WO2012095117A1

WO2012095117A1 - Micromechanical pressure sensor and method for producing same

Info

Publication number: WO2012095117A1
Application number: PCT/EP2011/000096
Authority: WO
Inventors: Joachim Knoch; Klaus KALLIS
Original assignee: Technische Universität Dortmund
Priority date: 2011-01-12
Filing date: 2011-01-12
Publication date: 2012-07-19
Also published as: EP2663849A1

Abstract

The invention relates to a micromechanical pressure sensor (1) and to a method for producing same, wherein the pressure sensor comprises a membrane (3) which is formed in a first semiconductor layer (2), and a sacrificial layer (4) to which the semiconductor layer (2) is applied. In the membrane (3), a drain region (7) and a source region (8) are present, which face each other and which are separated from each other by a conducting channel (12), wherein the sacrificial layer (4) forms an insulation layer (4) beneath which an additional layer (5) with a gate electrode (6) is provided, which is located opposite the conducting channel (12). A cavity (9) is formed in the insulation layer (4) between the additional layer (5) and the membrane (3), into which cavity the membrane can move when pressure is applied to the pressure sensor. The invention further relates to a method for measuring pressure.

Description

Micromechanical pressure sensor and method for its production

The present invention relates to a micromechanical pressure sensor with a membrane formed in a first semiconductor layer and a sacrificial layer on which the semiconductor layer is applied. Furthermore, the invention relates to a method for producing such a pressure sensor and a use thereof for high-resolution pressure measurement.

Micromechanical pressure sensors, so-called pressure cans, have been used in various technological fields for several years. For example, pressure cans in the automotive sector are used to monitor tire pressure. In the medical field, pressure sensors are used to monitor blood pressure. Three methods are used for reading today's micromechanical pressure cans. In a first method, the sensor is read out via the piezo-resistive change of the resistance of a silicon membrane. In a piezo-resistive sensor, the change of the electronic

Transport properties of a material, in particular silicon used to make a mechanical strain as an electrical signal measurable. By mechanical stress essentially a change in resistance is achieved, which can be measured. The reason for this is microscopic nature. Either the band structure can change and / or the frequency of scattering of the electrons in the strained material, which manifests itself as a change in resistance. With the resistance value or the determined

Resistance change can then be made a statement about the pressure.

CONFIRMATION COPY However, the piezo-resistive readout provides only a change in resistance of at most 30%, which is why so-called Wheatstone bridges must be used to ensure sufficient sensitivity of the sensor, as in the publications "A novel MEMS pressure sensor with MOSFET on chip", IEEE Sensors 2008 , 1564 (2008), Z.-H, Zhang, Y.-H. Zhang, LT. Liu and T.-L. Ren, and "A Silicon Piezoresistive Pressure Sensor", Proc. Is IEEE boarding. Workshop Electron Design, Test Appl. (2002), R. Singh, LL Ngo, HS Seng, FNC Mok. The Wheatstone bridge responds to changes in resistance with a change in voltage.

In a second method, the change in capacitance of the pressure sensor between the membrane and the underlying substrate is used for pressure determination. The capacitive reading is e.g. on the change in current of an integrated into the silicon membrane of the pressure sensor MOSFETs (Metal Oxide Semiconductor Field Effect Transistor) in the on state. The integration of a MOSFET in the silicon membrane of the pressure sensor is for example from the

Publications "A novel MOSFET pressure sensor", Smart Sensors, actuators and MEMS, vol. 5836, 363 (2005), J. A. Segovia, M. Femandez-Bolanos and J. M. Ouero, and "A novel MOSFET pressure microsensor", Internat. Conf. on Solid State and Integrated Circuit Technology 2006, 614 (2006), Y.-H. Zhang, L-T. Liu, Z.-H. Zhang, Z.-M. Tan, H.-W. Lin and T.-L. Ren known.

With this readout method, however, a maximum of 1 / d _0X dependence of the

Readout signal can be reached from the pressure, where d _{ox is} the decisive factor for the value of the capacitance oxide thickness. In addition, the operation of the MOSFET in the

Anzustand a comparatively high power loss.

A third possibility is the reading about the changed resonant frequency of a resonant circuit due to the capacitance change of the pressure cell, see eg the publication of H. Dudaicevs, M. Kandier, Y. Manoli, W. Mokwa and E. Spiegel in Sensors Actuators A, 43, 157 (1994). In this case, however, the pressure cell must have comparatively large lateral dimensions in order to realize the necessary capacity in the pF range. A further reduction and integration of pressure cans can not be achieved with this readout technology thus. The membranes of micromechanical pressure sensors are manufactured today from polycrystalline silicon. In this case, a sacrificial layer, for example made of silicon nitride, of suitable lateral and vertical dimensions is applied to an oxidized substrate (oxide) and subsequently polysilicon is grown over this structure. If the sacrificial layer is now etched away, a polysilicon membrane remains, which is closed by further growth of the polysilicon, so that a cavity arises in between the polysilicon and the oxidized substrate. Decisive disadvantages of this technique are a possible curvature of the polysilicon membrane and restrictions on the size scale of the

Membrane due to the variation in mechanical properties of polysilicon, which in turn is related to the varying morphology. In addition, a polysilicon membrane is mechanically much less stable than a crystalline silicon material membrane.

It is known from US Pat. Nos. US 7235456 B2 and US Pat. No. 7512170 B2, by etching holes in silicon wafers and then annealing them in a pure

Hydrogen atmosphere is a cavity in the silicon forming the substrate, i. in so-called bulk silicon, which is then bounded on all sides by the silicon. The use of the first two readout methods is not possible with such a sensor, since the membrane is not electrically insulated from the silicon lying opposite it, with which it is integrally formed. The application of a capacitive readout then leads to unacceptable leakage currents. Another disadvantage of this technology is that the dimensions of the cavity, in particular its height, depends on the thickness of the membrane and therefore is freely adjustable only to a small extent. A to a special application

adapted pressure sensor can be made with this lesson difficult and only with considerable effort.

It is an object of the present invention to provide an improved micromechanical pressure sensor which has a much higher measurement sensitivity and simultaneously increased measuring range, so that in particular the low pressure range for pressure measurements is accessible, the dimensions of the cavity and the membrane are independently adjustable, so that a reduction of the dimensions of the pressure sensor and a simple adaptation the dimensions of a particular use of the sensor is possible. Furthermore, it is an object of the invention to provide a method for producing such a pressure sensor and a method for pressure measurement by means of such a pressure sensor.

This object is solved by the features of claims 1, 10 and 20.

Preferred developments of the pressure sensor and the manufacturing and

Measurement methods are specified in the respective subclaims.

According to the invention, a micromechanical pressure sensor with a membrane formed in a first semiconductor layer and a sacrificial layer on which the

Semiconductor layer is proposed, proposed in which in the membrane, a drain region and a source region is present, which are opposite to each other and which are separated by a duct, wherein the sacrificial layer is a

Insulation layer forms, below which another layer is located with a gate electrode which is opposite to the duct, and wherein between the further layer and the membrane, a cavity is formed in the insulating layer.

Furthermore, a method for producing a micromechanical

Pressure sensor of the type mentioned proposed in the first one

Layer structure consisting of a first semiconductor layer, a sacrificial layer and a further layer is produced, wherein the sacrificial layer forms an insulating layer and is located between the first semiconductor layer and the further layer, then depressions are introduced into the first semiconductor layer such that a recess to the sides respectively is bounded by a web and extends to the insulating layer, then a Unterätzen a cavity in the insulating layer is such that the cavity is bounded by the first semiconductor layer and the further layer, wherein the first semiconductor layer for forming a in the

Substantially uniformly thick membrane, which limits the cavity to one side, is then annealed, and wherein a gate electrode is formed in the further layer, and a drain region and a source region are created in the membrane, which are opposite and which are separated by a duct which faces the gate electrode. A pressure sensor produced in this way forms a field-effect transistor (FET) and has an exponentially read-out signal dependent on the pressure. As a result, it has a much higher sensitivity than the state of the art allows.

In particular, due to this high sensitivity, the measurable range can be significantly increased, and especially the low pressure range becomes accessible. The dimensions of the cavity and the membrane can be determined by the thickness of the

Isolation layer and the width of the recesses or the webs are set independently, so that a reduction in the dimensions of the pressure sensor is possible. The use of an insulating layer makes it possible to produce a gate electrode isolated from the membrane underneath the membrane.

Preferably, the first semiconductor layer consists of silicon, in particular monocrystalline silicon. Silicon has the advantage that it is particularly easy to process. In particular, structures can be etched particularly simply and precisely in this structure, and limited regions can be easily doped. The use of monocrystalline silicon has the advantage that this results in a higher mechanical strength of the membrane compared to polycrystalline silicon (ie

Polysilicon), since there are no mechanical variations due to the varying morphology.

The further layer may consist of a second semiconductor layer, in particular of silicon. In this case, single- or polycrystalline silicon can be used. The formation of a gate electrode in the further layer can then take place by implanting ions into the second silicon layer through the membrane. This ion-implanted region becomes conductive, leaving a

Voltage can be applied to him, which generates an electric field acting substantially perpendicular to the duct. Alternatively, the gate electrode can also be produced by using a metal as the further layer. The layer structure can then be produced by oxidizing a silicon substrate, then applying a sufficiently thick, in particular several 10 μm thick, metal layer for sufficient mechanical stability, and then grinding away the substrate until the substrate is removed

resulting silicon thickness has reached the thickness required as a membrane. It should be noted, however, that the metal to be used the high Temperatures must withstand tempering, so that only high-melting metals such as tungsten can be used.

The insulating layer may consist of silicon oxide (SiO ₂ ). On the one hand, silicon oxide is a good insulator, which can easily be made of silicon, and on the other hand can be well etched.

The recesses to be introduced into the first semiconductor layer may preferably form trenches which run parallel to one another. These can be particularly easily and precisely determined by anisotropic etching, i. by vertical etching, in the first

Semiconductor layer can be generated. Alternatively, the recesses may be formed by holes, each forming a row. However, trenches have the advantage that with their width and spacing, the thickness of the desired membrane can be determined more easily.

Preferably, the drain region and the source region are the same

formed highly doped regions in the first semiconductor layer, wherein the thickness of the membrane, its distance from the gate electrode and the width and spacing of the doped regions are chosen such or chosen so that the capacitance between the gate electrode and the conduit equal to or is less than the capacitance between the drain region and the source region. The doped regions may therefore both be n-doped or both p-doped. In this way, a pressure sensor is formed as a field effect transistor (FET) with short channel behavior, which can be called short channel on nothing FET (SCHONFET) due to the lying above the cavity line channel.

The term "short channel" and "short channel effect" refers to the

geometric relationships between the distance d _{ox of} the membrane to the gate electrode (oxide layer thickness), the thickness dsi of the membrane (channel layer thickness) and the length L of the duct (channel length). A device exhibits long-channel behavior as long as the gate capacitance CG is significantly larger than the drain-source capacitance Cos. The gate capacitance CG for the pressure sensor _according to the invention corresponds to the product of relative dielectric constant ε _o and gate area through the distance d _{o of} the gate electrode to the membrane. The drain-source capacitance Cos can approximately from the product of relative dielectric constant ε _& , the width W of the doped regions and the membrane thickness dsi, divided by the channel length L are calculated. From the capacity relation follows in a first approximation:

C _G = (ε _οχ * (W ^* L) / dox) »C _DS = (£ _Si (W ^* d _Si ) / L) to avoid short channel effects. Here, Zo * is the relative dielectric constant of the insulating layer that is air between the gate and the membrane, so that

applies. si is the relative dielectric constant of the first semiconductor layer, in particular the silicon layer, W is the width of the FET, L is its channel length and d _{ox is} the distance between the membrane and the gate, ie the air gap. In order for enough short channel effects to occur, which are intentionally generated here, the two capacitances should be approximately equal. For the channel length follows as a guideline: L ² = Esi / eox ^* dsi ^* d _ox . The pressure sensor according to the invention can thus be desinged so that the built-SCHONFET short channel behavior, ie

Short channel effects shows.

In an alternative embodiment, the drain region and the source region may be formed by oppositely highly doped regions in the first semiconductor layer, wherein the thickness of the membrane, its distance from the gate electrode and the width and spacing of the doped regions are chosen such or ., the capacitance between the gate electrode and the conduction channel is much larger than the capacitance between the drain region and the source region. The one area is thus n-doped while the other area is p-doped. In this way, the pressure sensor is implemented as a field effect transistor (FET) without short channel behavior, in which the electrical charge carriers in the semiconductor structure of the

Membrane from conduction band to valence band or valence band to

Tunnels the conduction band so that the pressure sensor band-to-band tunnel FET can be called. The exponential dependence of the current on pressure is achieved by band-to-band tunneling, which depends on the pressure, i. depends on the distance of the membrane from the gate electrode.

Preferably, the doping of the drain and the source region is above an amount of 1 * 10 ¹⁹ charge carriers per cm ³ . A highly doped area allows the Depletion zone in the contacts relative to the bends of the

Neglecting conduction band / valence band in the channel, which simplifies the design of the pressure sensor. The ions which make the charge carriers available may be, for example, by diffusion or implantation into the respective ones

Silicon layer are introduced.

In another alternative embodiment, the drain region and the source region are formed by metals. In this way, a pressure sensor is obtained as a Schottky barrier MOSFET (SB MOSFET). At the metal-semiconductor junction, in particular at the source-channel junction, a potential barrier is formed. This barrier becomes thinner when pressure is applied to the sensor, resulting in an exponential dependence of the current on the pressure. Of the

Tunnel current through the barrier depends exponentially on its thickness. in the

Unlike the SCHONFET, the SB MOSFET does not have short channel effects

to provide an exponential sensor signal.

In an advantageous embodiment of the pressure sensor according to the invention according to one of the above-described embodiments, the drain region and the source region overlap the gate electrode respectively. This causes no areas to exist that are not properly affected by the gate.

Preferably, in the pressure sensor according to the invention a thin

Membrane used, which may have a thickness in the range between 5 to 50nm. This is particularly advantageous in the SB MOSFET or in the tunnel FET, since otherwise the tunnel currents may become very small.

If the further layer consists of silicon, ions can be introduced into the second silicon layer to form the gate electrode. The introduction can take place by implantation, wherein the ions are implanted through the membrane into the second silicon layer. This process step fits well into the process order after the annealing and before the formation of the drain region and the source region. This has the advantage that the layer structure is always only from above, i. must be handled from one direction. Turning over the

Layer structure is not necessary. The layer structure can preferably be produced by two different methods. According to a first method, the insulating layer is formed of silicon oxide, which can be produced by implantation of oxygen ions in silicon and subsequent annealing. This manufacturing method is known and is called SIMOX (Separation by Implantation of Oxygen). In this case, oxygen is implanted in a silicon wafer. In a subsequent annealing, the wafer oxidizes from the inside, in contrast to the conventional oxidation, which takes place only on the surface of the wafer, because there is the oxygen. By implanting oxygen, the oxygen is located below the wafer surface where it creates a buried oxide layer.

In an alternative embodiment variant, the layer structure can be produced from a bonded SOI wafer (silicone-on-insulator). These too

Manufacturing method is known. In this case, a wafer is oxidized from a silicon monocrystal and pressed a second wafer upside down on the oxidized wafer. This results in a silicon-oxide-silicon layer structure. This layer structure is then tempered and then a wafer is ground. This results in a thin, single-crystal silicon layer on an insulator (SiO ₂ ), which is located on a silicon single-crystal substrate. Such a layer structure is called SOI wafer.

The introduction of the recesses can by anisotropic etching of the first

Semiconductor layer done. Furthermore, the cavity can be made by isotropic etching. This can be done wet-chemically or by reactive ion etching. Anisotropic etching etches in one direction only. That is, the pits are made only by vertical etching. In the case of isotropic etching, etching is the same in all directions. This means that the cavity is etched both vertically and horizontally.

Preferably, the depressions are introduced into the first semiconductor layer in such a way that the width of the depressions is smaller than the width of the webs. This ensures that the webs during annealing to widen their height so that their sides to a closed membrane touch. The membrane thickness can be determined by the width of the depressions and the width of the webs.

The proposed pressure sensor can be used according to the invention for pressure measurement. For this purpose, a method for reading the pressure sensor is proposed in which applied between the drain region and the source region, a voltage and the current through the drain-source region at a

Voltage between the gate electrode and the source region is measured below the threshold voltage, wherein according to an exponential

Depending on the measured current from the pressure exerted on the pressure sensor ambient pressure, this ambient pressure is determined.

The field effect transistor integrated in the pressure sensor is consequently operated below the threshold voltage in the off state due to a voltage between the gate electrode and the source region, since the threshold voltage is the voltage at which the component changes over to the on state. This has the particular advantage that the power loss is minimized. Because the missing gate voltage leads to a very small current, the so-called outflow, which also decreases exponentially depending on the pressure exerted on the membrane. This allows the use of such pressure sensors in energy self-sufficient sensor networks. Together with the possibility of reducing the dimensions of the pressure sensor, sensors for the medical sector, for example, become possible in this way.

Further advantages and features of the pressure sensor according to the invention, its production method and the method for reading a pressure sensor will be explained in more detail with reference to embodiments and the accompanying figures. Show it:

Fig. 1a to 1h illustrating the manufacturing process of a

inventive pressure sensor

Fig. 2a: conduction band in a MOSFET with short channel effects

FIG. 2b: Drain current for a component with increasing gate oxide thickness. FIG

FIG. 3a: Schematic diagram of the buried gate pressure sensor. FIG FIG. 3b: conduction band course through the component integrated in the silicon membrane. FIG

Fig. 3c: drain current as a function of pressure

Fig. 4a: pressure sensor with integrated Schottky barrier MOSFET

Fig. 4b: Illustration of the exponential change of the current at a

Change in gate oxide thickness

Fig. 4c: exemplary experimental curves showing the strong dependence of

Prove transistor characteristics of the gate oxide thickness

Fig. 5a: top: pressure sensor with integrated tunnel FET

bottom: the band path for two different gate oxide thicknesses

Fig. 5b: experimental characteristics of a tunnel FET for two

different gate oxide thicknesses.

FIGS. 1a to 1h show the production steps for the production of a micromechanical pressure sensor 1 according to the invention with a membrane 3 formed in a first silicon layer 2 and an insulation layer 4 on which the silicon layer 2 is applied. Below the insulating layer 4 is a second silicon layer 5, in which ions are introduced to form a gate electrode 6. In the membrane 3, two separate areas 7, 8 separated from one another by a line channel 12 of the channel length L are formed, in which ions are introduced for forming a respective drain region 7 and a source region 8 overlapping the gate electrode 6. Between the second silicon layer 5 and the

Membrane 3, a cavity 9 is formed in the insulating layer 4, in which the membrane 3 can move into or bend in at an environmental pressure exerted on the pressure sensor 1. Under negative pressure, it moves out of the cavity.

The insulating layer 4 consists of silicon oxide (S1O2) and is also referred to below as the oxide layer. Being located between the first and second silicon layers 2, 5, it constitutes a so-called "buried layer." The first silicon layer 2 consists of monocrystalline silicon. The drain region 7 and the source region 8 are each with negatively charged ions n ⁺⁺ heavily doped. These regions 7, 8, the gate electrode 6 overlap each other by at least a quarter of its width. In a first production step, not shown in FIGS. 1a to h, the layer structure with the buried oxide layer 4 is produced. This can

for example, by implantation of oxygen ions in a silicon layer and subsequent annealing of this layer, whereby the oxygen ions combine with the silicon to form a layer of silicon oxide 4. Alternatively, bonded SOI (Silicon on Insulator) wafers can be used to seal the

Layer structure to realize. The thickness of the buried oxide layer 4 essentially determines the height of the cavity 3. The height of the cavity 3 can be adjusted completely freely, which is particularly crucial for the readout method according to the invention.

As shown in FIG. 1 a, trenches 10 are anisotropically etched out into the first silicon layer 2 in such a way that a trench 10 is delimited in each case by a web 11 and extends to the insulating layer 4. Subsequently, a cavity 9 is introduced into the buried oxide layer 4 with an isotropic etching step, FIG. 1 b. This can be done either wet-chemically or by means of reactive ion etching. In a subsequent annealing step in pure hydrogen (H2), the webs 11 increasingly round, see FIGS. 1c and 1d. Finally, they form a closed membrane 3 of thickness dsi, as can be seen in FIG. 1e. The

Dimension of the webs 11 and the trenches 10 are thus to be chosen so that the rounded webs 11 can touch by annealing to form the membrane.

In a subsequent method step, ions are implanted through the membrane 3 into the second silicon layer 5 below the buried oxide layer 4, which ions form a gate electrode 6 after activation, see FIG. 1f.

The last step involves defining a source region 7 and a drain region 8. This is done by a mask 12 covering those regions of the top silicon layer 2 into which no ions are to be implanted, see FIG. 1g. FIG. 1 g shows the subsequent ion implantation of similar ions n ⁺⁺ into the regions 7, 8 not covered by the mask 12, which form the source region 7 and the drain region 8 after the ion implantation. The finished pressure sensor 1 is shown in Fig. 1 h. The source region 7, drain region 8 and the gate electrode 6 form a transistor which is an integral part of the pressure sensor 1, wherein the

Current path of the transistor in the membrane 3 and in the conduction band 12 is located.

The height of the cavity 9 is determined essentially by the thickness of the oxide layer 4 and not by the silicon webs 11. As a result, the dimensions of the pressure sensor 1 can be flexibly optimized to the desired application. If e.g. a pressure sensor with a thin membrane and large cavity height is desired, this can be done with the known method of silicon forming

Do not realize tempering of a silicon substrate, since to do so deep webs would have to be etched, which must be either thin and close to each other or relatively thick. In both cases, there is a relationship between the cavity height and the lateral dimension of the pressure sensor. By generating the cavity in the oxide layer 4, membrane thickness dsi and cavity height can be adjusted independently of each other.

The membrane 3 is covered by a thermally grown silica 4, i. by the buried oxide, isolated from the substrate 5. The excellent insulating properties of silicon oxide make the SCHONFET technology presented here possible in the first place. As a result of the insulation between the substrate and the first silicon layer 2, which becomes the silicon membrane 3 in the region of the cavity 9 through the manufacturing process, it is possible to integrate an evaluation electronics in CMOS technology directly on a chip.

In order to realize the readout technique according to the invention, the length L of the channel 12 between the heavily doped source / drain regions 7, 8 is set such that the source channel and channel drain pn junctions strongly overlap. As shown in Figure 2 (a), these pn junctions have a spatial extent of the order L = (esi / ε _οχ d _S id _ox ). They therefore depend on the

Membrane thickness dsi and the distance between the membrane 3 and the gate electrode 6, hereinafter called gate oxide thickness d _ox , from. Due to the overlap, so-called short-channel effects occur, which lead to a significant lowering of the current-determining potential barrier in the channel 12. This is shown schematically in FIG. 2 (a). This in turn requires a strongly increasing

Drain current I _d between the source region 7 and the drain region 8 in the off state of the transistor, ie at a voltage between the gate electrode 6 and the source region 7 below the threshold voltage, the current in the first Approximation increases with decreasing ratio between the channel length L and the root of the thickness d _{ox of} the gate oxide, as shown in Figure 2 (b). This means that with a constant channel length L there is a strong dependence of the current ld on the gate oxide thickness d _o , ie on the height of the cavity 9, which is present without pressure through the buried oxide layer. By the above described

Manufacturing method can be both the diaphragm thickness dsi, the height d _{ox of} the cavity 9 (equal oxide thickness) and the channel length L set exactly defined and thus achieve the desired short channel effects. This will also be one

Scaling down the pressure sensors 1 allows.

Fig. 3 (a) schematically shows the device structure of the pressure sensor 1, and the variation of the outflow of the pressure sensor 1 is illustrated in Fig. 3 (c). If a pressure p is exerted on the membrane 3, then the effective oxide thickness d _ox , ie the air gap between the membrane 3 and the gate electrode 6, is reduced, as a result of which the capacitance CG between gate electrode 6 and conduction channel 12

is enlarged. This leads to a reduction of the short-channel effects and to an improved gate control, which in turn causes the potential barrier determining the current flow between source and drain to be raised, see FIG. 3 (b), without the gate-source voltage V _gs changing, ie also at

turned off transistor V _gs <V _th , where V _{th is} the threshold voltage of

Transistor is. This will drain current through the transistor, even in the

Off state, exponentially lowered, see Figure 3 (c).

The change in electricity is thus a consequence of the changing

geometric relationship between the gate capacitance CG and the drain / source capacitance Cos- With an increase in the gate capacitance CG, the influence of source and drain on the potential (band) - course in the line channel 12 is better "shielded" and accordingly follows the potential barrier rather the

Gate voltage V _gs as the drain voltage V _ds -

FIG. 3 (b) shows the conduction band profile in the silicon diaphragm 3 for the pressures p = 0 (top) and p * 0 (bottom). By way of example, FIG. 3 (c) shows the drain current as a function of the pressure for a pressure sensor 1 according to the invention. The strong dependence of the current on the pressure with a circumference of almost 6 is clearly evident Orders of magnitude. By suitable choice of the thickness of the oxide layer 4, the thickness of the first silicon layer 2 (SOI layer) and the width and spacing of the trenches 10 to be etched into the first silicon layer, see FIG. 1 (a), as already mentioned, the membrane thickness dsi, the lateral dimension of the pressure sensor 1 and the channel length L of the MOSFET integrated in the pressure sensor 1 are optimized for the respective field of application.

A MOSFET of the prior art without short channel effects must

On the other hand, it is mandatory to operate in the on state, ie V _gs > V _th , so that a changing gate capacitance CG has any influence on the current at all. The current is then proportional to the gate capacitance CG, resulting in a linearly changing current. However, such a MOSFET can not be operated in the off state, because the design does not allow current variation in the off state.

If the channel is suitably doped, a depletion zone capacitance is created parallel to the gate capacitance C _G. Now, when pressure is applied to the can, the ratio between gate and depletion zone capacitance changes, resulting in an exponentially changing outflow. On the other hand, in the SCHONFET, as described above, the relationship between gate and drain capacitance CG, C _D S is relevant.

An exponentially pressure-dependent sensor signal can also be achieved with a Schottky barrier MOSFET integrated in the pressure sensor 1, see FIG. 4 (a) and a band-to-band tunnel FET, see FIG. 5 (a). In the Schottky barrier MOSFET, the source region 7 and the drain region 8 are formed by metals, whereas the source and drain are highly doped in the band-to-band tunnel FET having different ion types n ⁺⁺ and p ⁺⁺ .

The switching behavior of a Schottky barrier MOSFET depends strongly on the thickness of the Schottky barrier, which in turn depends on the thickness of the gate oxide d _ox , as shown in Figures 4 (a) - (c). If pressure is exerted on the pressure sensor 1, then the thickness of the Schottky barrier, Figure 4 (b), which changes to a

exponentially dependent on the pressure current, Figure 4 (c). Even in a band-to-band tunnel FET, the current depends exponentially on the band-to-band tunnel barrier and thus on the gate oxide thickness d _ox . As shown schematically in Figure 5 (a), a thinner gate oxide results in a steeper one

p-n junction at the source-channel junction, resulting in a thinner barrier and thus exponentially increased current. Figure 5 (a) below shows that a reduction in oxide thickness results in an exponential increase in current.

Preferably, very thin membranes are used in the band-to-band tunneling MOSFET to get a significant tunneling current. Also in the case of a Schottky barrier MOSFETs thin membranes should preferably be used. Furthermore, although the sensitivity of a pressure sensor based on tunnels (Schottky barrier MOSFET or band-to-band tunnel FET) will be greater than achievable with existing piezoresistive pressure sensors, it will be worse compared to the SCHON-FET technology presented earlier fail.

The presented invention enables the realization of micromechanical pressure sensors with an exponentially dependent on the pressure readout signal, so that a much higher sensitivity of the pressure cans is achieved than is possible with the current state of the art. In particular, due to this high sensitivity, the measurable pressure range of a micromechanical pressure sensor can be significantly increased, which makes the low pressure range accessible in particular. Furthermore, due to the much lower power consumption in contrast to existing pressure sensors, the invention allows use in energy-autonomous sensor networks. In addition, the technique presented here is completely CMOS-compatible and can be produced in planar technology.

Claims

claims

1. Micromechanical pressure sensor (1) with a in a first semiconductor layer (2) formed membrane (3) and a sacrificial layer (4) on which the

Semiconductor layer (2) is applied, characterized in that in the membrane (3) a drain region (7) and a source region (8) is present, which are opposite and separated by a duct (12), and in that the sacrificial layer (4) forms an insulating layer (4) below which there is a further layer (5) with a gate electrode (6) facing the line channel (12), between the further layer (5) and the membrane (3) a cavity (9) is formed in the insulating layer (4).

2. Micromechanical pressure sensor (1) according to claim 1, characterized

characterized in that the first semiconductor layer (2) consists of monocrystalline silicon.

3. Micromechanical pressure sensor (1) according to claim 1 or 2, characterized

characterized in that the further layer (5) consists of silicon or metal.

4. Micromechanical pressure sensor (1) according to one of the preceding claims, characterized in that the insulating layer (4) consists of silicon oxide.

5. Micromechanical pressure sensor (1) according to one of the preceding claims, characterized in that the drain region (7) and the source region (8) are formed by similarly highly doped regions in the first semiconductor layer (2), wherein the thickness ( dsi) of the membrane (3), its distance (d _ox ) to the gate electrode (6) and the width (W) and distance (L) of the doped regions (7, 8) are selected such that the capacitance between the gate electrode (6) and the conduction channel (12) is equal to or smaller than the capacitance between the drain region (7) and the source region (8) ,

6. Micromechanical pressure sensor (1) according to one of the preceding claims 1 to 4, characterized in that the drain region (7) and the source region (8) by oppositely highly doped regions in the first

Semiconductor layer (2) are formed, wherein the thickness (dsi) of the membrane (3), their distance (d _ox ) to the gate electrode (6) and the width (W) and distance (L) of the doped regions (7, 8 ) are selected such that the capacitance between the gate electrode (6) and the conduction channel (12) is much larger than the capacitance between the drain region (7) and the source region (8).

7. Micromechanical pressure sensor (1) according to one of the preceding claims 1 to 4, characterized in that the drain region (7) and the source region (8) are formed by metals.

8. Micromechanical pressure sensor (1) according to one of the preceding claims, characterized in that the drain region (7) and the source region (8), the gate electrode (6) each overlap.

9. Micromechanical pressure sensor (1) according to one of the preceding claims 1 to 4 or 6 to 8, characterized in that the thickness (d _S i) of

Membrane (3) is between 5 and 50 nm.

10. A method for producing a micromechanical pressure sensor (1) according to any one of claims 1 to 9, characterized by the steps

a. Producing a layer structure consisting of a first

Semiconductor layer (2), a sacrificial layer (4) and a further layer (5), wherein the sacrificial layer (4) forms an insulation layer (4) and lies between the first semiconductor layer (2) and the further layer (5), b. Inserting depressions (10) into the first semiconductor layer (2) in such a way that a depression (10) is bounded to the sides by a web (11) in each case and reaches as far as the insulation layer (4),

c. Undercutting a cavity (9) in the insulating layer (4) such that the cavity (9) by the first semiconductor layer (2) and the further layer (5) is limited,

d. Annealing the first semiconductor layer (2) to form an im

Substantially uniformly thick membrane (3) which limits the cavity (9) to one side,

e. Forming a gate electrode (6) in the further layer (5), f. Generating a drain region (7) and a source region (8) in the membrane (3), which are opposite each other and separated from each other by a conduction channel (12), which is the gate electrode (6)

opposite.

11. The method according to claim 10, characterized in that the

Recesses by parallel trenches (10) are formed.

12. The method according to claim 10 or 11, characterized in that the further layer (5) is a second silicon layer (5), and for forming the gate electrode (6) ions are introduced into the second silicon layer (5).

13. The method according to claim 12, characterized in that the ions

implanted through the membrane (3) in the second silicon layer (5).

14. The method according to any one of claims 10 to 13, characterized in that the layer structure is prepared by implantation of oxygen ions in silicon and subsequent annealing.

15. The method according to any one of claims 10 to 13, characterized in that the layer structure is made of a bonded SOI Waver.

16. The method according to any one of claims 8 to 12, characterized in that the introduction of the recesses by anisotropic etching of the first semiconductor layer (2).

17. The method according to any one of claims 10 to 16, characterized in that the cavity (9) is produced by isotropic etching.

18. The method according to claim 17, characterized in that the cavity (9) is produced by wet-chemical etching or reactive ion etching.

19. The method according to any one of claims 10 to 18, characterized in that the recesses (10) are introduced into the first semiconductor layer (2) such that the width of the trenches (10) is smaller than the width of the webs (11).

20. A method for reading a pressure sensor (1) according to any one of claims 1 to 9, characterized in that between the drain region (7) and the source region (8) applied a voltage and the current through the drain-source Area (7, 8) is measured at a voltage between the gate electrode (9) and the source region (8) below the threshold voltage, wherein according to an exponential dependence of the measured current of the pressure applied to the pressure sensor (1) ambient pressure, this ambient pressure is determined.