RU2242728C2 - Heat flux sensor - Google Patents

Abstract

FIELD: microelectronics and optoelectronics.

SUBSTANCE: proposed sensor has silicon substrate that mounts diaphragm and thermocouples disposed in tandem. Hot and cold junctions of thermocouples are disposed on thin and thick parts of diaphragm. Thermocouples are interconnected to form pile and are made of single-crystalline silicon and aluminum; diaphragm is made of single-crystalline calcium fluoride insulating film.

EFFECT: enhanced sensitivity and manufacturability of device.

1 cl, 1 dwg

Description

The invention relates to the field of microelectronics and optoelectronics, in particular to the design of the heat flux sensor and can be used to measure heat transfer processes within the structure, to control the temperature inside thermostatically controlled volumes; temperature control in buildings to conserve energy used; to check the effectiveness of thermal insulation; to study and test the problems of radiation and convection (heat and mass transfer) in heat transfer systems; to study body exchange in wind tunnels.

Known designs of heat flux sensors of the thermoelectric type, made on the basis of a thermocouple battery. The main element of a thermocouple is a junction between two dissimilar materials. The temperature measurement using a thermocouple is based on the Seebeck effect.

where ΔE is the value of thermopower, V; α _{T is the} coefficient of thermoEMF, V / K; ΔТ = Т ₂ -T ₁ is the temperature difference between hot and cold junctions, K. Thus, measuring the temperature with a thermocouple reduces to measuring the thermopower, which the thermocouple develops at a strictly fixed temperature of one of the junctions [1].

As a thermoelectric pair, a combination of materials is selected that creates a sufficiently high thermoEMF (the difference in the coefficients of thermoEMF of materials should be as large as possible) and suitable for a given operating temperature range. To increase the sensitivity, the series connection of several thermocouples is used. Also, an increase in sensitivity can be achieved by using materials with a high thermoelectric coefficient as a thermoelectric pair. Such materials are known to be semiconductors.

The advantage of semiconductor thermoelectric sensors is the simplicity of manufacture and the absence of a power source, a linear output signal, which in most cases does not require amplification even with a small difference in the temperature of the junctions due to the large value of the thermoelectric coefficient of semiconductor materials. The falling stream is absorbed by the perceiving, blackened platform. In the dynamic mode of operation of the thermal sensor, thermal stabilization is absent. In static mode, thermal stabilization is carried out due to forced cooling.

A known design of a heat flux sensor, in which a semiconductor-metal contact is used as a thermoelectric pair, for example, silicon-aluminum [2]. The cold ends of this battery are located on a heat sink with a constant temperature (thermostat), and the heated ends are located in the region of the incident heat flux.

The sensor sensitive element is a silicon structure consisting of an n-type conductivity substrate in which a system of p-type conductivity paths is formed by diffusion. These silicon tracks by metallization with aluminum are connected in series by the output signal, and the pn junction is used to isolate hole-type tracks from the carrier base.

The sensitivity of these devices is about 15-20 mVm ² / kW, and the response time is about 5 · 10 ^-3 s.

The disadvantage of these heat flow sensors is the low sensitivity due to:

-high concentration of holes in diffusion layers;

- significant heat transfer due to the thick substrate (350-500 microns), and its high specific thermal conductivity, which leads to low thermal resistance of this structure and a decrease in the measured temperature difference.

The main method of reducing heat transfer is to reduce the heat flux along the substrate from the heated blackened spot to the periphery (contacts) by reducing the thickness of the substrate on which the thermocouple system is formed.

A known design of the heat flux sensor, the main feature of which is the use of a membrane as a carrier base, which provides a greater temperature drop, and, consequently, a larger output signal [3].

The device [3] is a 3.5 mm x 3.5 mm membrane, cold and hot junctions are located on the thick and thin parts of the membrane, respectively.

The disadvantage of the design is:

- the complexity of the technology for producing membranes of the same thickness throughout the field (from 30 to 40 microns)

- high thermal conductivity of silicon,

- the need to form p-type silicon elements in an n-type substrate by ion implantation or diffusion methods up to p = 10 ¹⁸ cm ^-2 . In this case, the coefficient of thermoEMF in silicon decreases by almost an order of magnitude.

- increase in leakage currents and own heating of thermocouples.

To increase the sensitivity of the heat flux sensors, it is desirable to use dielectric layers having less thermal conductivity and good dielectric insulation as membranes.

The known design of the heat flux sensor [4], selected as a prototype, containing a silicon substrate on which a membrane of an amorphous layer of silicon oxide (or nitride) silicon and polysilicon-aluminum (or gold) thermocouples are connected in parallel thermally and in series electrically. Hot junctions of thermocouples are located in the center of a thin oxide / nitride membrane, and cold ones around a silicon ring of rigidity [4].

On the underside of the substrate, silicon is etched to an oxide film. The polysilicon layer is doped to a concentration of 10 ¹⁸ cm ^-3 , after which mesa structures are etched in it and metal (Al or Au) contacts are created by photolithography methods.

The use of polysilicon structure on oxide eliminates the problem of isolation of the working elements of the sensitive element. Another advantage is that it is possible to obtain such a structure when all the working elements are located on thin dielectric silicon oxide isolated from the silicon monolith. Such a structure has a high thermal resistance, because the oxide film is thin and has low thermal conductivity.

The disadvantages of the known design [4] is the lack of manufacturability due to the need for an additional operation of ion doping, which complicates the manufacturing process, and low sensitivity due to:

- low coefficient of thermoEMF of a polysilicon film (116-154 μV / degree [3]) in comparison with single-crystal silicon (575 μV / degree [5]), which leads to a decrease in the sensitivity of the sensor by more than 5 times (with the same size of the receiving site) ,

- the presence of grain boundaries in polysilicon, which leads to segregation of the dopant, in particular, to an inhomogeneous distribution of resistivity in polysilicon, which, in turn, leads to instability of the sensor sensitivity both depending on the material and temperature,

- increased scattering in polysilicon, which reduces the mobility of the carriers (the capture of carriers by traps, the content of which is increased in polysilicon, works similarly) and leads to a decrease in the thermoelectric coefficient and sensitivity of the sensor,

часто большой и отрицательный ≈-2,5×10^-2К^-1 при температуре 300К, в то время как для монокристаллического кремния температурный коэффициент сопротивления составляет +2×10^-3K^-1 [6].- the fact that the temperature coefficient of resistance of polysilicon

often large and negative ≈-2.5 × 10 ^-2 K ^-1 at a temperature of 300 K, while for single-crystal silicon the temperature coefficient of resistance is + 2 × 10 ^-3 K ^-1 [6].

The objective of the invention is to increase the sensitivity of the sensor and its manufacturability.

This object is achieved in that the thermionic heat flux type sensor contains a silicon substrate on which the membrane and thermocouples are arranged in series, the hot and cold ends of which are located on the thin and thick parts of the membrane, respectively, and the thermocouples are connected to the battery, while the thermocouples are made of single-crystal silicon, and the membrane is made of a single crystal dielectric film of calcium fluoride.

The invention is illustrated by the drawing, which shows the construction of a thermoelectric type heat flux sensor based on a single-crystal Si-Al thermocouple measuring small-level heat fluxes.

The sensor contains a silicon substrate (1), on which a membrane of calcium fluoride (2) and thermocouples of single-crystal silicon (3) and aluminum (4) are sequentially placed. A layer of calcium fluoride (2) is grown on a silicon substrate (1) by MBE. Then, an epitaxial layer of single-crystal n-Si was grown on a CaF ₂ (2) layer, in which sensitive elements (3) in the form of mesostructures are formed using plasma-chemical etching, then Al is deposited (4).

The sensor operates as follows. The heat flux (5) falls on the receiving platform (6) located in the center, and the periphery is in thermal stabilization, which is provided by a refrigerator or a casing. A temperature gradient occurs in the sensor body. The output signal is proportional to the temperature difference between the hot junction (6) and the cold junction (7).

Sensors with an epitaxial CaF ₂ layer have an absolutely even membrane profile and a thinner sensor. The thin part of the plate, namely, it determines the temperature gradient, is 2 μm (the thickness of CaF _{2 is} 1 μm, the thickness of single-crystal Si over CaF _{2 is} less than 1 μm), which increases the output signal and, therefore, the sensitivity of the sensor. Being a good dielectric, CaF ₂ eliminates leakage currents through the substrate. Monocrystalline silicon has a higher coefficient of thermoEMF and carrier mobility than polysilicon, which also leads to an increase in the sensitivity of the sensor. The technological route of manufacturing the sensor is shorter by one operation, because the operation of ion doping is excluded, which increases the manufacturability of the proposed heat flow sensor.

To select the optimal geometric dimensions of the sensor, the thermal field in the membrane body was calculated. The location of hot junctions on the receiving site in the center of the membrane, and cold junctions at the membrane edge leads to the greatest output signal. The connection of thermocouples in parallel at the input and in series at the output allows you to get an output signal equal to the sum of the signals of a single thermocouple:

E = αΔTn

where n is the number of thermocouples.

Literature

1. Kireev P.S. Semiconductor Physics. - M., 1969.

2. Rubtsov, Dikareva R.P., Kiselev G.A., Averkov E.I. Semiconductor receivers of radiant heat fluxes, Bulletin of the Siberian Branch of the USSR Academy of Sciences, Series of Technical Sciences, vol.6, “Science”, Novosibirsk.

3. Jiangou Lu., Bing Xiong and Chenglu Lin, Structure and properties of silicon-metal thermopile, Sensors and Actuators A 35, 1993.

4. Choi and Wise, Micromachined thermopile IR Detector, Micromachined Transducers Sourebook, 1993.

5. Ioffe A. Selected Works. - L .: Nauka, 1975, p. 384.

6. Polycrystalline semiconductors. Physical properties and applications./ Ed. G. Harbeke. - M.: Mir, 1989, p. 252.

Claims (1)

The thermoelectronic type heat flux sensor contains a silicon substrate on which the membrane and thermocouples are arranged sequentially, hot and cold junctions of which are located on the thin and thick parts of the membrane, respectively, and the thermocouples are connected to a battery, characterized in that the thermocouples are made of single-crystal silicon and aluminum, and the membrane is made of a single crystal dielectric film of calcium fluoride.