Low-cost high-sensitivity integrated pressure and temperature sensor

Senrors and Actuators A, 41-42 (1994) 398-401

398

Low-cost high-sensitivity integrated pressure and temperature sensor P. Pons and G Blasquez CNRSfLAAS, 7, Avenue du Colonel Roche, 31077 Toulouse Cedtx (France)

Abstract A reslstlve temperature detector along with a capaatlve pressure sensor usmg a sdxon membrane have been Integrated on a glass substrate The metrologxal charactenstxs of this mmlatunzed sensor feature a pressure sensltnnty of 129 pF/bar between 1 and 6 bar and a temperature sensitMy of 6200 ppmPC between -40 and +lXl “C In the pressure and temperature ranges consldered, the main cause of error IS non-hneanhes that generate lmpreasions less than or equal to *3% of the full scales The technological process allows these new devices to be batch produced

1. Introduction

2. Technology

Metal resistors allow for a simple, accurate and lowcost measurement of temperature In ad&tlon, capacltlve pressure sensors offer a high sensltlvlty and a low thermal drift Integration of these two types of sensors mto the same chip seems to be wable and econormcally desirable, because m numerous apphcahens the pressure and temperature must be measured simultaneously Integration 1s a pmn feasible because both types of sensors are probably able to operate without qmficant interferences More specfically, the capaatlve sensors are characterued by low consumption and, therefore, neghgble energy disslpatlon Their operation cannot cause a sign&ant increase m the chip temperature and therefore cannot mfluence the response of the resistive detector Furthermore, reslstlve sensors usually feature low gauge coefficients Hence the mechamcal stress caused by the pressure apphed to the chip must not lead to consistent variations of the resistance If the detector 1s located outside the high-stress zones Fmally, because resstive detectors are designed to avoid excessive self heatmg, the operation of these detectors should not greatly affect the response of the pressure sensor In the following, the results of a feasiblhty study on a sensmg cell based on the aforementioned principles are gwen The technology described hes at the crossroads of the work done by Lee and Wise on pressure sensors [l] and by Popescu and Popescu on resistive temperature detectors [2] It has allowed us to fabncate an mtegrated miniature cell whose metrologcal features are highly promising

Figure 1 shows the sensor’s structure. It includes a capacitive pressure sensor place on the r&t-hand side and a resistance temperature detector (RTD) on the left In its prelunmary version, the mam steps of the fabrication process are as follows on 7740 Pyrex substrates, a 3000 A thick nickel film 1s deposited by magnetron sputtenng under argon Then photohthography and chemml etchmg are used to form smultaneously the fixed plate of the pressure sensor and the RTD Concurrently, the lower sides of the (100) s&on wafers are locally etched over a depth of approximately 2 pm m order to produce the mner camties of the pressure sensors Then the upper sides of these wafers are locally thinned or pierced by chenucal rnicromachmmg m a KOH bath, m order to obtam the deformable plates of the pressure sensors and the holes glvmg access to the contact pads These operations are

0924-4247/94/$07 00 0 1994 Elsevler Sequoia All rights reserved SSDl 0924-4247(93100519-A

Rg 1 Basx sensor architecture

399

8

5:

2-

2

1

3

P

4

5

6

(Bars)

Fig 3 Pressure sensor response A, expenmentd data, -, leastsquares straight he Fig 2 Picture of the demonstrator

followed by alummmm deposition on the upper side of slllcon and by thermal treatment to a&eve the ohnuc contact with the deformable plate Fmally, the Pyrex substrates and the slhcon wafers are assembled by anodlc bondmg at 400 “C and cut mth a diamond saw The chops are then glued to TO3 bases to measure theu metrologcal charactenstlcs The technolo@cal process bnefly described above pernuts the collectwe fabncatlon of a large number of chips The mmunum number of photohthographlc operations 1s four Most of the eqmpment used is commercially available and also found m slhcon foundnes Figure 2 shows the demonstrator produced to test the feaslbllrty of this technolo@cal process The size of this mock-up is 7 mmX 7 mm It can be greatly reduced by optmuzmg all the technological parameters if the concomitant reduction m pressure sensitivity 1s acceptable

-4-

* 1

I 2

.

, 3

-

I 4

-

I 5

* 6

P (Bars) Fig 4 Pressure sensor

lmeanty errors

3. Characterization 3 I Pressure sensor 3 11 Bessure sensrhvzty Figure 3 shows an example of response of the demonstrator between 1 and 6 bar In this range, the response 1s almost lmear and the mean sensltlvlty 1s equal to 129 pF/bar 3 12 Lmeanty error In Fig 3 the expernnental data are represented by tnangular symbols and the sohd lme represents the least-squares strmght lme The hneatlty error El [3], expressed as a percentage of the full scale, is gwen m Fig 4 In the range of pressures considered, this error does not exceed 53%

1

2

3

4

5

6

P (Bars) Fig 5 Inthence of temperature on pressure-sensor -40 “C (0), +40 “C (B) and 120 “C (+)

response

3 I 3 Temperature response sensrt~rty The responses (capacitance changes) measured at -40, +40 and 120 “C are plotted m Fig 5 The temperature sensltmty 1slow and does not exceed 0 8% for a temperature variation of 160 “C

400

3 14

Error due to the measurement srgnal

A voltage of amplitude V applied to the plate pads generates an electrostatic pressure equal to

where e0 1s the vacuum permlttlvlty and d the distance between the plates On the demonstrator considered, a 10 V voltage gives rise to an electrostatic pressure of 1 mbar The error 1sneghglble In the case of sensors designed for measurement ranges less than one bar, this error must be taken into account

0

10

20

Power 3 2 Temperature sensor

FIN. 8 Temperature

3 2 1 Temperature senrrhvqv

Figure 6 shows an example of response (resistance change) between - 40 and + 130 “C It IS almost linear The mean thermal coefficient 1s 111the vlcuuty of 6200 PPm/“C 3 2 2 Lmeary error This error has been calculated and formulated m the same manner as m the case of the pressure sensor The result obtained 1s shown m Fig 7 It can be seen

‘I

-1 4 -40

80

40

0

120

T (“0 Temperature-sensor

-&I

-

-40

I

response

-

I

40

0

.

I

a0

T (“Cl Fig 7 Temperature

sensor

Imeanly

errors

-

I

120

seosor

30

40

50

(mW) self-heatmg

errors

that the hnearlty error ranges from -2 the full scale.

to +3% of

3 2 3 Injknce of the measLuement sgnal The measurement signal induces a current m the RTD through the Joule effect, which generates self heatmg of the RTD and possibly of the pressure sensor through thermal conduction in the Pyrex substrate Figure 8 shows the error expressed as a percentage of full scale on the temperature measurement dehvered by the RTD as a function of the measurement signal power Clearly the temperature measurement error 1s not m excess of 1% if the power remams less than or equal to 20 mW In these condltlons, the thermal dnft of the pressure sensor sensltlvlty IS negligible relative to the hneanty errors Finally, application of a 6 bar pressure to the chip does not produce any measurable effect on the RTD’s response

4. Conclusions A new sensmg cell for a pressure and temperature sensor has been presented It mcludes on the same Pyrex substrate a reslstnre temperature detector and a capacltlve cell for the measurement of the absolute pressure The technolo@cal fabncatlon process used combines processes under development, hke silicon mlcromachmmg and ahcon/glass anodlc bonding, with proven thm-film deposition techniques It 1s collective and relatmely straightforward Reahzation of the RTD does not call for a spectlic deposition From the metrological standpomt, this sensing cell presents high pressure and temperature sensitlvlties Cross-sensltlvltles to measurands are neghgble The mam causes of error are the non-hnearttles of the materials and the measurement prmclples selected For an accuracy of the order of several percent and a temperature range of the order of a hundred degrees,

401

thermal compensation of the pressure-sensing cell 1s optional Among future developments for this type of cell 1s a reduction m size The &mate mmlatunzatlon louts that can be envisaged are, on the one hand, the resolution of the capacitance-measurmg electronic clrcmts and, on the other, the inherent hrmtatlon of thm-fihn technologies It should be noted that size reduction will Increase the thermal couplmg between the two sensors without severely downgrading the accuracy This property results from the conslderatlons developed 111the Introduction and their confirmation through this feaslblllty study Acknowledgements The work has been supported by the PROMETHEUS/ PROCHIP programme and the Centre National de la

Recherche Sclentfique (CNRS) The authors would hke to thank A Coustre and F Garajedagui for their fruitful advice and the TEAM semce for its techmcal support m the fabncation of the expenmental devices

References 1 Y S Lee and KS Wne, A batch-fabncated sdtcon capacltwe pressure transducer wtb low temperature sensltnnty, IEEE Tmns Electron Dewes, ED-29 (1982) 42-48 2 C Popescu and M Popescu, Thermo-reslstlve sensors, m P Cmreanu and S mddelhoek (eds ), Thm Aim Remtwe Sensors, Institute of Phywx, Bnstol, 1992, pp 214-252 3 G Blasquez, P Pans and R Behocarray, Feaslblllty of capacltwe pressure sensor wthout compensation cwxut, Senws and Achum~n A, 37-38 (1993) 112-115