DE10161214A1 - Gas sensor and method for the detection of hydrogen according to the principle of work function measurement, and a method for producing such a gas sensor - Google Patents

Abstract

The invention relates to a gas sensor which works on the principle of work function measurement for the detection of hydrogen, hydrogen sulfide and / or other gases which are chemically similar to hydrogen, the gas sensor comprising a sensitive layer (8) and by the adsorption of molecules of the the work function of the sensitive layer (8) can be changed on the sensitive layer to be detected. The gas sensor comprises a sensitive layer (8) which has platinum and possibly titanium. For example, a titanium-containing intermediate layer (8 ') is arranged between a platinum-containing layer (8) and a carrier (6). The invention also relates to a method for producing such a gas sensor and a corresponding detection method.

Description

The invention relates to a gas sensor and a method for the detection of Hydrogen, hydrogen sulfide and / or other gases that are chemically similar to Are hydrogen, according to the preambles of claims 1 and 21. Such a gas sensor and such a method are e.g. B. known from DE 42 39 319 C2. The invention also relates to a method for producing such a gas sensor.

Detection is the measurement of the presence and / or the concentration understood the gas in question. Since such a gas sensor is generally used for Examination of the composition of gases is used in the following Term "gas sensor" used; the sensor is also for the detection of substances suitable in liquids.

Hydrogen (H ₂ ) is a colorless, odorless gas which, at concentrations between 4% and 73% in air, becomes an explosive mixture with the oxygen contained in air. Since no greenhouse gases or other pollutants are released when hydrogen is burned with oxygen to water, hydrogen is a future energy source. B. for motor vehicles, airplanes, missiles and other mobile applications. For example, hydrogen can be obtained from water using solar energy and may replace fossil energy sources in the future.

However, the use of hydrogen as a mobile energy source is fraught with dangers, since even small amounts of hydrogen escaping from the energy store can lead to an explosion. In order to contain this danger, gas sensors could be used for control purposes, with which the hydrogen concentration in the vicinity of an energy store is measured and any leaks in the energy store can thus be detected. Due to the rapid dilution of hydrogen in air, these must respond well below 4% H ₂ concentration, e.g. B. at 1000 ppm (0.1%). Hydrogen is also a key gas for fire detection.

As gas sensors for hydrogen in the prior art, for. B. electrochemical Cells used, however, typically have a limited lifespan a year, a small measurable concentration range and high Show cross-sensitivity to other gases. Furthermore are Conductivity sensors are known, which consist of a gas-sensitive semiconductor, which accumulates of hydrogen changes its conductivity. However, such sensors require due to their high working temperature of over 200 ° C and a high heating output therefore unsuitable for applications whose high energy consumption is relative to the low energy densities of batteries are not covered for adequate periods of time can such. B. in cell phones, laptops or standing cars.

Another type of gas sensor that works on the principle of work function measurement work, on the other hand, have only a low energy requirement. With such Gas sensors, the detection is based on the fact that molecules of the to be detected Substance to be adsorbed on the surface of a sensitive material. hereby changes the work function of the sensitive material and thus the electrical potential, what can be measured for example by a field effect transistor (FET) structure can. In principle, all materials come from the insulator to the sensitive material Metal in question.

An example of such a so-called gas sensitive FET (GasFET), as z. B. is known from the above DE 42 39 319 C2, is shown schematically in Fig. 2. In a field effect structure 1 there is a channel 3 between a source region S and a drain region D, through which a drain current I _DS can flow. There is an air gap 4 between a passivation layer 2 applied thereon and the gate electrode G, into which the gas to be investigated is diffused or flowed in by external influences (for example a pump). On its side facing the air gap 4 , the gate electrode 6 is covered with a sensitive material 8 , to which molecules of the substance to be detected attach. This creates a dipole layer or a chemical compound on the surface of the sensitive layer 8 and thus an electrical potential which influences the conductivity of the channel 3 and thus the drain current I _DS across the air gap 4 . A change in I _DS can be used to infer the change in contact voltage and thus the change in work function Δφ on the surface of layer 8 , which in turn is a measure of the concentration of the substance to be detected.

The drain current is simply calculated from:

I _DS = µCW / L (U _G - (U _T + Δφ)) U _DS ,

where µ is the electron mobility in the channel, C the capacitance between gate G and channel 3 , W / L the width-to-length ratio of the channel, U _G the gate voltage, U _T the threshold voltage, Δφ the change in contact potential, Δφ times the elementary charge e the work function change and U _{DS is} the voltage between source and drain (drain voltage). If, for example, the change in the drain current I _DS is measured as a sensor signal when gas is applied, this gives a measure of the concentration of the substance to be detected.

The semiconductor components required for the production of GasFETs are included Mass production extremely inexpensive to manufacture.

Different types of GasFETs are known. The gas sensor shown with an air gap 4 between the gate G and the passivation layer 2 over a channel is generally referred to as a suspended gate FET (SGFET). From DE 42 39 319 C2 it is known to produce a SGFET from two components, namely a substrate with a field effect structure without a gate and a gate structure placed thereon (Hybrid Suspended Gate FET, HSGFET). Conversely, it is also possible to mount a prefabricated CMOS transistor using flip-chip technology on a substrate on which the gate electrode and the sensitive layer are applied, as described in DE 198 14 857 A1 (hybrid flip chip FET, abbreviated) HFC-FET). The contacting and fixing is done with a conductive adhesive. A somewhat different structure is known from DE 43 33 875. Here the sensitive layer is not directly above the channel of the FET, but is spatially separated from it, namely arranged on an extension of the gate electrode of the transistor. The potential changes caused by the adsorption of molecules on the sensitive layer are capacitively coupled into the channel area via the extended gate electrode (capacitive controlled FET, abbreviated CCFET). This structure has the advantage that the gate capacitance C is larger than in the SGFET, since there is no air gap between the channel and the gate, so that the measured change in the drain current and thus the sensor signal according to the above formula is greater than z. B. in a SGFET with otherwise the same parameters.

Before the development of gas-sensitive FETs with an air gap (suspended gate FETs), the gate electrode arranged directly above the channel was made of a sensitive material that was permeable to the gas to be detected. The change in work function occurs due to the dissolving of the gas in the sensitive layer. From I. Lundstrom, "A Hydrogen-Sensitive MOS Field Effecting Transistor" Applied Physics Letters 26 , 55-57 , 1975 , e.g. B. such a sensor with a palladium layer for the detection of hydrogen is known. However, since the hydrogen in this sensor first has to be dissolved in the palladium film before a sensor signal is triggered, the response times are relatively long. Furthermore, iridium is also known as a sensitive material for hydrogen, which also has long response times.

As an alternative to a GasFET, work function changes can also be made using the Kelvin Method ("Vibrating Capacitor Method") can be measured. Here is a transducer (e.g. a metal plate) excited to vibrate by means of a piezo element. At the A capacitor plate is attached to the end of the transducer gas-sensitive material is coated and a second, differently coated, capacitor plate opposite. The oscillation of the capacitor plates creates an alternating current generated which is the difference of the work functions on the two Capacitor plates depends.

The aim of the present invention is to provide an inexpensive and durable gas sensor and to provide an inexpensive process by which hydrogen, Hydrogen sulfide and other gases that are chemically similar to hydrogen with short Response times and over a wide concentration range can be detected can. Another goal is to provide a simple and inexpensive process for To provide manufacture of such a gas sensor.

To this end, the invention provides a gas sensor for the detection of hydrogen, Hydrogen sulfide and / or other gases that are chemically similar to hydrogen, ready to who works on the principle of work function measurement, with the gas sensor sensitive layer and by the adsorption of molecules of the detecting gas on the sensitive layer, the work function of the sensitive layer is changeable. The gas sensor comprises a sensitive layer that has platinum.

Furthermore, the invention provides a method for the detection of hydrogen, Hydrogen sulfide and / or other gases that are chemically similar to hydrogen, in which caused by the adsorption of the gas on a sensitive layer Change in work function is measured. The sensitive layer in turn points Platinum on.

Finally, the invention provides a method for producing a hybrid suspended Gate FETs (HSGFETs) with the following steps: (a) Fabricate one Field effect structure (measuring FET or measuring transistor) with a channel between a source and a drain area; (b) applying a platinum-containing layer to a support; (C) Placing the carrier on the field effect structure, so that the platinum layer the channel is facing and there is an air gap between the two.

Platinum is a precious metal. B. unsaturated organic Compounds are easily bound, which is the excellent catalyst properties explained by platinum. Due to its surface activity, platinum also helps Room temperatures quickly indicate hydrogen concentration changes, but does nevertheless a relatively low sensitivity to moisture. Furthermore you can Platinum layers can be produced simply and therefore inexpensively. For the sensitive layer of the gas sensor, platinum can be used both in pure form and in alloys.

The sensitive layer preferably also has titanium. One can do this Layer of a titanium-platinum alloy can be used, but is preferred a titanium-containing layer between a platinum-containing layer and a support Intermediate layer arranged. The carrier is, for example, a capacitor plate of a Kelvin Probe or the gate electrode of an SGFET. The titanium-containing intermediate layer facilitates the adhesion of the platinum-containing layer to the carrier. Because titanium against Hydrogen inert, titanium-containing materials provide a good base for thin ones Platinum layers, since hydrogen diffused through the platinum layer then at Underground does not respond. Therefore titanium and titanium alloys are also suitable for the non- sensitive layer in a referenceFET for referencing the sensor signal (see below).

In the preferred embodiments, the gas sensor is designed as a GasFET, which the additional functions described below for suppressing Interferences such. B. due to temperature fluctuations and humidity.

On the one hand, it has been shown that the gate voltage U _G drifts strongly at high air humidities. The reason for this is that a moisture film forms on the surface of the passivation layer 2 and / or the sensitive layer 8 , in which a leakage current can flow between regions of different potential. Such a current flow between the gate and the channel, as well as a change in the potential of the moisture film, have a strong influence on the drain current in the channel directly below, since the effect couples into the channel without an air gap and therefore via a high capacitance. As the measurement of the drain current shown in FIG. 3 in a SGFET according to the state of the art shows at atmospheric humidities of alternately 0% and between 10% and 90%, the baseline of the sensor signal (here the gate voltage U _G ) drifts at humidities of more than about 40 % so strong that a concentration measurement is no longer possible.

Second, the electron mobility µ in channel 3 and the threshold voltage U _{T of} the field effect structure 1 are strongly temperature-dependent, so that the drain current I _DS decreases sharply with increasing temperature, like the measurement of the drain current shown in FIG. 4 in a SGFET according to the prior art at temperatures between -5 ° C and 65 ° C shows. The sensitivity of the gate voltage to the temperature can be up to one volt / K. This means that with a temperature change of one degree, a signal change occurs that is equivalent to a strong gas application. This is insufficient for operation in the temperature range between 0 and 60 ° C, which is required for most applications. In order to be able to measure the influence of the work function change on I _DS at all, the sensor would have to be heated to a constant temperature, which would negate the advantage of the low energy requirement of such sensors.

Another interference effect with gas sensors based on the principle of work function measurement are so-called cross-sensitivities, i.e. H. the sensitive layer doesn't just speak a single substance. In particular, the adsorption of Water caused a work function change in weight, because at room temperature always There are high levels of humidity, so the moisture concentration is usually around one Is many times higher than that of the substances to be detected.

To suppress the interference effects due to temperature fluctuations, the Gas sensor preferably a first field effect structure with a channel between one Source and a drain area (MessFET); and a first gate electrode with a sensitive layer, with a change in the work function of the sensitive Layer, e.g. B. by the adsorption of hydrogen molecules on the sensitive layer, the drain current in the channel of the measuring FET can be influenced. In addition, the Gas sensor a second field effect structure with a channel between a source and a Drain area (referenceFET or reference transistor) for referencing the Temperature response of the drain current of the measuring FET.

This structure is based on the knowledge that the temperature response of the sensor signal essentially results from the temperature response of the electron mobility μ and the threshold voltage U _T, that is to say it has its cause in the channel of the field effect structure. Since the drain current in the channel of a reference FET is exposed to the same temperature influences as that in the measuring FET, the temperature effect can be eliminated from the measured sensor signal by comparing the two currents. The sensor can therefore be operated in any temperature range (in principle between approx. -60 and 200 ° C) and does not need to be heated to a constant temperature.

The referenceFET is preferably technologically, electrically and geometrically the same constructed like the MeßFET and can on a common substrate with this are produced so that there are hardly any additional costs for temperature referencing of manufacture.

According to one embodiment, the drain current is not in the channel of the referenceFET can be influenced by a gate electrode; d. H. for a SGFET, the channel is the ReferenceFETs are not covered by a gate electrode, or the air gap over the channel of the reference FET is so widened compared to the air gap above the measuring FET that the conductivity of the channel practically no longer by a Work function change is affected at the gate electrode. This structure has the advantage that the Drain current in the channel of the reference FET not from possible interference effects on one Gate electrode, but only depends on the temperature.

According to another preferred embodiment of the invention, the Drain current in the channel of the referenceFET, however, by changing the work function of one second gate electrode that can be influenced compared to the substance to be detected contains non-sensitive material. Non-sensitive means that the material is towards the Detecting substance responds at least significantly less than the sensitive Material platinum. Preferably, the sensitive material of the first and the non- sensitive material of the second gate electrode, however, roughly the same Cross-sensitivities, i.e. sensitivities to other substances. This has the great advantage that not only the Influence of temperature, but also the influences of cross-sensitivity are eliminated, since both transistors are equally exposed to these disturbances. In particular it is advantageous to choose a material for the non-sensitive layer that approximately responds to air humidity as strongly as platinum.

Therefore, for the non-sensitive material in the gate electrode of the reference FET preferably used titanium, since titanium is inert to many gases and platinum and Titan both have about the same sensitivity to moisture and ammonia. Alternatively, silicon nitride can also be used

The sensitive and the non-sensitive material are particularly preferred on one common gate structure applied. The gate structure is e.g. B. a substrate made of silicon or silicon carbide, which is used to produce a SGFET on a field effect structure is put on. The advantage of the silicon substrate is its smooth surface. Before the The gate structure is preferably initially placed overall with a titanium-containing one Material coated, and then in the area of the channel of the MeßFETs on the titanium-containing a platinum-containing layer applied. Aside from the general Simplicity in production, this has the particular advantage that the titanium-containing layer serves as an adhesion promoter for the sensitive platinum layer. Without one Intermediate layer, platinum adheres poorly to silicon.

In some embodiments, the measurement and referenceFETs each have one common drain or a common source area. Are preferred the drain and source regions of the two transistors, however, spatially apart separated so that they do not influence each other as much as possible. For example, they are in housed two separate doping wells of a silicon substrate.

In a particularly simple to manufacture embodiment, the gate structure made entirely of a platinum-containing material, e.g. B. made of platinum.

The temperature referencing is particularly preferably carried out with the aid of a sensor circuit with which the difference between the drain currents of the measurement and reference FETs is kept constant by readjustment of the voltage U _G at the gate electrode of the measurement FET. Alternatively, it is not the difference, but another linear combination of these currents that is kept constant. B. 1.5 scaled before it is subtracted from the drain current of the referenceFET. Such scaling can compensate for structural differences between the two transistors, so that greater freedom is given in the design of the gas sensor and the permissible manufacturing tolerances are greater. The said readjustment of the gate voltage U _{G also} has the advantage over a direct measurement of the drain current I _DS that the magnitude of the readjustment of U _G corresponds directly to the change in contact voltage Δφ. In addition, the gas space above the channel thereby always has the same potential (as I _DS = constant shows), and thus possible drifts due to changing potentials across the channel are excluded.

To compensate for the above The channel of the MeßFETs preferably from a so-called guard electrode to protect against surrounded by electrical interference.

The guard electrode prevents leakage of leakage currents and capacitive interference, and it prevents charge balancing operations on the Surface. In particular, at high atmospheric humidity, the formation of a Moisture film on the passivation layer of the FET and on the sensitive layer Leakage currents flow on the surfaces, which affect the electrical conductivity of the Channel. The result would be a falsification of the measurement result. The Guard electrode, which is kept at a constant potential, for example, interrupts this charge exchange, and the influence of moisture is at least considerably reduced.

The two above are particularly preferred. Suggested solutions to each other combined, i.e. the gas sensor with both a referenceFET and Guard electrode (s) equipped to both humidity and temperature influences compensate. Such a sensor delivers reproducible even at room temperatures Sensor signals and therefore does not need to be heated to keep the temperature constant hold and / or reduce the humidity. The gas sensor therefore has only in operation low energy consumption in the micro to milliwatt range is in production inexpensive and therefore ideal for mobile and battery-powered applications.

The guard electrode preferably forms a closed ring (guard ring) the channel of the measuring FET. If available, the referenceFET is also preferably included equipped with its own guard electrode or a guard ring. Alternatively, you can a single guard electrode surround both channels.

The guard electrode consists of a conductive material, preferably a metal, z. As aluminum, platinum or gold, and can by any thin film process applied to an insulator or passivation layer on the field effect structure be, e.g. B. by electrochemical deposition, sputtering = cathode ray Sputtering, reactive sputtering, vapor deposition, spin coating, sublimation, epitaxy or Spraying. Sputtering accelerates ions in a vacuum and as a beam directed onto a target, whereby atoms are shot out of the target and as a homogeneous, compact layer on the surface to be coated deposit. The thickness of the guard electrode thus produced is, for. B. between 10 and 500 nm.

A step in the insulator or passivation layer around the channel is preferred embedded on which the guard electrode is arranged. This arrangement offers itself especially when in the field effect structure Passivation layer is arranged over the channel and a recess in the step Side walls of the recess is integrated. With a SGFET, a recess in the Passivation layer serve as a spacer for the gate electrode.

The area enclosed by the guard electrode is advantageous whenever possible small, so that there are no charges or leakage currents within this range can flow. The guard electrode should therefore be as close as possible to the channel be placed. However, it has been shown that the potential of the guard electrode - if this z. B. is kept at a constant potential - a disturbing Can influence the drain current in the channel. The guard electrode is preferred therefore spaced so far from the respective channel that the drain current in the channel is not is significantly influenced by the potential of the guard electrode. Preferably the distance is 1 to 15 microns, z. B. about 5 microns.

Another preferred way to rule out the potential of the Guard electrode has an interference effect on the drain current in the channel, is that Potential of the guard electrode is the potential of the gate electrode of the measuring FET equate. In this way there are none between the gate and guard electrodes Differences in time and therefore none in the air gap due to the guard electrode generated electrical fields that could trigger a leakage current. Another It is possible to put the guard electrode at a constant potential, e.g. B. 0 V (ground)

The gas sensor of the present invention can be used as a Kelvin probe as well Suspended Gate FET (SGFET), or as a Capacitive Controlled FET (CCFET). In the case of the SGFET in particular, the channels of the measuring and possibly the referenceFET are preferably meandering, d. H. the area between source and drain areas is in the plane parallel to the passivation layer is serpentine. This will help space-saving utilization of the substrate area a favorable width-length ratio W / L of the transistor of 10,000, for example, so that a high signal-noise Ratio is achievable. With CCFET, such a widening of the channel is not absolutely necessary, as the one caused by the change in work function Potential change here is not transmitted via an air gap and therefore with a larger one Capacitance C is coupled into the channel. An alternative construction of the CCFET is preferred used, in which the extended gate electrode, by which the potential change on sensitive material is electrically coupled into the channel of the measuring FET, entirely covered by a passivation layer and therefore less exposed to interference is. This causes voltage fluctuations ("floating") of the gate electrode reduced.

The sensitive layer and possibly the titanium-containing intermediate layer are preferably as Thin layers formed. The layers are e.g. B. by electrochemical Deposition, sputtering, reactive sputtering, vapor deposition, spin coating, sublimation, Epitaxy or spraying applied to the carrier, whereby layers with approx. 10-500 nm layer thickness can be created. Alternatively, however, thick-film technology can also be used be used, e.g. B. platinum or titanium atoms in a polymer layer brought in.

The gas sensor is preferred in an application where a low Power consumption is important, e.g. B. a motor vehicle at rest or in systems in which the functionality in the event of a power failure must be guaranteed. It is also thought that the signal from the sensor via radio to be transmitted to a monitoring station. The gas sensor stands out opposite known sensors for hydrogen detection namely by an extremely low Energy consumption and is therefore also suitable for battery operation. The explosion Protection conditions are with a sensor that is like the gas sensor Room temperature or just slightly above it, also much easier and cheaper to perform. For example, the gas sensor in a motor vehicle with hydrogen as the fuel are used to detect leaks in the energy storage. Typical too detecting concentrations are then 100 ppm to 4% hydrogen concentration. Of the gas sensor can also be used for leak detection in ultra-high vacuum systems be used. To do this, the system is filled with hydrogen, and typical leaks result in a gas concentration of around 10 ppm. Also here is a portable device with low power consumption is an advantage. Another application is z. B. with oil cooled high-voltage transformers. In these, z. B. at one Splitting arcs of hydrogen from methane. The hydrogen can correspondingly high concentration lead to an explosion of the transformer. Here the sensor is suitable for monitoring the hydrogen content.

The invention will now be described with reference to exemplary embodiments and the accompanying Drawing explained in more detail. The drawing shows:

Figure 1 is a schematic sectional view through a gas sensor according to a first embodiment.

Figure 2 is a schematic sectional view through a gas sensor in SGFET design according to the prior art.

Fig. 3 is a graph of the drain current at a gas sensor according to the prior art at various relative humidities;

Fig. 4 is a graph of the drain current at a gas sensor according to the prior art as a function of temperature;

Fig. 5 is a schematic plan view of an embodiment of a field effect structure;

Fig. 6 is a schematic diagram of a sensor circuit;

Fig. 7a, b show schematic sectional views of a second embodiment of a gas sensor;

Fig. 8 is a schematic sectional view of a third embodiment of a gas sensor;

9 is a graph of the drain current at a gas sensor with guard electrode and at a gas sensor with guard electrode and ReferenzFET at various relative humidities.

10 is a diagram of the drain currents in the reference and in MeßFET and whose variation in a gas sensor with Temperaturreferenzierung in dependence on the temperature.

11 is a diagram of the sensor signal U as a function of the applied hydrogen concentration.

Fig. 12 is a diagram of the sensor signal U versus time t partial pressures of hydrogen p for different applied;

Fig. 13 is a bar graph of the work function change of platinum when exposed to various gases at room temperature and at 130 ° C.

Figure 14a, b show schematic sectional views of a fourth embodiment of a gas sensor during the preparation of (a) and in finished state (b).

Fig. 15 current-voltage characteristics of a field-effect structure of FIG. 8 at different voltages U _k.

Functionally identical or similar parts are the same in the drawing Reference signs marked.

Fig. 1 shows a GasFET according to the invention in suspended-gate type, both with Temperaturreferenzierung by a ReferenzFET (Reffet), is as well provided with guard electrodes 10. In this example, platinum was used for the sensitive layer 8 of the measuring FET (measuring FET) and titanium for the non-sensitive layer 8 'of the reference FET, so that the gas sensor responds to hydrogen, hydrogen sulfide and other substances which are chemically similar to hydrogen. The titanium layer 8 'covers the entire underside of a gate structure 6 , and a sensitive platinum layer 8 is applied to the titanium layer in the area above the measuring FET.

The measuring FET and the reference FET are in two separate, e.g. B. p-doped wells 11 and 11 'arranged in a silicon substrate 12 . A channel 3 , 3 'runs between the corresponding n ⁺ -doped source-S and drain regions D of the measuringFET and the RefFET in the p-doped well 11 , 11 '. The arrangement of the two field effect structures in separate troughs 11 , 11 'has the advantage that the FETs cannot influence one another electrically. In particular, no current can flow between the transistors, since barrier layers form at the boundaries of the wells 11 , 11 'to the substrate 12 . In addition, the field effect structures z. B. by applying a voltage to the tubs 11 , 11 'specifically in their electrical properties, in particular their threshold voltage U _T , are influenced.

A passivation layer 2 is applied to the substrate 12 , which on the one hand electrically isolates the field effect structures and on the other hand protects against environmental influences such as e.g. B. Protects oxidation. Silicon nitride is preferably used for the passivation, since it is inert to most substances, ie it does not show its own work function change when gas is applied. Above the channels 3 , 3 ', depressions are arranged in the passivation layer 2 , each of which is covered by a part of the gate structure 6 (also called sensor cover) and thereby form the air gaps 4 , 4 '. In the area of the depressions, the passivation layer 2 is only a few micrometers thick, and the distance between the passivation layer 2 and the gate structure 6 (air gap height) is approximately 1-3 μm. Here, the gate structure 6 as a whole forms the gate electrode G and is manufactured, for example, from highly doped silicon or a highly conductive metal. In the areas that lie opposite the channels of the transistors, the bottom of the gate structure 6 is coated with a platinum layer 8 in the case of the measuringFET and with a titanium layer 8 ′ in the case of the RefFET.

The gas to be examined or the liquid to be examined reaches the air gaps 4 , 4 ′ via the gas inlets 14 in the gate structure. In other (not shown) examples, the air gaps 4 , 4 'extend to the edge of the gate structure 6 , so that the medium can be fed from the side. Despite the low air gap height of only 1 to 3 µm, gas exchange takes place in less than one second due to diffusion.

In the example shown, the gate structure 6 lies directly on the substrate 12 and the air gaps 3 , 3 'are realized by depressions in the passivation layer 2 ; In other exemplary embodiments, spacers with a corresponding height of 1 to 3 μm are used for this.

To produce the sensor shown, the substrate 12 with the field effect structures and the gate structure 6 are first produced separately from silicon wafers. The gate can also be made of another material, e.g. B. plastic. The wells 11 , 11 'and the S and D regions are doped by a standard microelectronic method such as diffusion, ion implantation or epitaxy. When assembling the two components, the gate structure 6 is positioned upside down with the aid of a swivel arm on the substrate 12 , which is held on a heating plate. The components are positioned laterally by means of a beam splitter optics and a cross table before the swivel arm is folded over and the gate structure 6 comes to lie at the desired position above the substrate 12 . For permanent fixation, the connection using a two-component adhesive - designated 16 in the drawing - has proven to be the most reliable, simplest, and most cost-effective option. The height h of the adhesive space, which is filled by the adhesive, is approximately 20 µm. After fixing the gate structure on the substrate, the contact surfaces (not shown) of the source and drain regions of the field effect structures and the gate electrode are contacted and z. B. connected to a sensor circuit shown in Fig. 6.

In the gas sensor shown in Fig. 1, the channels 3 , 3 'in each of the field effect structures RefFET and MeßFET are protected by a guard electrode 10 against electrical interference, which is applied to the passivation layer 2 in the area of the air gap 4 , 4 ' and in this Level encloses the channel 3 or 3 '. The smallest distance d between the channel areas 3 , 3 'and the guard electrodes 10 is at least 5, preferably 10 µm, so that the guard electrode does not control the channel 3 , 3 '.

Fig. 5 shows a plan view of one of the transistors Reffet or MeßFET. The field effect structure shown has the special feature that the channel 3 does not run in a straight line between drain and source regions, but in a meandering shape. This increases the width-to-length ratio W / L of the transistor with the same overall size and thereby increases the sensor signal. With a channel length L of 0.2 µm and a channel width of 2 mm increased due to the meandering, a ratio W / L of 10,000 results, for example.

The channel region 3 is surrounded as a whole by a guard electrode 10 at a distance of 10 μm. The guard electrode is shown here in plan view as a continuous rectangle, but of course other configurations such as. B. an open ring or several individual electrodes possible.

The SGFET shown is operated with a sensor circuit, the principle of which is shown in FIG. 6. The open-drawn gates of the RefFET and MeßFETs symbolize the hybrid structure of the gas sensor with an air gap. With this circuit, the gas sensor operates in the so-called feedback mode, ie that the drain current I _DS at a constant drain voltage U _DS of z. B. 100 mV is kept constant by readjusting the gate voltage U _G , z. B. to about 100 µA. For this purpose, the drain currents I _{DS of} the RefFET and the measuring FET are each converted into an equivalent voltage in an I / U converter and the two voltages are compared in an integrator. Possibly. the two voltages are scaled differently before the comparison.

The integrator adjusts the gate voltage (usually the same gate voltage is present at both transistors) when one of the two input voltages from the I / U converters changes. It stops regulating when the input voltages are the same again. The size of the readjusted gate voltage ΔU _G is output as a sensor signal.

In the event of a temperature change, the currents I _{DS of} the RefFET and the measuring FET change in the same way, which means that both input voltages of the integrator change proportionally to one another. The integrator is therefore not active, the gate voltage is not readjusted and therefore no sensor signal is output. The sensor signal is therefore not temperature-dependent. When a substance to be detected is acted upon, only the drain current I _{DS of} the measuring FET and thus the associated input voltage at the integrator changes, and this now adjusts the gate voltage until the original current flows again in the channel.

Expressed in the physical quantities of the above formulas, a change in the contact potential by Δφ can be regarded as a shift in the threshold voltage U _T. In order for the drain current to be kept constant, the gate voltage must be readjusted by a value ΔU _G , which means that

I _DS = µCW / L ((U _G + ΔU _G ) - (U _T + Δφ)) U _DS = Const.

results.

If the contact potential of the sensitive layer changes by Δφ when gas is applied, this can be done directly on the basis of the change in the gate voltage

ΔU _G = Δφ

be measured.

A corresponding readjustment of the gate voltage is of course also possible in the case of a gas sensor without a referenceFET, in which case half of the circuit shown is omitted. The gate voltage U _G and thus the sensor signal is simply readjusted whenever the drain current changes in the measuring FET. Even without the readjustment, the sensor can be operated by reading out the changing drain current I _DS .

FIG. 7 shows such a gas sensor according to the invention with only one field effect structure, and configurations of the guard electrode 10 that are slightly modified compared to FIG. 1. In this example too, the gate structure rests on a passivation layer 2 of a substrate 12 , the air gap 4 again being realized by a recess in the passivation layer 2 . A step 20 , on which the guard electrode 10 is arranged, is formed in the side walls 18 of the depression, which run approximately along the boundaries between the channel and the source and drain regions. Like the lateral spacing from the channel in the example in FIG. 1, the step serves to minimize the interference effect of the guard electrode on channel 3 . In the embodiment of FIG. 7b, the guard electrode 10 is processed as the highest layer on the field effect structure and therefore extends up to the sensitive layer 8 and thus also protects it from disruptive charge shifts. Thus can form no electrical interference field and thus leakage currents on the surfaces in the air gap 4 as possible, the guard electrode 10 may be electrically connected to the gate electrode G. This helps to keep all the surfaces surrounding the air gap 4 at the same potential, so that the only change in potential that the work function change on the sensitive layer 8 has on the drain current in the channel. Alternatively, the guard electrode potential can be at a constant potential, e.g. B. mass.

FIG. 8 schematically shows a form of a CCFET which is modified compared to DE 43 33 875 C2. In contrast to DE 43 33 875 C2, the elongated gate electrode 22 is buried in the hatched passivation layer 2 and thus encapsulated, so that it is less susceptible to interference from the presence of gases or from electrical currents and charges. The voltage of the gate electrode protected in this way therefore "floats" less. The gas sensitive layer is mounted on a support above and separated by an air gap through which the gas flows. The sensor shown is equipped with a sensitive layer 8 made of a platinum-containing material, a measuring and a reference FET, and guard electrodes 10 according to the type of FIG. 7b, but the applicants reserve the right to claim the modified structure of the CCFET independently of these features ,

Over the passivation layer is a sensor cover 6 using z. B. support feet 26 arranged and z. B. attached by adhesive (not shown). The support feet are formed as elevations on the sensor cover 6 , the cross section of which is as small as possible in the plane of the passivation layer 2 , so that no dust particles which change the distance are caught between the support feet 26 and the passivation layer 2 during assembly. The underside of the sensor cover is in the area of the air gap 4 of the measuring FET with a sensitive layer, for. B. platinum, and in the area of the air gap 4 'of the reference FET with a non-sensitive layer, for. B. titanium coated. In the example shown, the titanium layer covers the entire underside of the sensor cover, so that a titanium layer runs between the platinum layer 8 'and the sensor cover 6 , which facilitates the adhesion of the platinum layer to the sensor cover. The thickness of the titanium layer is e.g. B. 20 nm, that of the platinum layer 100 nm.

The sensor functions as follows: If gas molecules are adsorbed on the sensitive or non-sensitive layer 8 , 8 ', their work function changes. The change in contact potential at the layer 8 , 8 'acts across the air gap 4 on the buried gate electrode 22 and is transmitted through the electrode 22 to the part 22 a of the electrode lying above the channel region 3 . Since there is no air gap between the part 22 a and the channel 3 , the potential change couples into the channel 3 with a large capacitance, thereby causing relatively large changes in the drain current. The signal-to-noise ratio should therefore be at least as good as that of the HSGFET variant. The calculation and referencing of the sensor signal can, for. B. with the circuit shown in Fig. 6.

The so-called capacitive coupling of the change in contact potential takes place here via a capacitance formed by the layer 8 and the gate electrode 22 , which is connected in series with a capacitance between the gate electrode 22 and a doped well 24 (called CC well) in the substrate. By positioning the buried gate electrode 22 differently within the passivation layer 2 , these two capacitances can be varied and the capacitive voltage divider formed by them can be set differently. Due to the capacitive voltage divider layer 8 , gate electrode 22 and CC well 24 , some signal is lost.

A voltage U _{k can also be} applied to the well 24 doped in the substrate 12 under the air gap 4 . By changing the voltage U _k , the threshold voltage U _{T of} the FET can be shifted, which is equivalent to a shift in the current-voltage characteristic of the FET. This is shown by way of example in FIG. 15, in which I _DS is plotted on the well 24 against the gate voltage U _G at different voltage U _k . While U _{K is} varied between 0 and 0.5 volts in 0.025 V steps, U _T shifts by a total of approx. 15 V. A change in U _k by approx. 25 mV results in a shift in the threshold voltage by 0.5 V. Effect can e.g. B. can be used to compensate for differences in the characteristic curves due to technological manufacturing tolerances between the measuring FET and the reference FET or between different gas sensors on the wafers or batches. In addition, the threshold voltage U _T can be shifted by the voltage U _k on the well 24 to such an extent that the FET blocks and is therefore effectively switched off. In this idle state, it does not consume any current even though the voltages and potentials are still present. In this operating mode, the gas sensor can be switched on and measured immediately.

Setting the properties of the field effect structure via U _k is also advantageous if the gas sensor is heated slightly above the ambient temperature by the drain current I _DS . In this case, the heating can be regulated without power via the gate. This slight heating of the gas sensor - which can of course also be carried out in the embodiments of FIGS. 1 and 7 - is particularly advantageous since it condenses e.g. B. prevents water in the air gap, but uses little energy, since the sensor temperature should only be slightly above the ambient temperature. In addition, the above-mentioned temperature compensation by means of a reference FET works even more effectively the lower the temperature fluctuations are. A relatively strong drain current I _{DS is} set to heat the gas sensor. In order to be able to detect gases and regulate the heating power at the same time, the characteristic curve of the measuring FET can be shifted in the embodiment of FIG. 8 by adjusting the voltage U _k on the well 24 such that the measuring FET despite different source-drain voltages and currents with the aid of the shift in the threshold voltage, it also responds to changes in the contact potential at the sensitive layer. Alternatively, the gas sensor can also be operated in such a way that the regulation of the heating is carried out alternately in a timed manner or the gas sensor is read out.

In Fig. 14a and b show a further construction variant of the gas sensor is shown, in which the gate electrode is made entirely of a platinum-containing material 8, thus in the finished sensor (Fig. 14b) no carrier or gate structure 6 is present. Rather, the platinum layer 8 lies directly on a passivation layer 2 , in which a recess is formed above the channel 3 of the measuring FET to form an air gap 4 . However, a carrier structure 6 is used as an assembly aid for the platinum layer 8 . For this purpose, a support structure 6 , for. B. a silicon substrate, coated with a platinum layer and positioned in the same way as a gate structure 6 on the substrate 12 ( Fig. 14a). Since the platinum layer 8 does not adhere well to the silicon substrate, it falls off the carrier 6 and comes to lie correctly positioned on the passivation layer 2 of the substrate 12 ( FIG. 14b). The carrier 6 is then removed and the platinum layer 8 is contacted. The associated reference transistor works in this case, for. B. without gate. The preferred reading method would then be to measure the change in current from los.

Fig. 11 and 12, the good suitability demonstrate a gas sensor having a platinum layer for the detection of hydrogen. Fig. 11 shows the sensor signal U of a gas-FETs in the manner of Fig 1 with a 100 nm thick platinum layer in function of the applied concentration of hydrogen at room temperature. As you can see, there is in the concentration range between 0 and 2% an approximately logarithmic relationship between the sensor signal and the H ₂ concentration, the sensor being particularly sensitive at low hydrogen concentrations of less than 1%. Fig. 12 shows the sensor signal at different concentrations of hydrogen in function of time, whereby the time response of the sensor is visible. The response time of the sensor is in the range of a few seconds, the decay time is approx. 30 seconds.

Fig. 13 shows the cross-sensitivities shows the platinum layer with respect to the gases NO _2, NH _3, CO, SO _2, Cl ₂ and O _3, as well as the sensitivity to H ₂ at room temperature (gray bars) and at 130 ° C (black). The diagram shows that a platinum-containing layer as a sensitive layer for a gas sensor can cover two gases according to the principle of work function measurement: at room temperature it is very sensitive and selective to hydrogen, at a temperature of 130 ° it shows good sensitivity and selectivity to ozone.

The diagrams shown in FIGS. 9 and 10 demonstrate the good results of a sensor of the type of the example of FIG. 1 or 7 with a sensitive layer 8 made of platinum and possibly a non-sensitive layer made of titanium with regard to the compensation of moisture and temperature influences. FIG. 9 shows a measurement curve 30 of the drain current I _DS in the case of a gas sensor in the manner of the example in FIG. 7, that is to say without a reference FET, but with a guard ring, for humidities between alternately 0% and between 10 and 90%. In comparison to the prior art (see FIG. 3), the baseline is considerably more stable; only a rash can be seen, which is caused by the known effect that water molecules accumulate on the sensitive layer and / or the passivation layer and thereby cause a sensor signal. Curve 32 , on the other hand, shows the estimated (not measured) sensor signal of a gas sensor according to the type of the example from FIG. 1, which is equipped with guard electrodes and also has a referenceFET with a non-sensitive layer, which has approximately the same sensitivity to moisture as the sensitive one Has layer of MeßFETs. With this referencing, the moisture signal can be suppressed almost to the level of the noise.

Fig. 10 shows the temperature profile of the drain currents (channel currents) in the channel of the measurement FET according to the invention, measurement I _D , and the reference FET, Ref I _D , and the difference ΔI _D , these currents, with a gate voltage of U _G = -1 V and one Drain voltage of U _D = -200 mV, measured with a gas sensor similar to that of Fig. 1. It can be seen from the figure that the drain currents in the measuring and reference FET have approximately the same temperature dependency, so that the calculation of the two quantities Temperature dependence can be reduced by a factor of 20.

Claims (25)

1. Gas sensor, which works on the principle of work function measurement, for the detection of hydrogen, hydrogen sulfide and / or other gases which are chemically similar to hydrogen, the gas sensor comprising a sensitive layer ( 8 ) and by the adsorption of molecules of the detected Gas on the sensitive layer, the work function of the sensitive layer ( 8 ) is variable, characterized in that the sensitive layer ( 8 ) has platinum. 2. Gas sensor according to claim 1, wherein the sensitive layer (8) further comprises titanium, wherein arranged in particular a titanium-containing intermediate layer (8) between a platinum-containing layer (8) and a carrier (6). 3. Gas sensor according to claim 1 or 2, comprising:
a first field effect structure with a channel ( 3 ) between a source and a drain region (MessFET); and
a first gate electrode (G) comprising the sensitive layer ( 8 ), with a change in the work function of the sensitive layer ( 8 ), e.g. B. by the adsorption of hydrogen molecules on the sensitive layer, the drain current in the channel ( 3 ) of the MeßFETs can be influenced. 4. Gas sensor according to one of the preceding claims, comprising:
a second field effect structure with a channel ( 3 ') between a source and a drain region (reference FET) for referencing the temperature response of the drain current (I _DS ) of the measuring FET. 5. Gas sensor according to claim 4 with a second gate electrode which comprises a titanium-containing layer ( 8 '), wherein the drain current in the channel ( 3 ') of the reference FET can be influenced by changing the work function of the titanium-containing layer. 6. Gas sensor according to claim 2 and 5, wherein the titanium-containing layer ( 8 ) and the platinum-containing layer ( 8 ') are applied to a common gate structure ( 6 ) through an air gap ( 4 ) from the channel regions ( 3 , 3 ') of the measuring and the reference FET is separated. 7. Gas sensor according to one of claims 3-5, wherein the gate electrode ( 6 ) consists of a platinum-containing material. 8. Gas sensor according to one of the preceding claims, which has a sensor circuit with which a difference between the drain current (I _DS ) of the measuring FET and the drain current (I _DS ) of the reference FET, or another linear combination of these currents, by a readjustment of the potential is constantly stable at the gate electrode (U _gate ) of the measuring FET. 9. Gas sensor according to one of the preceding claims, wherein the channel ( 3 ) of the measuring FET is surrounded by a guard electrode ( 10 ) for protection against electrical interference. 10. Gas sensor according to claim 4 and 9, wherein the channel ( 3 ') of the referenceFET is also surrounded by a guard electrode ( 10 ). 11. Gas sensor according to one of claims 9 or 10, wherein the gate electrode (G) is spaced by an air gap ( 4 ) from the field effect structure (MessFET, RefFET) and the guard electrode ( 10 ) is arranged on a passivation layer applied to the field effect structure ( 2 ) , e.g. B. is evaporated or sputtered. 12. Gas sensor according to claim 11, wherein the guard electrode ( 10 ) is arranged on a step ( 20 ) embedded in the passivation layer ( 2 ). 13. Gas sensor according to one of claims 9-12, in which the guard electrode ( 10 ) is stable at a constant potential (U _K ), or in which the guard electrode ( 10 ) is at the same potential as the gate electrode (G). 14. Gas sensor according to one of the preceding claims, which as a suspended Gate FET (SGFET), especially as a Hybrid Suspended Gate FET (HSGFET) is trained. 15. Gas sensor according to claim 14, wherein the channels ( 3 , 3 ') of the measuring and possibly the reference FET are meandering. 16. Gas sensor according to one of claims 1-13, which is controlled as capacitive FET (CCFET) is formed. 17. Gas sensor according to claim 1 or 2, which is designed as Kelvin insode. 18. Gas sensor according to one of the preceding claims, wherein the platinum-containing sensitive layer ( 8 ) and possibly the titanium-containing intermediate layer ( 8 ') are formed as thin layers. 19. Gas sensor according to one of the preceding claims for use in a mobile application, such as B. a motor vehicle, or in a tunnel. 20. A method for the detection of hydrogen, hydrogen sulfide and / or other gases which are chemically similar to hydrogen, in which the change in the work function caused by the adsorption of the gas to be detected on a sensitive layer (Δφ.e) is measured, characterized that the sensitive layer ( 8 ) has platinum. 21. The method according to claim 20, which with a gas sensor according to one of the Claims 1-20 is executed. 22. A method for producing a hybrid suspended gate FET (HSGFET) according to any one of claims 1-16, comprising the following steps: a) producing a field effect structure (measuring FET) with a channel ( 3 ) between a source and a drain region; b) applying a platinum-containing layer ( 8 ) to a carrier ( 6 ); c) placing the carrier ( 6 ) on the field effect structure (MeßFET), so that the platinum layer ( 8 ) faces the channel ( 3 ) and there is an air gap ( 4 ) between the two. 23. The method according to claim 22, wherein the carrier ( 6 ) is finally removed and the platinum-containing layer ( 8 ) remains on the field effect structure (MeßFET). 24. The method according to claim 22, wherein a titanium-containing layer ( 8 ') is applied before the application of the platinum-containing layer ( 8 ) to the carrier ( 6 ). 25. The method according to any one of claims 22-24, wherein the platinum and possibly the titanium-containing layer ( 8 , 8 ') by electrochemical deposition, sputtering, reactive sputtering, vapor deposition, spin coating, sublimation, epitaxy or spraying onto the carrier ( 6 ) is applied.