rpm to get a thickness of approximately

1.8 m. To improve the etching efficiency of PI2610

polyimide in O2 plasma, polyimide is fully cured in a

convection oven. To prevent the polyimide film from

deformation due to a sudden change in curing temperature,

polyimide is baked on the hotplate at 90 C for 3 min and then

at 180 C for 90 s before curing in a convection oven. Finally

polyimide is loaded to a convection oven, where temperature is

initially set to 180 C to match the temperature before loading.

The temperature is increased at a rate of 4 C min1 up to

400 C. Note that the bulk of the solvent in the polyimide is

evaporated at the curing temperature of 200 C. However, in

our process, some still remains and polyimide is cured at a

higher temperature than 200 C, because of further process

(a)

(b)

(c)

(d)

(e)

Figure 2. Fabrication of an adaptive microbolometer. (a) Spin coatand pattern polyimide for anchors. (b) Deposition of PECVD nitrideand Ti/Au for beams. (c) Spin coat and pattern polyimide as secondsacrificial layer. (d) Deposition of PECVD nitride for top andbottom plates with TCR layer (Ti). (e) Dry etch releasing of pixeland beams.

issues such as outgassing during PECVD nitride deposition,

bubble forming during the annealing process and so on.

Furthermore, N2 gas is introduced in the convection oven to

prevent any surface reactions. During the curing process, the

thickness of the polyimide decreases to approximately 1.5m.

The polyimide is patterned by using a 300 nm thick aluminum

masking layer which is evaporated and patterned using a lift-

off process. The polyimide is etched with an O2 plasma to

form anchor points. During etching, the substrate is heated to

300 C to accelerate the etch rate as well as prevent the

formation of residue. Finally the aluminum layeris completely

etched in aluminum etchant.

1075


4/7


3.2. The support beams and their metallization

For the structural material, PECVD nitride is chosen for

two reasons: PECVD nitride has a relatively low thermal

conductivity compared to LPCVD nitride film. Also, it is

easy to control stress in PECVD nitride by changing process

parameters such as temperature and pressure [21, 22]. To form

support beams, a 500 nm thick PECVD nitride is deposited onthe top of patterned polyimide and etched using a CF4+O2plasma as shown in figure 2(b). We determined the shape that

yields the lowest actuation voltage by designing and testing

several shapes and lengths of support beams. To ensure a

low resistance for electrical read-out, the support beams are

coated with 50 nm of Ti as an adhesion layer and 300 nm of

Au as a conducting layer by e-beam evaporation. These layers

are patterned by lift-off. Due to the poor step coverage of

the sidewall between the anchor and the edge of the support

beams, a 50 nm of Ti and 1 m of Au layer is deposited on

the sidewall of anchor.

3.3. The second sacrificial layer and the vias

A second polyimide layer plays an important role in isolating

the pixel plate from the support beams. As shown in

figure 2(c), thesecondpolyimide layer is processed in thesame

manner as the first. After curing, it is etched to define vias

between theTi/Au and the TCR material (Ti) of the pixel plate.

The contact between Ti/Au and the TCR material through the

vias is very important, because it enables the reading of the

resistance change of the TCR material. Therefore, etching of

polyimide is done very carefully so that there is no polyimide

residue on the top of Ti/Au after etching. The polyimide is

etched in an O2 plasma with the substrate heated to 300C, as

was done for the first polyimide layer. The complete etchingof polyimide was verified by measuring the resistance of the

contact between Ti/Au and the TCR material. For example,

the contact resistance would be higher with the existence of

the polyimide residue.

3.4. Top and bottom layers of the pixel plate and the TCR

layer

The pixel plate consists of two PECVD nitride layers which

encapsulate the TCR layer as shown in figure 2(d). This

structure helps to protect the TCR layer from moisture and

other contamination. A 200 nm thick PECVD nitride is

deposited on the top of the second patterned polyimide layer

and etched in an STS etcher using a CF4+O2 plasma. This

defines the vias from the Ti/Au metal of the support beams to

the 100 nm Ti metal TCR layer, which is deposited using

e-beam evaporation. To encapsulate the Ti layer, another

200 nm of PECVD nitride is deposited. The two PECVD

nitride layers and the Ti layer are patterned and etched in

a CF4+O2 plasma using the STS etcher to form pixel plates

with a size of 100 m 100 m. In the final step, the

double polyimide layers are isotropically etched in an O2plasma with substrate heating at 300 C to release the pixel

plate and its support beams from the substrate. During this

etching process, we found that the polyimide layer around

the vias was not easy to remove due to the small gap.

This limited to achieving the thermal conductance value we

(a)

(b)

Figure 3. (a) SEM image of an individual microbolometer device.The dimensions of the microbolometer pixels are shown in table 1.The microbolometer pixels are designed to be compatible with highfill-factor arrays, and shown in part (b), an optical image of a portionof a 4 4 high fill factor array.

expected. This might be improved by adding a thick Ti

metal layer before TCR layer deposition to increase the gap

around the vias. This would enhance the detectivity to

its theoretical values by achieving a good thermal isolation

of the devices. A completed pixel and a 4 4 test

array with a fill factor of 91% are shown in figures 3(a)

and (b), respectively. Typical device characteristics are given

in table 1.

4. Measurement

The thermal conductances of the devices were measured

under vacuum (


5/7


Table 1. Measured and calculated parameters for the microbolometer.

Parameter Value

Size of pixel plate 100 m 100 m

Thickness of pixel plate Top plate (PECVD nitride): 0.2mTCR layer (titanium): 0.1 mBottom plate (PECVD nitride): 0.2 m

Length of support beam 277 mWidth of support beam 6 mThickness of support beam 0.5 mAir gap spacing Plate to support beam: 1.5 m

Support beam to substrate: 1.5 m

Thermal coefficient of resistance 0.2% K1 (Titanium)Thermal conductivity of PECVD nitride 0.0380.051 W cm1 K1

Density of PECVD nitride 2.22 g cm3

Specific heat of PECVD nitride 1.311.73 J g1 K1

Heat capacity of device 2.55 108 J K1

Thermal time constant 1.5 ms

Signal

Analyzer ConstantCurrentSource+ -

FunctionGenerator

DC PowerSupply

Laser Diode

with moduleLenses Beam Splitter

AdaptiveBolometer

Pinhole Aperture

VacuumChamber

Si Photodetector

Ref.

Preamplifier

Lock-inAmplifier

Figure 4. Experimental setup for the measurement of responsivityand detectivity.

the photodetector where a pinhole aperture of similar size to

the pixel under test was placed in front of it. The device was

biased with a current of 100A using a constant current source

to read an output voltage of the device. The microbolometer

output voltage was amplified using a PARC113 preamplifier

and then fed into the lock-in amplifier. For the calculation of

the detectivity, the voltage noise in the frequency of interest

was measured using an HP35660A dynamic signal analyzer.

5. Results and discussion

5.1. Tunable thermal conductance

As mentioned in the previous section, a thermal conductance

of a typical device was extracted from the plot of 1/R versus

I2 as shown in figure 5. In an unactuated state, a typical device

gives a thermal conductance, G = 1.68 105 W K1. This

is much lower value than the designed thermal conductance of

7 107 W K1 in its unactuated state. There are two possible

reasons for this. First the vacuum level in our apparatus was

only able to go to 10 mTorr, which places it in the transition

zone between air and solid-state limited conduction. It is

0.0 2.0x10-6 4.0x10-6 6.0x10-6 8.0x10-6

8.4x10-3

8.6x10-3

8.8x10-3

9.0x10-3

9.2x10-3

9.4x10-3

G = 2.99x10-4 W/K

G = 1.68x10-5 W/K

Vact = 0V

Vact = 17V

1/R

I2

Figure 5. A plot of the inverse resistance versus current squared fora microbolometer in its unactuated (Vact = 0 V) and soft actuation(Vact = 17 V) states.

further suspected that the top layer of polyimide between the

plate and the support beam was not fully removed during

etch release. With an actuation voltage of 17 V, the thermal

conductance of the device changes to G = 2.99 104 W

K1. This shows that the thermal conductance of a typical

device is changed by about 1.25 orders of magnitude. Full

device snap down of the support beams would have increased

the detectivity tuning, but in usestiction would have been seen.

The snap down and restoration of the support beam happens at

the moment of turn-on/off of power supply, but the numerical

value for the actuation time was not measured. Given the sizeof the device, the mechanical resonance is expected to be the

order of 10 kHz, making a transition time of the order of 1 ms

or less.

5.2. Tunable responsivity

The responsivity of the same device from figure 3(a) is shown

in figure 6. The responsivity varies by over an order of

magnitude from 1.5 V W1 to 0.2 V W1 (the absolute values

are low, but explained by the artificially low responsivity

described previously), when the support beam of the device

snaps down at a voltage of 17 V. In commercial devices, one

would likely use vanadium oxide instead of Ti because of its

higher TCR and place a special absorption layer rather than

1077


6/7


10 100 100010-2

10-1

100

Vact = 0V

Vact = 17V

Responsivity(V/W)

Frequency (Hz)

Figure 6. Responsivity versus frequency for an adaptiveresponsivity microbolometer. The response varies by over an orderof magnitude when the lowest device support is snapped downat 17 V.

counting on the intrinsic absorption of titanium and PECVD

nitride at = 658 nm. These factors represent limitations of

our available fabrication processes rather than the architectureitself. In figure 3(a), the reponsivities of the device taken at

frequencies below 10 Hz arenot plotted because of high scatter

and too high 1/fnoise.

5.3. Detectivity characterization

Once the responsivity had been measured, the noise spectrum

of the devices was obtained directly from the signal analyzer

and used to extract the detectivity as described in (3).

Figure 7 shows the detectivity versus frequency of the same

device as in figure 3(a). The detectivity varies more than an

order of magnitudebetween 1.8106 cmHz1/2 W1 and1.5105 cm Hz1/2 W1, at 0 V actuation and 17 V actuation,

respectively. As mentioned above, the detectivities below

10 Hz were not plotted because of difficulties in measuring

1/f noise at extremely low sampling rates, which made the

uncertainty in the detectivity values very large. This problem

might be solved by replacing the Ti TCR material with VOx,

because VOx gives a relatively low 1/fnoise characteristic at

low frequency range [6] as well as a high TCR of3.0% K1

at room temperature [2]. In this measurement, Johnson noise

was around 120 nV Hz1/2, which is somewhat higher than

in typical microbolometers. This noise will be reduced in the

future with a high TCR element and a low-noise resistor for

electrical read-out.

Lastly, the NEPs of adaptive microbolometers have been

measured versus frequency in high and low sensitivity states

with a bias current of 100 A as shown in figure 8. The lowest

NEP of 5.6 108 W Hz1/2 was obtained in an unactuated

state. This is higher than in commercial microbolometers,

but reasonable for the responsivity and TCR material of

our devices. At higher frequencies, NEPs for both states

converge.

The devices described are the first microbolometers

made with a fully adaptable detectivity, but they have a

relatively low performance relative to commercial devices.

In particular, the device was designed to have a much

lower thermal conductance of 7 107 W K1 than

measured in its unactuated state. This would directly

translate to another 1.2 orders of magnitude improvement

10 100 1000

104

105

106

Vact = 0V

Vact = 17V

Detectiv

ity(cmHz1/2/W)

Frequency (Hz)

Figure 7. Detectivity versus frequency for a typical device. Thedetectivity is in a high state at 0 V actuation and a low state at 17 Vactuation.

10 100 1000

10-7

10-6

Vact = 0 V

Vact = 17 V

NEP(W/Hz

1/2)

Frequency (Hz)

Figure 8. Noise equivalent power (NEP) versus frequency for atypical device. The lowest NEP is obtained in a high state at 0 Vactuation.

10 20 30 40 50 60 70 80200

205

210

215

220

225

230TCR of Ti = 0.21% K-1

Resistance(ohm)

Temperature (C)

Figure 9. Resistance versus temperature measured by point probefor titanium. Measured TCR of Ti is 0.21% K1. For TCRmeasurement, the temperature was scanned from 26 C to 70 C.

in responsivity and detectivity. In particular, it is suspected

that the top layer of polyimide between the plate and

the support beam was not fully removed during etch

release. This will be corrected in future processing. In

addition, the performance could be further improved by

using vanadium oxide instead of titanium as the TCR layer.

Commercial VOx has a TCR that is approximately 2%

K1 compared to a TCR for Ti of approximately 0.2% K1

as shown in figure 9. VOx also has a very low 1/f noise.

With these changes, the architecture presented here could

provide commercial thermal imaging performance with full

intensity adaptivity, extending the dynamic range of present-

day microbolometers by at least an order of magnitude.

1078


7/7


6. Conclusions

Adaptive microbolometers in which thermal conductance,

responsivity and detectivity are electronically controlled have

been made. These bolometers have been designed to have

extended dynamic range to handle both very hot and cool

areas within the same image. For characterization of the

devices, the thermal conductance, responsivity, detectivity and

NEP have been measured versus frequency in high and low

sensitivity states. The high sensitivity state corresponded to

normal microbolometer operation. In the low sensitivity state,

a portion of the support beam was actuated to be partially in

contact with the substrate and thereby increased the thermal

conductivity and reduced the detectivity. We reported that

the thermal conductance of a typical device could be changed

by 1.25 orders of magnitude, the responsivity by one order

of magnitude and the detectivity by 1.2 orders of magnitude

with the minimum NEP of 5.6 108 W Hz1/2 in the high

sensitivity state.

Acknowledgments

This work was supported by Defense Advanced Research

Project Agency under grant no. F33615-00-1-1625 and by the

Army Research Office under grant no. DAAD19-03-1-0343.

The authors would like to thank the staff at Nanofabrication

center at the University of Minnesota.

References

[1] Rogalski A 2003 Third-generation infrared photon detectors

Opt. Eng. 42 3498516[2] Wood R A, Han C J and Kruse P W 1992 Integrated uncooledinfrared detector imaging array Technical Digest of the1992 Solid-State Sensor and Actuator Workshop (Hilton

Head Island, SC, 26 June 1992) (Cleveland, OH:Transducer Research Foundation) pp 1325

[3] Almasri M, Butler D P and Celik-Butler Z 2001Self-supporting uncooled infrared microbolometers withlow-thermal mass J. Microelectromech. Syst. 10 46976

[4] Burgess D E, Manning P A and Watton R 1985 The theoreticaland experimental performance of a pyroelectric arrayimagerProc. SPIE 572 26

[5] Choi I H and Wise K D 1986 A silicon-thermopile-basedinfrared sensing array for use in automated manufacturing

IEEE Trans. Electron Devices 33 729[6] Wood R A 1997 Monolithic silicon microbolometer arrays

Uncooled Infrared Imaging Systems and Arrays vol 47ed P W Kruse and D D Skatrud (New York: Academic)pp 43121 chapter 3

[7] Shan P C, Celik-Butler Z, Butler D P and Jahanzeb A 1995Semiconducting YBaCuO thin films for uncooled infraredbolometers J. Appl. Phys. 78 73349

[8] Changhong C, Xinjian Y, Xingrong Z and Bifeng X 2000Preparation and properties of vanadium dioxide thin filmsfor uncooled microbolometerIEEE the Conf. Digest of the

25th Int. Conf. on Infrared and Millimeter Waves (1215September 2000) pp 1456

[9] Ichihara T, Watabe Y, Honda Y and Aizawa K 1997 A highperformance amorphous Si1xCx:H thermistor bolometerbased on micro-machined structure IEEE the Conf. Digestof Int. Conf. on Solid-state Sensors and Actuators (1619

June 1997) pp 12536[10] Lisauskas A, Khartsey S I and Grishin A 2000 Tailoring the

colossal magnetoresistivity: La0.7(Pb0.63Sr0.370.3MnO3thin-film uncooled bolometerAppl. Phys. Lett.77 7568

[11] Murphy D F et al 2001 High-sensitivity (25-m pitch)microbolometer FPAs Proc. SPIE 4454 14759

[12] Tezcan D S, Eminoglu S and Akin T 2003 A low-costuncooled infrared microbolometer detector in standardCMOS technology IEEE Trans. Electron Devices50 494502

[13] Lee H-K, Yoon J-B, Yoon E, Ju S-B, Yong Y-J, Lee W andKim S-G 1999 A high fill-factor infrared bolometer usingmicromachined multilevel electrothermal structures IEEETrans. Electron Devices 46 148991

[14] Hanson C M, Dudley D and Robinson J E 1996 Thermalimaging system with integrated thermal chopperUS patentno 5486698 (23 January 1996)

[15] Wood A 2000 private communication[16] Leonov V N and Butler D P 2001 Two-color thermal detector

with thermal chopping for infrared focal-plane arrays Appl.Opt. 40 260110

[17] Jerominek H, Picard F and Vincent D 1993 Vanadium oxidefilms for optical switching and detection Opt. Eng.32 20929

[18] Song W-B and Talghader J J 2002 Adjustable responsivity forthermal infrared detectors Appl. Phys. Lett. 81 5502

[19] Song W-B and Talghader J J 2002 Interface thermalconductance for microactuated devices Technical Digest ofthe 2002 Solid-State Sensor and Actuator Workshop (Hilton

Head Island, SC, 26 June 2002) (Cleveland OH:Transducer Research Foundation) pp 32932

[20] Song W-B and Talghader J J 2002 Thermal contactconductance of actuated interfaces Appl. Phys. Lett.81 12168

[21] Davies R R, McNie M E, Brunson K M and Combes D J 2001Engineering in- and out-of-plane stress in PECVD siliconnitride for CMOS-compatible surface micromachining

Proc. SPIE 4557 3208[22] Scheeper P R, Voorthuyzen J A and Bergveld P 1991 PECVD

silicon nitride diaphragms for condenser microphonesSensors Actuators B 4 7984

[23] Erickson P, Andersson J Y and Stemme G 1997 Thermal

characterization of surface-micromachined silicon nitridemembranes for thermal infrared detectors J.

Microelectromech. Syst. 6 5561

1079
http://dx.doi.org/10.1117/1.1625377http://dx.doi.org/10.1117/1.1625377http://dx.doi.org/10.1117/1.1625377http://dx.doi.org/10.1109/84.946808http://dx.doi.org/10.1109/84.946808http://dx.doi.org/10.1063/1.360382http://dx.doi.org/10.1063/1.360382http://dx.doi.org/10.1063/1.127109http://dx.doi.org/10.1063/1.127109http://dx.doi.org/10.1117/12.448169http://dx.doi.org/10.1117/12.448169http://dx.doi.org/10.1109/TED.2002.807453http://dx.doi.org/10.1109/TED.2002.807453http://dx.doi.org/10.1109/16.772496http://dx.doi.org/10.1109/16.772496http://dx.doi.org/10.1117/12.143951http://dx.doi.org/10.1117/12.143951http://dx.doi.org/10.1063/1.1493223http://dx.doi.org/10.1063/1.1493223http://dx.doi.org/10.1063/1.1499518http://dx.doi.org/10.1063/1.1499518http://dx.doi.org/10.1117/12.442962http://dx.doi.org/10.1117/12.442962http://dx.doi.org/10.1016/0925-4005(91)80180-Rhttp://dx.doi.org/10.1016/0925-4005(91)80180-Rhttp://dx.doi.org/10.1109/84.557531http://dx.doi.org/10.1109/84.557531http://dx.doi.org/10.1109/84.557531http://dx.doi.org/10.1016/0925-4005(91)80180-Rhttp://dx.doi.org/10.1117/12.442962http://dx.doi.org/10.1063/1.1499518http://dx.doi.org/10.1063/1.1493223http://dx.doi.org/10.1117/12.143951http://dx.doi.org/10.1109/16.772496http://dx.doi.org/10.1109/TED.2002.807453http://dx.doi.org/10.1117/12.448169http://dx.doi.org/10.1063/1.127109http://dx.doi.org/10.1063/1.360382http://dx.doi.org/10.1109/84.946808http://dx.doi.org/10.1117/1.1625377

Documents

Design and characterization of Adaptive microbolometers