rus
DOI: 10.22184/1993-7296.FRos.2021.15.5.396.409
The article presents the results of studying the possibilities of reducing the response time (time constant of thermal relaxation) to the thermal effect of a bolometric sensor. The analysis was carried out within the framework of the generally accepted theory of absorption of IR radiation and thermophysical processes occurring in membrane MEMS structures. The results of modeling the optimal design of high-performance microbolometric elements, taking into account the requirements of specific applications and its implementation using modern technological processes of formation, are presented. The discovered regularities will be applied in the production of microbolometers of the Astron family.
The article presents the results of studying the possibilities of reducing the response time (time constant of thermal relaxation) to the thermal effect of a bolometric sensor. The analysis was carried out within the framework of the generally accepted theory of absorption of IR radiation and thermophysical processes occurring in membrane MEMS structures. The results of modeling the optimal design of high-performance microbolometric elements, taking into account the requirements of specific applications and its implementation using modern technological processes of formation, are presented. The discovered regularities will be applied in the production of microbolometers of the Astron family.
Теги: direction finding of fast moving thermal objects microbolometers uncooled microbolometric photodetector vanadium oxide микроболометры неохлаждаемые микроболометрические фпу оксид ванадия пеленгация быстродвижущихся тепловых объектов
Fast Bolometric Focal Plane Arrays
R. Z. Khafizov, V. V. Startsev, V. Yu. Moskvichev
EDB “Astrohn” JSC, Lytkarino, Moscow region, Russia
The article presents the results of studying the possibilities of reducing the response time (time constant of thermal relaxation) to the thermal effect of a bolometric sensor. The analysis was carried out within the framework of the generally accepted theory of absorption of IR radiation and thermophysical processes occurring in membrane MEMS structures. The results of modeling the optimal design of high-performance microbolometric elements, taking into account the requirements of specific applications and its implementation using modern technological processes of formation, are presented. The discovered regularities will be applied in the production of microbolometers of the Astron family.
Keywords: microbolometers, uncooled microbolometric photodetector, direction finding of fast moving thermal objects, vanadium oxide
Received on: 21.06.2021
Accepted on: 05.08.2021
INTRODUCTION
The degree of integration and sensitivity of modern bolometric detectors have reached values typical of similar photonic devices. Compared to photonic detectors, bolometers provide energy absorption in a wider spectral range, can operate without cooling, and are very attractive in terms of cost and power consumption. At the same time, they are significantly inferior to photon detectors in terms of speed. Today, leading technology companies are working on the development of uncooled bolometric array detectors for applications that require the capture of rapidly changing events [1–3], for example, for use in optoelectronic systems (OES) for the direction finding of fast moving thermal objects.
This article is devoted to the determination of the main characteristics of sensitive microbolometric structures, the development of which is being carried out at EDB “Astron” JSC. The possibilities of optimizing the parameters and characteristics of uncooled microbolometric focal plane arrays are considered, constructive and technological solutions for their development are determined, aimed at meeting the requirements for the functioning of promising OES.
1. Electrophysical, thermal and design-technological parameters of a microbolometric sensor – problem statement
A microbolometer is a microelectromechanical sensor (MEMS) that is sensitive to thermal radiation and is manufactured using silicon micromachining techniques. A major advantage of MEMS fabrication processes is that they fit well with standard CMOS silicon technology. The main element of the sensor is a thin dielectric membrane on which an active thermoresistive layer is deposited. As a result of etching the sacrificial layer located under the membrane, it is hung over the surface of the substrate. It is connected to the substrate by narrow micro support consoles that detect and control the thermal conductivity of the sensor. Having a low thermal mass, the membrane is able to quickly heat up under the influence of infrared radiation. In order to increase the sensitivity of the structure, a coating is applied to the membrane that efficiently absorbs IR radiation. The use of sectoral (localized in certain sectors of the array) absorbing coatings suggests the possibility of manufacturing multispectral microbolometric sensors.
A number of factors determine the parameters of a microbolometric sensor as part of an IR image receiver, namely:
temperature resistance coefficient (TRC) α of the active layer;
absorption coefficient η of infrared radiation;
sensor area S, including the area of the sensitive membrane A and the area Sc occupied by the consoles, S = A + Sc;
heat capacity C of the membrane (including the absorbing coating), equal to C = cA, where c is the heat capacity of the membrane per unit area;
thermal conductivity of the consoles G, determined primarily by the thermal conductivity of the consoles;
electrical resistance of the sensor R;
time constant of thermal relaxation of the sensor τr;
frame time τf of the array detector, determined through the time constant of thermal relaxation of the sensor τr as τf = πτr.
The listed parameters are closely related to each other, influencing the sensitivity of the sensor.
When designing the microbolometric focal plane arrays receiver, the following characteristics are set as the main ones: the number N and the pitch d of the array elements, which determine the required spatial resolution, the frame time τf, which gives the time resolution, and the sensitivity in terms of the noise equivalent temperature difference (NETD – Noise Equivalent Temperature Difference) device. The number of array elements at a certain size of its sensitive array limits the maximum cell area and, accordingly, the area A of its active part. The detector response required for specific applications imposes an upper limit on the thermal relaxation time of the radiation detector τf.
The amount by which the active part of the film structure changes its resistance when heated depends on the temperature coefficient of the material resistivity (TRC). Providing a high TRC value is necessary to achieve high sensor sensitivity. The active layer is also characterized by a specific electrical resistance ρ, which it is desirable to optimize to achieve, on the one hand, the maximum TRC, and on the other, not to greatly increase the noise minimization.
Vanadium oxide VOx is one of the most widely used materials for the formation of active thermoresistive films in modern microbolometers. It has a fairly low electrical resistivity with a sufficiently high and stable TRC (films are not subject to hysteresis phenomena with temperature changes). At the same time, intensive research is currently underway to find new materials, since the implementation of thermoresistive films based on VOx requires rather complex technological processes, the development of which causes significant difficulties.
2. Thermophysical analysis
The conversion of infrared radiation in thermal sensors, in contrast to quantum analogs, occurs in two stages. First, the membrane is heated, and then the resulting temperature difference ΔT between the substrate and the membrane is converted into a signal carrying information about the radiation intensity. The temperature difference ΔT resulting from the absorption of infrared radiation by the membrane is determined by the thermal balance between the absorbed thermal energy and the heat loss associated with the thermal conductivity of the consoles.
The heat balance equation is as follows:
, (1)
where C is the membrane heat capacity [J / K], ΔP(t) is the change in the thermal radiation power [W] in the detector plane, η is the IR absorption coefficient, and ∆T is the temperature change between the detector and the heat sink (substrate), G is the thermal conductivity of the consoles [W / K]. The reaction of the membrane temperature to a sharply increasing change in the radiation power can be determined by integrating equation (1) taking into account the boundary condition t = 0, ΔT = 0. The result of this integration is the expression:
,
where is a time constant of thermal relaxation characterizing the reaction time of the membrane to a change in the power of thermal radiation.
In a steady state, when
.
The heat capacity of a membrane is determined through the specific volumetric heat capacity cV, which depends on the thermal properties of the materials from which it is made:
C = cV = cVAth = cAA,
where A is its area, ht is thickness, and cA is specific heat per unit area.
When registering an alternating signal at a frequency f, the expression for ΔT can be written in the form:
,
where ΔPf is the corresponding frequency component of the thermal radiation power absorbed by the sensor membrane, w = 2πf.
This ratio is generally applicable for all thermal sensors, and thus the first stage of conversion of thermal radiation, which consists in heating the membrane, is identical for all types of thermal sensors. The differences between them are determined by the physical mechanisms of converting the temperature difference ΔT between the substrate and the membrane into an electrical signal, which is the essence of the second stage of the conversion of infrared radiation. It is these mechanisms that influence the possibility of achieving the limiting characteristics of thermal sensors in terms of their sensitivity and speed.
In bolometric sensors using the thermoresistive effect, the change in the electrical resistance of the thermosensitive layer under the influence of thermal radiation is determined. The relationship between the change in temperature ΔT and the change in the resistance ΔRB of the bolometer is determined by the ratio:
ΔRB = αΔRBΔT,
where α is the temperature coefficient of electrical resistance (TRC). α is defined as:
.
The TRC value of vanadium oxide films is usually in the range from –0.02 K‑1 to –0.03 K‑1.
For the formation of the electrical signal ΔU of the bolometer, a voltage is required that displaces it by the value UB. With this in mind, the relationship between ΔU and ΔT is determined by the expression:
ΔU = αUBΔT.
It is important to remember an essential circumstance associated with the necessity of electrical displacement of the bolometric sensor. It consists in the fact that the flow of current through the sensor causes its additional heating. The Joule heat released during heating is determined by the expression P = U2B / RB and, accordingly, the additional temperature heating induced by the displacement is equal to:
.
The element with area S of the bolometric array is schematically shown in Fig. 1. A membrane with an IR-absorbing layer of area A is suspended on micro consoles of length L and width wt made of heat-insulating material. A thermistor is formed on the surface of the membrane, the conductive buses of which are laid along the consoles. The readout of changes in the thermistor current when the membrane is heated under the influence of infrared radiation is carried out by the CMOS IS. The IS is located directly on the chip and is connected to the membrane using indium columnar contacts. The supporting dielectric layer of the membrane and the consoles are made, for example, of silicon nitride by etching a sacrificial layer of polyimide or of thermal silicon dioxide by local etching of a silicon substrate under the membrane.
Here and in what follows, we will consider an element with two consoles of width wt, the length of each of which is L. In this case, we will assume that the electrical resistance of the element is determined mainly by the resistance of the thermoresistive layer, and the resistance of the busbars can be neglected.
Taking into account the ratio , to increase the response speed, it is necessary to decrease the heat capacity and increase the thermal conductivity of the structure. The heat capacity of the membrane is determined by its volume, and the thermal conductivity depends on the geometry and material of the consoles. Reducing the heat capacity for a given element area is possible only through the use of thinner layers included in the membrane structure. This is limited by the capabilities of the technology and limitations imposed by the need to provide sufficient structural strength. At the same time, an increase in thermal conductivity can be realized by fairly simple design solutions (see Fig. 2). On this path, however, one must proceed from the fact that the decrease in sensitivity, which is inevitably associated with an increase in thermal conductivity, should not go beyond the limits determined by the requirements of the problem solved with the help of the photodetector. It is also important to emphasize that the use of design possibilities to increase the speed of response makes sense, along with the search for technologies and materials that provide the maximum values of the TRC, the absorption coefficient of infrared radiation and the spectral sensitivity of the element that is optimal for solving specific problems.
Fig. 3 shows the plots of relaxation of a thermal sensor with an area of 50 × 50 μm2 when exposed to thermal power ∆P = 2.5 ∙ 10–9 W for different values of the thermal conductivity of the consoles.
The plot shows that there is a one-to-one correspondence between the achievable sensor sensitivity and its speed. Therefore, for each specific application, when designing a sensor, the optimal ratio between its sensitivity and speed is chosen. In [4, 5], for a given array pixel size and frame time, the optimal ratio between the membrane area and the area occupied by the consoles was revealed, which provides the highest possible sensitivity.
Focusing on modern technological capabilities for conducting assessments, the following technological parameters of the membrane were adopted:
bearing layer Si3N4 150 nm thick,
thermoresistive layer VOx 50 nm thick,
protective layer SiO2 100 nm thick,
a layer of absorbing material with a thickness of 10 nm.
Taking into account the thermophysical parameters of the materials of the structure for the heat capacity of the membrane, we obtain the value: C = 6 ∙ 10–5 ∙ A [J / K], where A is expressed in cm2 (9).
To reduce the thermal resistance of the consoles, the conductive sections of the consoles are considered to be made of VOx with a specific thermal conductivity gt = 0,05 W / cm · K, then for the thermal conductivity we get:
[W / К],
and, accordingly, for τr we get:
, (2)
where d is expressed in cm (11)
Fig. 4 shows the dependence of the relaxation time on the pitch of the elements, determined by formula (2).
3. TRC and electrical resistance
of thermoresistive material
The electrical resistance of the microbolometric element is selected based on the possibility of achieving high TRC values in accordance with the dependence of TRC on the resistivity of the thermoresistive material, and minimizing the bias power and noise to achieve low NETD values. The relationship between TCS and resistivity for VOx is illustrated in Fig. 5.
The resistance of a microbolometer with a near-square topology of a thermosensitive element can be represented as:
,
where ht is the thickness of the thermistor material, ρt is its electrical resistivity. Typical values for VOx films used in modern bolometers are ρt ≈ 1 Ω · cm,ht ≈ 50 nm, which gives for a Rt value of ~200 kΩ.
4. Sensitivity
The sensitivity of microbolometers in an OES is characterized by a noise equivalent temperature difference (NETD – Noise Equivalent Temperature Difference), which is determined based on the signal-to-noise ratio (SNR) for a given incident power. A higher SNR for a particular instrument indicates better sensitivity. Therefore, to increase the sensitivity of the instrument, high response (output) values are needed along with a reduction in noise.
The output signal of the bolometric sensor is:
.
Then for NETD it is true:
,
where ∆p / ∆T is the change in irradiance in focal plane of the array when the scene temperature changes, which depends on the parameters of the lens and the IR spectrum, Unoise is the noise voltage.
Fig. 6 shows the calculated voltage dependences of the main types of noise in a single frequency band [7] for a bolometric sensor with typical parameter values: T = 300 K, Δf = 30 30 Hz, TRC α = –2.0%, thermal conductivity G = 5.0 · 10–7 W / K, thermal relaxation time τr = 5 ms, absorption coefficient η = 0.9, active area of the sensor Α = 35 × 35 μm2 with thickness ht = 100 nm. The applied bias voltage is VB = 1 В.
It can be seen that the noise characteristics of the bolometer are mainly determined by the Johnson noise. The well-known analytical expression for this type of noise allows one to estimate NETD, considering its contribution to be decisive.
In this approximation for NETD we get:
. (3)
Here k is the Boltzmann constant, ∆f is the frequency band depending on the frame time and the number of array elements. It is also important to note that the NETD value is determined by the parameters of the materials included in the membrane structure, namely, α, gt and ρB, the topological dimensions of the sensor (A, L, wt) and technological constraints that specify the value of the membrane structure thickness ht. As part of the equipment, the NETD value is significantly influenced by the characteristics of the optical system and the conditions of its operation (background environment, thermal characteristics of objects, etc.)
Fig. 7 shows the calculated dependence of NETD on the pitch of the bolometric array elements, obtained taking into account the dependence of the frequency band on the integration time of the array signals . Plots of NETD versus element spacing are shown together with similar dependencies for relaxation times shown in Fig. 4.
It can be seen from the figure that acceptable NETD values (less than 100 mK) at sufficiently low values of thermal relaxation time (less than 2 ms) can be achieved for matrices with an element pitch of 35–40 μm and, thus, the frame rate can be increased to at least ~ 200 Hz. In this case, you should not increase the number of array elements, limiting yourself to an array with the number of pixels no more than 64 × 64, or, if it is critically important to increase the array format, then you can implement, for example, a 160 × 120 format with division into four sectors with individual outputs, without increasing the bandwidth frequencies when reading signals. Such small array formats, however, are sufficient for a number of special applications, in particular, when they are used in thermal direction-finding OES operating with small fields of view, where the need to fix objects moving in space at high speeds comes to the fore. It should also be noted that the NETD dependencies shown in Fig. 5, were obtained under the assumption that the change in the surface radiation flux density at the receiver with a change in the object temperature, determined in expression (3) by the ratio Δp/Δt is equal to:
,
where η is the fraction of incident radiation absorbed by the membrane, q is the optical factor q = Hto / 4, H is the aperture ratio, and to is the transmission of the objective. It was assumed that Н = 1 and to = 1, and for the value of ≈ 2.62 · 10–4 W / cm2 ∙ K [8] was taken for the range 8–14 μm. When working in this range with objects heated to 600 K, the estimate gives the value ≈ 1.7 · 10–3 W / cm2 ∙ K, which significantly (up to 6.5 times) improves the capabilities thermal direction finding of such objects at high frame rates (see Fig. 8). And when recording infrared radiation in a wider spectral range, an improvement of more than an order of magnitude can be achieved. In this case, however, it is necessary to optimize the spectral sensitivity of the bolometric sensor by using a broadband absorbing coating, for example, gold black [9]. Gold black has high absorption in a wide spectral range (see Fig. 9) and is characterized by low heat capacity, without affecting the noise level and thermal relaxation time. As the results of a number of studies show, films of nanomaterials are also promising and, in particular, films of carbon nanotubes [10]. They are characterized by a large integral surface area and high porosity for extremely low heat capacity. It is also necessary to optimize the size of the interference vacuum gap between the membrane and the IR-reflecting layer on the substrate.
5. Possibilities of using signal processing
Additional improvements in the tactical and technical characteristics of OES with bolometric photodetectors used in heat direction finding equipment can be implemented through adaptive signal processing, taking into account the current change in the conditions for registering thermal objects. So at significant distances from the target, its thermal point image in the focal plane of the optical system shifts at low speeds, which makes it possible to work with low frame rates. A bolometric array optimized for high frame rates can be converted to low-frame mode by summing signals in an external microprocessor, thereby raising its sensitivity. This is important, since at large distances from the target, the thermal signals coming from it to the input of the OES are significantly weakened. When the OES approaches the target, its angular velocity increases, and the speed of the array comes to the fore, the growth of which, as can be seen from the previous consideration, is achieved due to a certain decrease in sensitivity. The reduced sensitivity value is not so critical in this case, since the signal from the target becomes significantly higher.
6. Conclusion
The paper considers the theoretical relations that optimize the design of the microbolometric element of the array detector, providing an improvement in its performance characteristics, taking into account specific applications. Requirements for the parameters of elements and materials used for its manufacture are formulated, which provide a fundamental possibility of minimizing the thermal relaxation time of the bolometric sensor in order to increase the frame rate of the matrices. The found patterns will be applied in the production of microbolometers of the Astron family.
AUTHORS
Renat Zakirovich KHAFIZOV, Candidate of Science (Phys.&Math), Head of the Research Center «Bolometric detectors», EDB “Astron” JSC, www.astrohn.com, Lytkarino, Moscow region, Russia.
ORCID: 0000-0002-8319-5901
Vadim V. STARTSEV, Candidate of Technical Science, Chief Designer of
EDB “Astron” JSC, www.astrohn.com, Lytkarino, Moscow region, Russia.
ORCID: 0000-0002-2800-544X
Vadim Yurievich MOSKVICHEV, Deputy Chief Designer of EDB “Astron” JSC,
www.astrohn.com, Lytkarino, Moscow region, Russia.
R. Z. Khafizov, V. V. Startsev, V. Yu. Moskvichev
EDB “Astrohn” JSC, Lytkarino, Moscow region, Russia
The article presents the results of studying the possibilities of reducing the response time (time constant of thermal relaxation) to the thermal effect of a bolometric sensor. The analysis was carried out within the framework of the generally accepted theory of absorption of IR radiation and thermophysical processes occurring in membrane MEMS structures. The results of modeling the optimal design of high-performance microbolometric elements, taking into account the requirements of specific applications and its implementation using modern technological processes of formation, are presented. The discovered regularities will be applied in the production of microbolometers of the Astron family.
Keywords: microbolometers, uncooled microbolometric photodetector, direction finding of fast moving thermal objects, vanadium oxide
Received on: 21.06.2021
Accepted on: 05.08.2021
INTRODUCTION
The degree of integration and sensitivity of modern bolometric detectors have reached values typical of similar photonic devices. Compared to photonic detectors, bolometers provide energy absorption in a wider spectral range, can operate without cooling, and are very attractive in terms of cost and power consumption. At the same time, they are significantly inferior to photon detectors in terms of speed. Today, leading technology companies are working on the development of uncooled bolometric array detectors for applications that require the capture of rapidly changing events [1–3], for example, for use in optoelectronic systems (OES) for the direction finding of fast moving thermal objects.
This article is devoted to the determination of the main characteristics of sensitive microbolometric structures, the development of which is being carried out at EDB “Astron” JSC. The possibilities of optimizing the parameters and characteristics of uncooled microbolometric focal plane arrays are considered, constructive and technological solutions for their development are determined, aimed at meeting the requirements for the functioning of promising OES.
1. Electrophysical, thermal and design-technological parameters of a microbolometric sensor – problem statement
A microbolometer is a microelectromechanical sensor (MEMS) that is sensitive to thermal radiation and is manufactured using silicon micromachining techniques. A major advantage of MEMS fabrication processes is that they fit well with standard CMOS silicon technology. The main element of the sensor is a thin dielectric membrane on which an active thermoresistive layer is deposited. As a result of etching the sacrificial layer located under the membrane, it is hung over the surface of the substrate. It is connected to the substrate by narrow micro support consoles that detect and control the thermal conductivity of the sensor. Having a low thermal mass, the membrane is able to quickly heat up under the influence of infrared radiation. In order to increase the sensitivity of the structure, a coating is applied to the membrane that efficiently absorbs IR radiation. The use of sectoral (localized in certain sectors of the array) absorbing coatings suggests the possibility of manufacturing multispectral microbolometric sensors.
A number of factors determine the parameters of a microbolometric sensor as part of an IR image receiver, namely:
temperature resistance coefficient (TRC) α of the active layer;
absorption coefficient η of infrared radiation;
sensor area S, including the area of the sensitive membrane A and the area Sc occupied by the consoles, S = A + Sc;
heat capacity C of the membrane (including the absorbing coating), equal to C = cA, where c is the heat capacity of the membrane per unit area;
thermal conductivity of the consoles G, determined primarily by the thermal conductivity of the consoles;
electrical resistance of the sensor R;
time constant of thermal relaxation of the sensor τr;
frame time τf of the array detector, determined through the time constant of thermal relaxation of the sensor τr as τf = πτr.
The listed parameters are closely related to each other, influencing the sensitivity of the sensor.
When designing the microbolometric focal plane arrays receiver, the following characteristics are set as the main ones: the number N and the pitch d of the array elements, which determine the required spatial resolution, the frame time τf, which gives the time resolution, and the sensitivity in terms of the noise equivalent temperature difference (NETD – Noise Equivalent Temperature Difference) device. The number of array elements at a certain size of its sensitive array limits the maximum cell area and, accordingly, the area A of its active part. The detector response required for specific applications imposes an upper limit on the thermal relaxation time of the radiation detector τf.
The amount by which the active part of the film structure changes its resistance when heated depends on the temperature coefficient of the material resistivity (TRC). Providing a high TRC value is necessary to achieve high sensor sensitivity. The active layer is also characterized by a specific electrical resistance ρ, which it is desirable to optimize to achieve, on the one hand, the maximum TRC, and on the other, not to greatly increase the noise minimization.
Vanadium oxide VOx is one of the most widely used materials for the formation of active thermoresistive films in modern microbolometers. It has a fairly low electrical resistivity with a sufficiently high and stable TRC (films are not subject to hysteresis phenomena with temperature changes). At the same time, intensive research is currently underway to find new materials, since the implementation of thermoresistive films based on VOx requires rather complex technological processes, the development of which causes significant difficulties.
2. Thermophysical analysis
The conversion of infrared radiation in thermal sensors, in contrast to quantum analogs, occurs in two stages. First, the membrane is heated, and then the resulting temperature difference ΔT between the substrate and the membrane is converted into a signal carrying information about the radiation intensity. The temperature difference ΔT resulting from the absorption of infrared radiation by the membrane is determined by the thermal balance between the absorbed thermal energy and the heat loss associated with the thermal conductivity of the consoles.
The heat balance equation is as follows:
, (1)
where C is the membrane heat capacity [J / K], ΔP(t) is the change in the thermal radiation power [W] in the detector plane, η is the IR absorption coefficient, and ∆T is the temperature change between the detector and the heat sink (substrate), G is the thermal conductivity of the consoles [W / K]. The reaction of the membrane temperature to a sharply increasing change in the radiation power can be determined by integrating equation (1) taking into account the boundary condition t = 0, ΔT = 0. The result of this integration is the expression:
,
where is a time constant of thermal relaxation characterizing the reaction time of the membrane to a change in the power of thermal radiation.
In a steady state, when
.
The heat capacity of a membrane is determined through the specific volumetric heat capacity cV, which depends on the thermal properties of the materials from which it is made:
C = cV = cVAth = cAA,
where A is its area, ht is thickness, and cA is specific heat per unit area.
When registering an alternating signal at a frequency f, the expression for ΔT can be written in the form:
,
where ΔPf is the corresponding frequency component of the thermal radiation power absorbed by the sensor membrane, w = 2πf.
This ratio is generally applicable for all thermal sensors, and thus the first stage of conversion of thermal radiation, which consists in heating the membrane, is identical for all types of thermal sensors. The differences between them are determined by the physical mechanisms of converting the temperature difference ΔT between the substrate and the membrane into an electrical signal, which is the essence of the second stage of the conversion of infrared radiation. It is these mechanisms that influence the possibility of achieving the limiting characteristics of thermal sensors in terms of their sensitivity and speed.
In bolometric sensors using the thermoresistive effect, the change in the electrical resistance of the thermosensitive layer under the influence of thermal radiation is determined. The relationship between the change in temperature ΔT and the change in the resistance ΔRB of the bolometer is determined by the ratio:
ΔRB = αΔRBΔT,
where α is the temperature coefficient of electrical resistance (TRC). α is defined as:
.
The TRC value of vanadium oxide films is usually in the range from –0.02 K‑1 to –0.03 K‑1.
For the formation of the electrical signal ΔU of the bolometer, a voltage is required that displaces it by the value UB. With this in mind, the relationship between ΔU and ΔT is determined by the expression:
ΔU = αUBΔT.
It is important to remember an essential circumstance associated with the necessity of electrical displacement of the bolometric sensor. It consists in the fact that the flow of current through the sensor causes its additional heating. The Joule heat released during heating is determined by the expression P = U2B / RB and, accordingly, the additional temperature heating induced by the displacement is equal to:
.
The element with area S of the bolometric array is schematically shown in Fig. 1. A membrane with an IR-absorbing layer of area A is suspended on micro consoles of length L and width wt made of heat-insulating material. A thermistor is formed on the surface of the membrane, the conductive buses of which are laid along the consoles. The readout of changes in the thermistor current when the membrane is heated under the influence of infrared radiation is carried out by the CMOS IS. The IS is located directly on the chip and is connected to the membrane using indium columnar contacts. The supporting dielectric layer of the membrane and the consoles are made, for example, of silicon nitride by etching a sacrificial layer of polyimide or of thermal silicon dioxide by local etching of a silicon substrate under the membrane.
Here and in what follows, we will consider an element with two consoles of width wt, the length of each of which is L. In this case, we will assume that the electrical resistance of the element is determined mainly by the resistance of the thermoresistive layer, and the resistance of the busbars can be neglected.
Taking into account the ratio , to increase the response speed, it is necessary to decrease the heat capacity and increase the thermal conductivity of the structure. The heat capacity of the membrane is determined by its volume, and the thermal conductivity depends on the geometry and material of the consoles. Reducing the heat capacity for a given element area is possible only through the use of thinner layers included in the membrane structure. This is limited by the capabilities of the technology and limitations imposed by the need to provide sufficient structural strength. At the same time, an increase in thermal conductivity can be realized by fairly simple design solutions (see Fig. 2). On this path, however, one must proceed from the fact that the decrease in sensitivity, which is inevitably associated with an increase in thermal conductivity, should not go beyond the limits determined by the requirements of the problem solved with the help of the photodetector. It is also important to emphasize that the use of design possibilities to increase the speed of response makes sense, along with the search for technologies and materials that provide the maximum values of the TRC, the absorption coefficient of infrared radiation and the spectral sensitivity of the element that is optimal for solving specific problems.
Fig. 3 shows the plots of relaxation of a thermal sensor with an area of 50 × 50 μm2 when exposed to thermal power ∆P = 2.5 ∙ 10–9 W for different values of the thermal conductivity of the consoles.
The plot shows that there is a one-to-one correspondence between the achievable sensor sensitivity and its speed. Therefore, for each specific application, when designing a sensor, the optimal ratio between its sensitivity and speed is chosen. In [4, 5], for a given array pixel size and frame time, the optimal ratio between the membrane area and the area occupied by the consoles was revealed, which provides the highest possible sensitivity.
Focusing on modern technological capabilities for conducting assessments, the following technological parameters of the membrane were adopted:
bearing layer Si3N4 150 nm thick,
thermoresistive layer VOx 50 nm thick,
protective layer SiO2 100 nm thick,
a layer of absorbing material with a thickness of 10 nm.
Taking into account the thermophysical parameters of the materials of the structure for the heat capacity of the membrane, we obtain the value: C = 6 ∙ 10–5 ∙ A [J / K], where A is expressed in cm2 (9).
To reduce the thermal resistance of the consoles, the conductive sections of the consoles are considered to be made of VOx with a specific thermal conductivity gt = 0,05 W / cm · K, then for the thermal conductivity we get:
[W / К],
and, accordingly, for τr we get:
, (2)
where d is expressed in cm (11)
Fig. 4 shows the dependence of the relaxation time on the pitch of the elements, determined by formula (2).
3. TRC and electrical resistance
of thermoresistive material
The electrical resistance of the microbolometric element is selected based on the possibility of achieving high TRC values in accordance with the dependence of TRC on the resistivity of the thermoresistive material, and minimizing the bias power and noise to achieve low NETD values. The relationship between TCS and resistivity for VOx is illustrated in Fig. 5.
The resistance of a microbolometer with a near-square topology of a thermosensitive element can be represented as:
,
where ht is the thickness of the thermistor material, ρt is its electrical resistivity. Typical values for VOx films used in modern bolometers are ρt ≈ 1 Ω · cm,ht ≈ 50 nm, which gives for a Rt value of ~200 kΩ.
4. Sensitivity
The sensitivity of microbolometers in an OES is characterized by a noise equivalent temperature difference (NETD – Noise Equivalent Temperature Difference), which is determined based on the signal-to-noise ratio (SNR) for a given incident power. A higher SNR for a particular instrument indicates better sensitivity. Therefore, to increase the sensitivity of the instrument, high response (output) values are needed along with a reduction in noise.
The output signal of the bolometric sensor is:
.
Then for NETD it is true:
,
where ∆p / ∆T is the change in irradiance in focal plane of the array when the scene temperature changes, which depends on the parameters of the lens and the IR spectrum, Unoise is the noise voltage.
Fig. 6 shows the calculated voltage dependences of the main types of noise in a single frequency band [7] for a bolometric sensor with typical parameter values: T = 300 K, Δf = 30 30 Hz, TRC α = –2.0%, thermal conductivity G = 5.0 · 10–7 W / K, thermal relaxation time τr = 5 ms, absorption coefficient η = 0.9, active area of the sensor Α = 35 × 35 μm2 with thickness ht = 100 nm. The applied bias voltage is VB = 1 В.
It can be seen that the noise characteristics of the bolometer are mainly determined by the Johnson noise. The well-known analytical expression for this type of noise allows one to estimate NETD, considering its contribution to be decisive.
In this approximation for NETD we get:
. (3)
Here k is the Boltzmann constant, ∆f is the frequency band depending on the frame time and the number of array elements. It is also important to note that the NETD value is determined by the parameters of the materials included in the membrane structure, namely, α, gt and ρB, the topological dimensions of the sensor (A, L, wt) and technological constraints that specify the value of the membrane structure thickness ht. As part of the equipment, the NETD value is significantly influenced by the characteristics of the optical system and the conditions of its operation (background environment, thermal characteristics of objects, etc.)
Fig. 7 shows the calculated dependence of NETD on the pitch of the bolometric array elements, obtained taking into account the dependence of the frequency band on the integration time of the array signals . Plots of NETD versus element spacing are shown together with similar dependencies for relaxation times shown in Fig. 4.
It can be seen from the figure that acceptable NETD values (less than 100 mK) at sufficiently low values of thermal relaxation time (less than 2 ms) can be achieved for matrices with an element pitch of 35–40 μm and, thus, the frame rate can be increased to at least ~ 200 Hz. In this case, you should not increase the number of array elements, limiting yourself to an array with the number of pixels no more than 64 × 64, or, if it is critically important to increase the array format, then you can implement, for example, a 160 × 120 format with division into four sectors with individual outputs, without increasing the bandwidth frequencies when reading signals. Such small array formats, however, are sufficient for a number of special applications, in particular, when they are used in thermal direction-finding OES operating with small fields of view, where the need to fix objects moving in space at high speeds comes to the fore. It should also be noted that the NETD dependencies shown in Fig. 5, were obtained under the assumption that the change in the surface radiation flux density at the receiver with a change in the object temperature, determined in expression (3) by the ratio Δp/Δt is equal to:
,
where η is the fraction of incident radiation absorbed by the membrane, q is the optical factor q = Hto / 4, H is the aperture ratio, and to is the transmission of the objective. It was assumed that Н = 1 and to = 1, and for the value of ≈ 2.62 · 10–4 W / cm2 ∙ K [8] was taken for the range 8–14 μm. When working in this range with objects heated to 600 K, the estimate gives the value ≈ 1.7 · 10–3 W / cm2 ∙ K, which significantly (up to 6.5 times) improves the capabilities thermal direction finding of such objects at high frame rates (see Fig. 8). And when recording infrared radiation in a wider spectral range, an improvement of more than an order of magnitude can be achieved. In this case, however, it is necessary to optimize the spectral sensitivity of the bolometric sensor by using a broadband absorbing coating, for example, gold black [9]. Gold black has high absorption in a wide spectral range (see Fig. 9) and is characterized by low heat capacity, without affecting the noise level and thermal relaxation time. As the results of a number of studies show, films of nanomaterials are also promising and, in particular, films of carbon nanotubes [10]. They are characterized by a large integral surface area and high porosity for extremely low heat capacity. It is also necessary to optimize the size of the interference vacuum gap between the membrane and the IR-reflecting layer on the substrate.
5. Possibilities of using signal processing
Additional improvements in the tactical and technical characteristics of OES with bolometric photodetectors used in heat direction finding equipment can be implemented through adaptive signal processing, taking into account the current change in the conditions for registering thermal objects. So at significant distances from the target, its thermal point image in the focal plane of the optical system shifts at low speeds, which makes it possible to work with low frame rates. A bolometric array optimized for high frame rates can be converted to low-frame mode by summing signals in an external microprocessor, thereby raising its sensitivity. This is important, since at large distances from the target, the thermal signals coming from it to the input of the OES are significantly weakened. When the OES approaches the target, its angular velocity increases, and the speed of the array comes to the fore, the growth of which, as can be seen from the previous consideration, is achieved due to a certain decrease in sensitivity. The reduced sensitivity value is not so critical in this case, since the signal from the target becomes significantly higher.
6. Conclusion
The paper considers the theoretical relations that optimize the design of the microbolometric element of the array detector, providing an improvement in its performance characteristics, taking into account specific applications. Requirements for the parameters of elements and materials used for its manufacture are formulated, which provide a fundamental possibility of minimizing the thermal relaxation time of the bolometric sensor in order to increase the frame rate of the matrices. The found patterns will be applied in the production of microbolometers of the Astron family.
AUTHORS
Renat Zakirovich KHAFIZOV, Candidate of Science (Phys.&Math), Head of the Research Center «Bolometric detectors», EDB “Astron” JSC, www.astrohn.com, Lytkarino, Moscow region, Russia.
ORCID: 0000-0002-8319-5901
Vadim V. STARTSEV, Candidate of Technical Science, Chief Designer of
EDB “Astron” JSC, www.astrohn.com, Lytkarino, Moscow region, Russia.
ORCID: 0000-0002-2800-544X
Vadim Yurievich MOSKVICHEV, Deputy Chief Designer of EDB “Astron” JSC,
www.astrohn.com, Lytkarino, Moscow region, Russia.
Readers feedback
