Method for manufacturing a semiconductor pressure sensor

0001The subject of the present invention is a method for manufacturing a semiconductor pressure sensor.

0002As is known, a pressure sensor is a device that converts a variation in pressure into a variation of an electrical quantity (a resistance or a capacitance). In the case of a semiconductor sensor, the pressure variation is detected by a membrane of semiconductor material, which overlies a cavity and is able to undergo deflection under mechanical stress.

0003Pressure sensors using semiconductor technology typically find their application in medicine, in household appliances, in consumer electronics (cell-phones, PDAs -Personal Digital Assistants-), and in the automotive field. In particular, in the latter sector, pressure sensors are used traditionally for detecting the pressure of the tyres of motor vehicles, and are used by the control unit for alarm signalling. Pressure sensors are, on the other hand, also used for monitoring airbag pressure, for controlling the breakdown pressure of the ABS, and for monitoring the pressure of oil in the engine, the pressure of injection of the fuel, etc.

0004Currently existing sensors manufactured using the semiconductor technology are basically of two types: piezoresistive and capacitive sensors.

0005Operation of piezoresistive sensors is based upon piezoresistivity, i.e., upon the capability of some materials to modify their resistivity as the applied pressure varies. Piezoresistors are normally formed on the edge of a suspended membrane (or diaphragm) and are connected to one another in a Wheatstone-bridge configuration. Application of a pressure causes a deflection of the membrane, which in turn generates a variation in the offset voltage of the bridge. By detecting the voltage variation with an appropriate electronic circuit, it is possible to derive the desired pressure information. An example of a piezoresistive sensor of the above type is described in EP-A-822 398.

0006Sensors of a capacitive type are based upon the change in capacitance that occurs when a pressure is applied on a flexible membrane suspended above a support and separated therefrom by a region that is empty or filled with gas (air gap). Two examples of silicon sensors of a capacitive type are described in "A MEMS-Based, High-Sensitivity Pressure Sensor for Ultraclean Semiconductor Applications", A.K. Henning, N. Mourlas, S. Metz published on: http://www.redwoodmicro. com/Papers/ASMC.pdf and "Application of High-Performance MEMS Pressure Sensors Based on Dissolved Wafer Process" A. Tadigadapa, S. Massoud-Ansari, published on: http://www.memsissys.com/pdf/issystech2.pdf.

0007In the case where the dielectric present between the two electrodes is a vacuum, an absolute pressure sensor is obtained, whereas, if gas is present, which is generally introduced hermetically at a known reference pressure, the detected capacitance variation is linked to the difference between the external pressure and the internal pressure, and, consequently, a relative pressure sensor is obtained.

0008Application of a pressure causes a deflection of the membrane with consequent reduction in its distance from the bottom electrode. In this way, the capacitance of the pressure sensor increases. By measuring the difference between the capacitance thus obtained and the rest capacitance (i.e., in the absence of stress), the pressure variation detected by the sensor is obtained.

0009Also in this case, a circuit for processing the electrical signals generated by the capacitive sensor provides the information of pressure sought.

0010In general, capacitive technology presents a lower current consumption than does piezoresistive technology. Consequently, capacitive sensors are preferable in those applications where power consumption is an important parameter, for example, in the automotive field, wherein accurate control of the load power is required. Moreover, capacitive pressure sensors present smaller overall dimensions and lower costs, as is required in numerous applications.

0011Both piezoelectric sensors and capacitive sensors hence call for the construction of a cavity underneath the flexible membrane.

0012Currently, various solutions have been proposed: <ul id="ul0001" list-style="none" compact="compact"><li>1. use of silicon-on-insulator (SOI) substrates; for instance, pressure sensors using this solution are described in US-A-5 369 544; US-A-5 510 276; US-A-6 131 466;</li><li>2. use of porous silicon (see, for example, US-A-5 242 863);</li><li>3. wet etching from the front (see, for example, US-A-4 766 666);</li><li>4. wet etching from the rear, using tetramethyl ammonium hydroxide (TMAH);</li><li>5. other methods (see, for example, US-A-4 744 863).</li></ul>

0013In all known solutions, the use of semiconductor technology for making cavities underneath suspended structures and layers calls for processes that are complex, costly and, in some cases, far from compatible with the manufacturing steps currently used in the semiconductor industry for manufacturing integrated circuits.

0014The aim of the present invention is hence to provide a manufacturing method which will overcome the disadvantages of known solutions.    According to the present invention a method for manufacturing a semiconductor pressure sensor and the respective pressure sensor are provided, as defined in claims 1 and 13, respectively.    For an understanding of the present invention, preferred embodiments thereof are now described purely by way of nonlimiting example, with reference to the attached drawings, wherein: <ul id="ul0002" list-style="dash" compact="compact"><li>Figure 1 is a cross-section through a wafer of semiconductor material in an initial manufacturing step;</li><li>Figure 2 is a top view of the wafer of Figure 1;</li><li>Figure 3 is a cross-section of details of Figure 2, at an enlarged scale;</li><li>Figures 4-9 are cross-sections through the wafer of semiconductor material of Figure 1, in subsequent manufacturing steps, for a pressure sensor of capacitive type; and</li><li>Figure 10 is a cross-sectional view through a wafer of semiconductor material for a pressure sensor of piezoelectric type.</li></ul>    Hereinafter, an embodiment of a process for manufacturing a semiconductor material sensor of capacitive type is described. The present process is based upon the process disclosed in EP-A-1 324 382 for manufacturing a SOI wafer, and, more precisely, refers to the second embodiment shown in Figures 11-14 of said document.

0015Figure 1 shows a wafer 1 of semiconductor material, preferably monocrystalline silicon, comprising an N-type substrate 2, designed to form the bulk of the device. A resist mask 3 (visible in the enlarged detail of Figure 3) is formed on the top surface of the substrate 2. The mask 3 has two circular areas, designated by 4a and 4b and hereinafter referred to as sensor area and reference area, in each of which a honeycomb lattice is defined, the two lattices being of different sizes.

0016In particular, as appears in the enlarged detail of Figure 2, the sensor area 4a has mask regions 5a with an hexagonal shape arranged close to one another (see also the cross-section of Figure 3), while the reference area 4b has mask regions 5a that are more widely spaced. For example, the distance <u>t</u> between opposite sides of the mask regions 5a and 5b may be 2 µm, the distance d1 between sides facing adjacent mask regions 5a may be 1 µm, and the distance d2 between sides facing adjacent mask regions 5b may be 2 µm.

0017Using the mask 3, trench etching of silicon of the substrate 2 is performed, so forming a sensor trench 6a and a reference trench 6b at the sensor area 4a and at the reference area 4b, respectively. The channels of the sensor and reference trenches 6a, 6b may have, for example, a depth of approximately 10 µm, are of different width, as may be seen in Figure 3, and delimit silicon columns 7a and 7b, respectively, which are identical and have a shape corresponding to that of the mask regions 5a and 5b.

0018Next (see Figure 4), the mask 3 is removed and an epitaxial growth is performed in a deoxidizing environment (typically, in an atmosphere with a high concentration of hydrogen, preferably using trichlorosilane - SiHC ₁₃). Consequently, an epitaxial layer 10 (shown only in Figure 4 and hereinafter not distinguished from the substrate 2) of N type, grows on top of the silicon columns 7a and 7b and closes, at the top, the sensor and reference trenches 6a, 6b, trapping the gas present therein (here, molecules of hydrogen H₂). The thickness of the epitaxial layer 10 may be, for example, 9 µm. The epitaxial layer 10, of N type, is shown only in Figure 4 and hereinafter is not distinguished from the substrate 2.

0019An annealing step is then carried out, for example for 30 minutes at 1190°C.

0020As discussed in the aforementioned patent EP-A-1 324 382, annealing causes a migration of the silicon atoms, which tend to arrange themselves in lower-energy position. Consequently, at the sensor trench 6a, where the columns 7a are arranged close together, the silicon atoms migrate completely and form a sensor cavity 11, closed at the top by a membrane 13. On account of the presence of the sensor cavity 11 (having, for example, a diameter of 600 and 200 µm, according to the pressure to be applied), the membrane 13 is flexible and can be deflected under external stresses.

0021Instead, at the reference trench 6b, where the columns 7b are arranged at a bigger distance from one another, the migration of silicon atoms causes only a thinning of the columns 7a, hereinafter indicated as pillars 15. In practice, a labyrinthine cavity 12 is formed, wider than the reference trench 6a. Furthermore, the pillars 15 in the labyrinthine cavity 12 prevent any movement to the overlying region, hereinafter referred to as electrode region 14.

0022Preferably, annealing is performed in an H₂ atmosphere so as to prevent the hydrogen in the sensor and reference trenches 6a, 6b from escaping through the epitaxial layer 10 to the outside and so as to increase the concentration of hydrogen in the cavities 11 and 12, should the hydrogen trapped inside during the epitaxial growth be not sufficient. Alternatively, annealing can be carried out in a nitrogen environment.

0023The crystallographic quality of the membrane 13 is excellent, as is evident from tests carried out by the present applicant.

0024Next (see Figure 5), the membrane 13 and the electrode region 14 are doped via implantation of P-type dopant species, for example boron. Subsequently (see Figure 6), an access trench 20 is dug just in the electrode region 14, from the surface of the wafer 1 as far as the labyrinthine cavity 12. The access trench 20 preferably has the shape shown in Figure 6, and hence extends, by stretches, near the periphery of the area occupied by the labyrinthine cavity 12.

0025Thermal oxidation of the columns 7b is then carried out so as to form an oxidized region 21 underneath the electrode region 14. The necessary oxygen is fed to the labyrinthine cavity 12 through the access trench 20. In this step, there is a gradual growth of the oxidized region 21 at the expense of the columns 7b and of the silicon of the substrate 2 surrounding the access trench 20 and the labyrinthine cavity 12. In particular, the columns 7b are completely oxidized and increase in volume. As shown in Figure 7, the labyrinthine cavity 12 and the access trench 20 are filled in part with thermal oxide, but remain partially open (remaining portions 12' and 20' of the labyrinthine cavity and of the access trench).

0026Next (see Figure 8), the remaining portions 12' and 20' of the labyrinthine cavity and of the access trench are filled with insulating material 22, for example TEOS, forming, as a whole, an insulating region 24. In Figure 8, for clarity, the demarcation line between the insulating material 22 and the oxidized region 21 is represented by a dashed line. As an alternative, the labyrinthine cavity 20' can remain empty, thus avoiding the filling step.

0027A P-type implantation, an N-type implantation and respective diffusion steps are then carried out in order to form contact regions 25a, 25b of P⁺-type above the membrane 13 and the electrode region 14 as well as contact regions 25c, 25d of N⁺-type above the substrate 2 (see Figure 9). The contact regions 25c, 25d preferably have an annular shape and extend, respectively, around the membrane 13 and around the electrode region 14. Next, metal contacts 26a, 26b, 26c and 26d are formed and contact the contact regions 25a to 25d, respectively.

0028In practice, the structure of Figure 9 forms two capacitors, designated by C1 and C0, which have as first electrode the membrane 13 and the electrode region 14, respectively; as second electrode, the bulk region underlying the membrane 13 and the bulk region underlying the electrode region 14, respectively; and as dielectric, the sensor cavity 11 and the insulating region 24 (or the oxidized region 21 and the labyrinthine cavity 20'), respectively.

0029The capacitor C1 (referred to also as sensing capacitor) represents the element sensitive to the pressure that is applied on the membrane 13, while the capacitor C0 (reference capacitor) represents the reference element, which provides the rest capacitance. Since the areas of the P/N junctions of the sensing capacitor C1 and of the reference capacitor C0 are equal, these capacitors have the same junction capacitance and the same leakage currents. In addition, the reference capacitor C0 undergoes a trimming step at the wafer level, using one or more known capacitors arranged in parallel and using a one-time programmable (OTP) device.

0030If so desired, prior to forming the contact regions 25a-25d, it is possible to integrate the electronic components making up the control circuitry on the same chip of the pressure sensor.

0031Finally, in a way not shown, the wafer 1 is cut into dice, each containing a sensing capacitor C1 and a reference capacitor C0 (as well as, if envisaged, the control circuitry), and the dice are encapsulated in such a way that the membrane 13 is accessible from the outside.

0032Operation of the pressure sensor according to the invention is described hereinafter.

0033If a pressure is applied on the membrane 13, the latter is deflected, reducing its distance from the bottom electrode (substrate 2). Consequently, the capacitance of the sensing capacitor C1 increases. If the difference between the signal supplied by the sensing capacitor C1 and the signal supplied by the reference capacitor C0 is measured via an electronic circuit for signal processing of the "fully differential" type, there is rejection of the common-mode components and amplification of the differential ones, and hence an indication of the pressure applied is obtained.

0034The advantages afforded by the described pressure sensor emerge clearly from the foregoing description. In particular, thanks to the described manufacturing process, the silicon pressure sensor is of low cost and reduced dimensions, and hence can be used in numerous applications where these requirements are important.

0035If so desired, it is possible to integrate on the same chip the sensing capacitor C1 and the reference capacitor C0 and the relative control circuitry, thus reducing the manufacturing and assembly costs, as well as the overall dimensions of the device.

0036Finally, it is clear that numerous modifications and variations may be made to the pressure sensor described and illustrated herein, all falling within the scope of the invention, as defined in the annexed claims.

0037In particular, the described technique can be used, with just a few modifications, for producing a pressure sensor of a piezoelectric type. In this case, in fact, it is sufficient, during the final manufacturing steps, for example, simultaneously with the manufacture of the components of the control circuitry, to form piezoresistive elements near the periphery of the membrane 13. Preferably, the piezoresistive elements are of P-type, and the membrane is of N-type. In case of a pressure sensor of piezoelectric type, however, it is not necessary to provide a reference element, and hence all the steps necessary to form the labyrinthine cavity 12 and the insulating region 24, as well as boron implantation, are omitted. An embodiment of a pressure sensor of piezoelectric type is shown in Figure 10, wherein the elements in common with the embodiment of Figure 9 are designated by the same reference numbers.

0038As is known, in Figure 10 no insulating region 24 is present, and resistors 30, implanted or diffused, here of a P-type, are formed on the periphery of the membrane 13 (of an N^--type). In Figure 10, only three resistors 30 can be seen, but it must be understood that a fourth resistor is formed in the non-visible part of the wafer 1 and is connected to the visible resistors 30 in a bridge configuration. In Figure 10, the interconnections between the resistors 30 (typically metal regions extending above an insulating layer, not shown) are represented schematically.

0039As an alternative, the resistors 30 may be made of polysilicon above the membrane 13.

0040For a capacitive sensor, the membrane may be of any shape, for example square or generically polygonal, even though the preferred shape is circular, since it prevents any stress concentration.

0041The contact of the capacitors C0 and C1 with the second electrode can be made, instead of on the front of the device, on the rear, as indicated by the dashed line 35 in Figure 9; in this case, it is expedient to use a substrate 2 of N⁺-type in order to reduce the access series resistance. Also in this case, the epitaxial layer is of N^--type.

0042The shape of the columns 7a, 7b may vary with respect to what is illustrated. For example, they can be replaced by diaphragms of semiconductor material with reduced thickness, or, in general, by other thin structures (referred to also as walls) capable of enabling migration of silicon during the annealing step and forming the sensor cavity 11 and the labyrinthine cavity 12 (for a capacitive implementation).