Thermal detectors CHAPTER 07 1 Thermopile detectors 1-1 1-2 1-3 1-4 1-5 1-6 Features Structure Characteristics How to use New approaches Applications 2 Bolometers 2-1 Operating principle and structure 2-2 Characteristics 1 Thermal detectors Thermal detectors have an absorption layer that absorbs and converts light into heat, and provide an electric signal output that represents the change in absorption layer temperature. Because thermal detectors have no wavelength dependence, they can serve as infrared detectors when used with a window material such as Si that transmits infrared light. Thermal detectors are mainly classified into: (1) thermopile detectors that change in electromotive force, (2) bolometers that change in resistance, (3) pyroelectric detectors that change in dielectric surface charge, and (4) diodes that change in voltagecurrent characteristics. Hamamatsu manufactures two types of thermal detectors: thermopile detectors and bolometers. These two types of thermal detectors are different in terms of operating principle, structure, and characteristics. Thermopile detectors have a structure in which a large number of thermocouples are serially connected on a silicon substrate and their sensitivity increases as more thermocouples are used. This means that the larger the photosensitive area, the higher the sensitivity, because the number of thermocouples is proportional to the size of the photosensitive area. In bolometers, the photosensitive area uses a bolometer resistance made up of thermoelectric conversion materials, so the resistance temperature coefficient is the primary cause in determining bolometer sensitivity. Since bolometer sensitivity does not depend on the size of the photosensitive area, detectors can be fabricated that have a small photosensitive area yet no drop in sensitivity. Thermopile detectors are usually manufactured as single-element detectors with an ample photosensitive area or arrays with a small number of elements, while bolometers are manufactured as arrays with a larger number of elements than thermopile detectors. Hamamatsu thermopile detectors and bolometers 2 Product name Multi-element array Sensitivity enhancement Supply current Package atmosphere Rise time Thermopile detector Possible with larger pixel size (pixel size: 200 × 200 µm or larger) Possible Not required (thermal electromotive force) Nitrogen 1 ms or more Bolometer Possible with smaller pixel size (pixel size: 75 × 75 µm or smaller) Possible Required Vacuum 2 ms or more [Figure 1-1] Thermally isolated structure (thermopile detector) Thermocouple Thermopile detectors are thermal detectors that utilize the Seebeck effect in which a thermal electromotive force is generated in proportion to the incident infrared light energy. Thermopile detectors themselves have no wavelength dependence and so are used with various types of window materials for diverse applications such as temperature measurement, human body sensing, and gas analysis. Hot junction Metal A Metal B Cold junction ΔV Hot junction • Operates • Spectral ΔT Features 1-1 ΔT Thermopile detectors 1. at room temperature Cold junction n × ΔV response characteristics that are not KIRDC0046EA dependent on wavelength • No Our thermopile detector structures differ according to the type of device, namely, single/dual/quad element types and linear/area arrays. optical chopping is required, and voltage output can be obtained according to input energy. • Low cost • Long Single/dual/quad element types life 1-2 Structure In order to obtain a large output voltage, Hamamatsu thermopile detectors have many thermocouples that are serially connected on a silicon substrate to magnify the temperature difference between the hot and cold junctions. The hot junction side (photosensitive area) is designed to be a thermally isolated structure on which an infrared absorption film is attached. To make the thermally isolated structure, MEMS technology is used to process the membrane (thin film) to make it float in a hollow space. Our thermopile detectors use materials that have a large Seebeck coefficient (thermal electromotive force) and are easily formed by the semiconductor process. When infrared light enters a thermopile detector having the above mentioned structure, the hot junction on the membrane heats up and produces a temperature difference (ΔT ) between the hot and cold junctions accompanied by generation of a thermal electromotive force (ΔV). Single/dual/quad element types have large photosensitive areas and are manufactured by bulk processing technology with high workability. Etching is performed from the backside of the substrate to form the photosensitive area in a membrane state so that the hot and cold junctions are thermally isolated from each other to achieve high sensitivity. [Figure 1-2] Cross-sectional view (single/dual/quad element types) Infrared absorption film Cold junction Photosensitive area Isolation layer Substrate Hot junction Thermocouple KIRDC0106EA [Table 1-1] Hamamatsu thermopile detectors Type Number of elements Window material Spectral response range Package Main applications Single element 1 Anti-reflection coated Si 3 to 5 µm TO-18 Gas analysis, temperature measurement Dual element 2 3.9 µm, 4.3 µm TO-5 3 to 5 µm TO-8 Band-pass filter Quad element Linear array 4 16, 32 5 µm long-pass filter Area array Gas analysis 8×8 Flat package Temperature measurement TO-8 Temperature measurement, human body sensing 5 to 14 µm 3 Temperature characteristics To manufacture linear and area arrays, the gap between each element must be made narrow in order to reduce non-sensitive areas. To do this, the portion directly under each photosensitive area is selectively bored by surface processing technology so that the photosensitive area becomes a thin membrane-like structure. CMOS process technology is utilized to lay out the signal processing circuits on the same chip where the thermopile section is formed. [Figure 1-4] Temperature characteristics of sensitivity (single element type T11262-01, typical example) 60 58 56 Sensitivity (V/W) Linear and area arrays [Figure 1-3] Cross-sectional view (linear and area arrays) Thermocouple Cold junction Hot Infrared junction absorption film Isolation layer 54 52 50 48 46 44 Etching hole 42 Metal wiring 40 -20 -10 0 10 20 30 40 50 60 70 80 Element temperature (°C) Signal processing circuit Substrate Thermopile section Thermopile section KIRDC0107EA KIRDB0522EA [Figure 1-5] Temperature characteristics of element resistance (T11262-01, typical example) 150 Characteristics Sensitivity Thermopile sensitivity (Rv) is determined by the number of thermocouples as expressed by equation (1). 140 Element resistance (kΩ) 1-3 130 120 110 Rv = η: n: α: G: ω: τ: ηn α G 1 + ω2 τ 2 [V/W] ……… (1) 100 -20 -10 emissivity number of thermocouples Seebeck coefficient thermal conductivity angular frequency thermal time constant 0 10 20 30 40 50 60 70 80 Element temperature (°C) KIRDB0523EA Linearity Noise Thermal noise called Johnson noise in the element resistance is predominant in thermopile detector noise. Noise (VN) is expressed by equation (2). VN = k : T : Rd: Δf : 4 4k T Rd Δf [V rms] ……… (2) Boltzmann’s constant absolute temperature element resistance bandwidth Figure 1-6 shows an example of the relation between the input energy and output voltage. Thermopile detector output voltage is proportional to the input energy. [Figure 1-6] Output voltage vs. input energy (T11262-01, typical example) [Figure 1-7] Frequency characteristics (Typ. Ta=25 ˚C) 2 (Ta=25 ˚C) 10-1 1 0.3 × 0.3 mm 0 Relative output (dB) 10-2 Output voltage (V) 0.25 × 0.25 mm 10-3 10-4 -1 -2 2 × 2 mm -3 1.2 × 1.2 mm -4 10-5 -5 0.1 10-6 10-5 10-4 10-3 10-2 1 10 100 1000 10-1 Frequency (Hz) Input energy (W/cm2) KIRDB0561EA KIRDB0560EA Spectral response Frequency characteristics Since thermopile detectors have no wavelength dependence, their spectral response is determined by the transmittance characteristics of window materials. Spectral transmittance characteristics of typical window materials are shown in Figure 1-8. Figure 1-7 shows the frequency characteristics of thermopile detectors each having a different photosensitive area. Frequency response tends to decrease as the photosensitive area becomes larger. How to use 1-4 Single/dual/quad element types (1) Circuit not using thermistor In cases where the ambient temperature is constant or high precision measurement is not required, thermopile detectors can be used with a circuit that does not include a thermistor. [Figure 1-8] Spectral transmittance characteristics of window materials 100 Anti-reflection coated Si T11262-01 90 5 μm long-pass filter 80 8 to 14 μm band-pass filter Transmittance (%) 70 3.9 μm band-pass filter T11722-01 (reference light) 60 Si 50 4.3 μm band-pass filter T11722-01 (CO2) 40 30 4.4 μm band-pass filter 20 10 0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 Wavelength (μm) KIRDB0512EA 5 • Dual-polarity power supply type [Figure 1-9] Amplifier circuit (dual-polarity power supply type) C1 R2 -V 4 R1 2 method using a microcontroller is more common. Figure 1-11 shows a circuit example where the thermistor output signal is fed into the amplifier circuit. This type of circuit is used when high measurement accuracy is not required. The circuit shown in Figure 1-11 applies to both cases where the thermistor is externally connected to the thermopile detector or the thermistor is built into the thermopile detector. [Figure 1-11] Amplifier circuit with thermistor - 6 Vout 3 + +V +V 7 Ra Thermopile detector 3 7 + 6 2- +V Thermopile detector R1 Rb Gain = 1 + (R2/R1) fhigh = 1/(2πC1 R2) Vout 4 Thermistor Rth -V R2 C1 Rd Gain = 1 + (R2/R1) fhigh = 1/(2πR2 C1) KIRDC0049EA • Single KIRDC0051EB [Figure 1-10] Amplifier circuit (single power supply type) C1 R2 GND R1 +V 2 3 + Finding the resistance values (Ra, Rb, Rc, Rd) for the circuit with a thermistor To find the resistance value of Rd at which the thermistor output Vth is linear in the operating temperature range 1) Determine the operating temperature range (Tmin to Tmax). 2) Find the resistance value (Rh) of the thermistor Rth at Tmax. 3) Find the resistance value (Rl) of the thermistor Rth at Tmin. 4) Find the resistance value (Rm) of the thermistor Rth at an intermediate temperature between Tmin and Tmax. 5) Find the resistance value of Rd from equation (3). Rd = Rh Rm + Rl Rm - 2Rh Rl ……… (3) Rh + Rl - 2Rm 4 6 Vout 7 +V Thermopile detector To measure the thermopile output voltage Vout and check the voltage range where the thermistor output voltage (Vth) varies in the operating temperature range (when the measurement object's temperature is 25 °C) 1) Measure the thermopile detector output voltage (Voutmin) at Tmax. 2) Measure the thermopile detector output voltage (Voutmax) at Tmin. Gain = 1 + (R2/R1) fhigh = 1/(2πC1 R2) R4 KIRDC0050EB (2) Circuit using thermistor Output signals of a thermopile detector are temperature dependent. When detecting the temperature of an object in locations where the thermopile element temperature may drastically fluctuate, some means of making the output signals constant is required to ensure stable temperature detection. There are two methods to compensate for the temperature: one is to directly input the thermopile detector and thermistor signals into a microcontroller, and the other is to feed the thermistor output signal into the amplifier circuit. If high accuracy is necessary, the 6 Vth power supply type When using an op amp that operates from a single power supply, an error occurs near ground potential which is caused by the op amp’s offset voltage and nonlinearity. To cope with this, the thermopile detector is operated with one terminal biased. In the circuit shown in Figure 1-10, the op amp supply voltage is biased with dividing resistors R3 and R4. R3 Rc To find the resistance values of Ra, Rb and Rc 1) Find the voltage drop (Vb) of Rb and the voltage drop (Vc) of Rc, from the simultaneous equations (4) and (5). Voutmin = Vb Rd + Vc ……… (4) Rh + Rd Voutmax = Vb Rd + Vc ……… (5) Rl + Rd 2) Find the voltage drop (Va) of Ra. (b) T11264 series V: supply voltage 3) Determine the Rb value which should be smaller than Rth + Rd by at least two orders of magnitude. 4) Find the Ra and Rc values from the simultaneous equations (7) and (8). Temperature sensor (+) Vdd Vref Az_in Naz_hold_in Naz_on Thermopile detector Video LPF Hsp Vclk Reset Hclk KIRDC0113EA (1) Driving method V Rc ……… (8) Ra + Rb + Rc Vc = Vsp Horizontal shift register Preamplifier V Ra ……… (7) Ra + Rb + Rc Va = Vdd GND Vertical shift register Temperature sensor COM Va = V - Vb - Vc ……… (6) Linear and area arrays Linear and area arrays consist of a one- or two-dimensional thermopile array, shift registers and temperature sensor, and to which a preamplifier is hybrid-connected to amplify the output signal. Since the preamplifier is built into the same package, this reduces external noise and also simplifies the circuit configuration connected subsequent to the sensor. The voltage to linear and area arrays is supplied from a 5 V single power supply. The timing pulse to each terminal is input as shown in the timing chart below so that the signal (Video signal) from each element of the thermopile array is output in a time series. [Figure 1-13] Timing char t (T11264 series) Vsp Vclk Hsp Hclk Reset Az_in Naz_hold_in Naz_on [Figure 1-12] Block diagram 1 2 3 4 5 6 7 8 (a) T11263 series Vsp Vclk Hsp Hclk Reset Az_in Naz_hold_in Naz_on AD_trig Temperature sensor Temperature sensor (+) Thermopile detector Address switch Shift register LPF Video Preamplifier Video 1 2 1 2 3 4 5 6 7 8 1 1 2 3 4 5 6 7 8 9 Naz_on Naz_hold_in Vref Az_in Reset Phinor Sp Clk Vdd GND KIRDC0114EA (2) Temperature sensor KIRDC0112EA In linear and area arrays, a temperature sensor is mounted on the same chip where the thermopile is formed. When a constant current flows in this temperature sensor, a voltage signal can be obtained that is inversely proportional to the temperature. A Si diode is used in the temperature sensor, which typically has a temperature coefficient of approximately -2.2 mV/°C. [Table 1-2] Digital input description (linear and area arrays) Digital input Description Vsp, Hsp Input logic signals needed to start the vertical/horizontal shift register scans. These inputs are required to scan the shift registers. Vclk, Hclk Input logic signals needed to switch the vertical/horizontal shift register channels. The shift register scan speed can be adjusted by changing the clock rate. Reset Input logic signal for setting the video output to a fixed potential while the pixel output signals are not read out Video This outputs the thermopile detector signals in synchronization with the vertical/horizontal shift register scan timing. Az_in Naz_hold_in Logic signals for driving the internal amplifier Naz_on AD_trig Timing signal for acquiring the video signal into an external A/D converter, etc. This is used in synchronization with the logic signals to the input terminals. 7 [Figure 1-14] Temperature sensor forward current vs. temperature (T11264 series, typical example) 0.9 Forward voltage (V) 0.7 0.6 0.5 0.4 0.3 0.2 2-1 0.1 0 0 20 40 60 80 Temperature (°C) KIRDB0563EA New approaches We are further enhancing our in-house CMOS and MEMS technologies to develop thermopile detectors with higher performance and more sophisticated functions yet at a lower cost. We also plan to offer small thermopile detectors with a lens, which come in a wafer level package. 1-6 Bolometers Bolometers are small infrared sensors that do not require cooling. When infrared light enters a bolometer, the bolometer resistance heats up, causing a change in its resistance. This change is converted into a voltage for readout. Bolometers include a readout circuit to minimize intrusion of external noise. 0.8 1-5 2. (If=10 μA) 1.0 Applications Operating principle and structure Since bolometers are thermal detectors, the membrane (thin film) that absorbs infrared light must be thermally isolated from the substrate, so the structure has two long, thin legs called “beams” to support the membrane. The membrane is formed by sacrificial layer etching and floats about 2 µm from the substrate. Infrared light radiated from an object is absorbed by the infrared absorber on the membrane, causing the membrane temperature to increase and the bolometer resistance to decrease. The incident light level can be read out as a voltage signal by applying an electric current to the bolometer resistance. As the bolometer resistance material, we use a-Si (amorphous silicon) whose resistance greatly varies with temperature. A CMOS readout circuit (ROIC: readout integrated circuit) is fabricated on the substrate. [Figure 2-1] Cross-sectional view (one pixel) CO2 sensors Thermopile detectors are used for non-dispersive infrared (NDIR) detection type CO2 sensors. These CO2 sensors allow precision measurements with high accuracy (minimal error deviation from the true value). Bolometer resistance Infrared absorber Infrared light Isolation film a-silicon Temperature and human body sensing in specific areas Electrode Substrate (CMOS readout circuit) Thermopile linear and area arrays are used for temperature and human body sensing in specific areas such as for air conditioner operation control. These can detect locations where persons are present and the direction that a person moves. 8 Beam KIRDC0115EA [Figure 2-2] Enlarged photograph of photosensitive area (bolometer) If there are gas molecules such as air around the membrane, the heat absorbed by the membrane is conducted to the gas molecules, causing the bolometer sensitivity to drop. To prevent this, bolometers are sealed inside a vacuum package. Figure 2-3 shows a block diagram of a bolometer. The vertical shift register selects each pixel, and the amplifier array converts current changes to voltage changes, which are sampled and held, and are finally output from one line of the horizontal shift register. Figure 2-4 shows the readout circuit (one pixel) from the photosensitive area to the amplifier. This circuit has a reference resistance that is equivalent to the bolometer resistance, and both are connected in series so that changes in current equivalent to changes in the bolometer resistance are converted to voltage signals by a later stage amplifier. Vertical shift register Amplifier array Sample-and-hold circuit Output Horizontal shift register α : η : Vb : G : t : τ : ( ) 1 - exp - t τ ……… (1) temperature coefficient of resistance infrared absorptance supply voltage thermal conductance readout time thermal time constant If t >> τ in equation (1), the voltage sensitivity is expressed by equation (2). Rv = α η Vb ……… (2) G Noise Bolometer noise comes from several sources including thermal noise caused by temperature fluctuations in the bolometer resistance, 1/f noise resulting from factors such as bolometer materials and electrical conductivity of contact, temperature fluctuation noise caused by temperature changes in the membrane, photon noise caused by photons in the environment, and readout circuit noise caused by the amplifier and other devices. Temperature fluctuation noise and photon noise are low compared to other sources of noise, so thermal noise and 1/f noise are usually predominant. The total noise (VN) of a bolometer is given by equation (3). [Figure 2-3] Block diagram (bolometer) Photosensitive area α η Vb G Rv = KIRDC0116EA [Figure 2-4] Readout circuit (one element) Vref2 V J2 + V 1/f 2 + V TH2 + V PN2 + V ROIC2 ……… (3) VN = VJ : V1/f : VTH : VPN : VROIC : thermal noise 1/f noise temperature fluctuation noise photon noise readout circuit noise (1) Thermal noise Reference resistance 4k T Rbol (f 2 - f 1) ……… (4) VJ = Vdd C − + SR Bolometer resistance To sample-and-hold circuit k : T : Rbol : f1 : f2 : Boltzmann’s constant membrane temperature bolometer resistance lower limit of frequency bandwidth upper limit of frequency bandwidth (2) 1/f noise Vref K Vb 2 ln V 1/f = KIRDC0117EA ( ) f2 f1 ……… (5) K: 1/f coefficient 2-2 Characteristics Voltage sensitivity The voltage sensitivity of a bolometer is defined as the output voltage divided by the infrared light level incident on the photosensitive area. The voltage sensitivity (Rv) is expressed by equation (1). f1 and f2 are respectively defined by equations (6) and (7). f1 ≈ 1 4tstare ……… (6) t stare: correction period of reference output f2 = 1 2t ……… (7) 9 (3) Temperature fluctuation noise RV η V TH = 4k T 2 G f TF ……… (8) f TF : thermal equivalent noise bandwidth fTF is defined by equation (9). f TF = 1 4τ ……… (9) (4) Photon noise V PN = R V 8A σ k (T 5 + T BKG5) f TF ……… (10) η A : photosensitive area σ : Stefan-Boltzmann constant (5.67 × 10-8 W/m2 K4) TBKG : background temperature Noise equivalent power Noise equivalent power (NEP), which corresponds to the S/N, is used to indicate the relation between the bolometer signal output and noise. The NEP signifies the incident light level required to obtain a signal output equivalent to the total noise level and is given by equation (11). V N ……… (11) RV NEP = Noise equivalent temperature difference Noise equivalent temperature difference (NETD) is used to express the bolometer performance including the readout circuit. NETD indicates the temperature change that occurs when infrared light with a power equivalent to NEP enters the bolometer, and is given by equation (12). NETD = Fno : ɸλ1-λ2 : L λ1-λ2 : λ : 10 4Fno 2 V N RV A ɸλ1-λ2 π Lλ1-λ2 ……… (12) F number of optical system transmittance of optical system temperature contrast in wavelength interval wavelength