History:

35 years ago, the Author found his first job inthe pressure measurement industry, and thecost of a typical pressure transmitter was inthe range of $350 to $600. The predominanttechnology at that time was the unbondedstrain gauge from Bell and Howell or thebonded strain gauge from Hottinger andBaldwin. There was also an LVDT fromSchaevitz, and capacitance, potentiometerand reluctance pick ups connected to deflec-ting diaphragms and bellows; apart fromdozens of other measuring systems whichroamed around the market place.

These technologies were either too expensiveor possessed inadequate stability for mostrequirements.

At the beginning of the 60s, Statham broughtthe first thin-film transducers with good sta-bility onto the market and at about the sameprice as existing strain gauge types. Thismarked the start of the breakthrough for thistechnology. Large price reductions for highquantities were not aimed for, however.

The demand for inexpensive, stable trans-ducers however could not be ignored andthere were hectic developments at variouslevels. The majority were based on the piezo-resistive effect discovered and published byPfann and Thursten. The Gauge Factor (theunit change of resistance with unit change oflength) for metallic wires is ~ 2 and is mainlycaused by the extension and "thinning" of awire under strain, with a small additionalpiezoresistive (change of resistivity withstrain) effect. However, the piezoresistive effect in dopedsemiconductor resistors can be up to 50 timesgreater, giving Gauge Factors of ~100.

There were many developments to processthe brittle semiconductor materials into thinwhiskers (semiconductor strain gauges) andto bond them to carriers such as diaphragmsor flexible beams. Some Companies, whomastered this difficult problem, such as DataInstruments, Sensometrics and Microgaugestill offer this technology today.

The researchers became extremely enthusias-tic when, in 1964 the first planar transistor

came onto the market- the beginning ofintegrated circuit (IC) technology, and com-plete resistance bridges could be integratedinto a small silicon circuit. There were manyattempts at this time to bond these small flatchips onto metal diaphragms etc. Renownedcompanies such as Philips stuck their necksout and went under. These failures gavepiezoresistive technology a disastrous imageand strengthened the general conviction thatonly thin-film gauges could be really stable.An attitude which continues to be felt today...

Honeywell first came up with the idea ofusing the silicon chip with diffused resistorsas the pressure diaphragm. Now the tinybrittle diaphragm has to be bonded to a carrierof steel or glass. The problem of bondingmaterials as different as silicon and steel waspartly overcome because the sensor elementswere only indirectly influenced by the bodystresses of the bond and the majority of thesestresses could be eliminated by the configu-ration of the resistors on the diaphragm. Tony Kurtz left Honeywell at this stage ofthe technology and founded Kulite.Druck, in England also founded their com-pany on this technology.

The great breakthrough in piezoresistivetechnology came with the reproducible, inte-grated, homogenous silicon measuring cell,for which Honeywell designated the Authoras inventor on the original patent. It wasimmediately foreseen that these pressure sen-sors could be produced cheaply and in highquantities. Art Zias brought this technologyfrom Honeywell to National Semiconductor,who expected this to bring in big business.The first large-scale production for thesemeasuring cells was, however launched byDelco for the MAP sensor (Manifold Pres-sure). In 1966, the Clean Air Bill was passedby the American Congress, which restrictedthe pollutant discharge for cars. The automotive industry demanded adeferment because the technology, and aboveall economic sensors, were not yet available. At the beginning of the 70s, the first carscame onto the market with electronic controlsystems and MAP silicon sensors, whichcould fulfill the conditions of the Clean AirBill. The silicon sensors were built into thecars either naked or with gel protection. Thisprovides insufficient protection for industrial

applications or level measurements, but appa-rently sufficient for this requirement.

Considerable indecision predominated atNational Semiconductor about how the pro-duct should be brought to the industrialmarket. The Parylene coating promoted aswater protection had to be recalled after 6months, for example!The handbook from National Semiconductorwritten by Art Zias – the Mister Piezo ofAmerica - showed, for example, instructionsfor a level transmitter. You take a rubberglove, fill it with silicon or olive oil, stick thecable with the sensor and electronic circuitinto the glove, tie together the opening of theglove with string and submerge the wholething in the sauce.In theory however, there was a lot of geniusin the handbook and, according to a state-ment from Art Zias, the income which Natio-nal Semiconductor got from the sales of thebook was for a time larger than the sales ofthe measuring cells, which just shows howmany were occupying themselves with theproblem of pressure sensor technology! In alast desperate attempt, National started anadvertising campaign under the slogan:

"A brand-new way to measure pressure:With the naked silicon sensor."

The Author parodied back:"A brand-new way to make love: Naked"

Finally, National Semiconductor sold thesensor division, which passed throughvarious hands before finally emerging asSensym.

The inability of the US manufacturers todevelop a protective housing for the siliconsensor enabled new technologies to break-through into the industrial as well as theautomotive sector: The capacitive ceramicmeasuring cell from Kavlico, (which is alsoproduced today by Texas Instruments) inquantities of millions.

The Europeans and Japanese were not muchcleverer. In Europe, Magnetti- Marelli (Fiat)developed the ceramic thick-film straingauge measuring cell, which however has notachieved the success of the capacitive measu-

1/2

The piezoresistive technologyon the right track.Statement:The pressure transmitter which will dominate the industrial market, has a silicon sensor chip with integrated electronicsmanufactured using IC (Integrated Circuit) technology, and built into an oil-filled housing with an isolating diaphragm.

ring cell. Technologically speaking, this isdue to instabilities caused by the influence atthe point of adhesion of the ceramic/sensorelement, and also by its extremely highbridge resistance. Here again, the experiencesof silicon technology were neglected.

From Japan came a product from NipponDenso where the pressure on the nakedsilicon sensor is applied from behind,allowing the electronics and the sensorgauges to be protected by a gel layer, butthey are still exposed to the referencepressure atmosphere. Apart from the fact thatabsolute pressure sensors can hardly bemanufactured at all, the Author’s analysis ofthe transmitter design creates the same im-pression of helplessness as shown byNational Semiconductor.

The Present and the Future

In 1993, in the article "Marketing of PressureSensors" in the Trade Fair News we hadalready predicted the following:

Sooner or later, the market positions will bedetermined by the competing technologies. Inan open market, we estimate that the majorityof industrial applications in the range of 10mbar to 1 bar will be dominated by thecapacitive ceramic measuring cell. In the range of 1 bar to 1000 bar, monolithicsilicon piezoresistive technology (IC) willdominate the field.Other technologies will have to operate inniche markets because, for example, the costsof thin-film technology are too high for verylarge-scale projects.

That was written 6 years ago. In certain areas- such as cooling compressors - silicon piezo-resistive technology has succeeded in brea-king the dominance of capacitive ceramictechnology. But large volume automotiveprojects continue to run with ceramic capa-citive sensors. Nevertheless, we stand by ourstatement.

KELLER AG introduces a new silicon piezo-resistive transmitter (Figure 1) which, in ouropinion, has fundamental technologicaladvantages over the capacitive ceramic sensorand will in the future attempt to break itsdominance, even in demanding automotiveapplications.

The development of the housing with a sepa-rating diaphragm has been the focal point ofdevelopment at KELLER AG since itsfoundation 25 years ago. Today, KELLERhas processes in-house for the completesilicon sensor housing, which have compa-

rable manufacturing costs to the ceramicsensor housing. In a patented process, a continuous, belt fedfurnace brazes together the brass housing, asteel insert and a nickel diaphragm in oneoperation. The process can be fully auto-mated.

As far as electronics and trimming adjust-ment are concerned, no cost advantage can bedetermined for either technology. Accuracyis also no longer a topic for either of them.The stabilities are excellent, and customer-specific circuits (ASICs) can compensate forinaccuracies.

The advantages and disadvantages are there-fore determined only by the costs of thesensors and of housing them.However, while 5000 silicon sensors can fiton one 6 inch silicon wafer, 10 large bakingtrays are necessary for 5000 ceramic cells. 5000 silicon sensors pass through the variousscreening processes together, but 5000 cera-mic sensors have to pass through individually.

Mounting silicon sensors onto a TO5 headerwith glass feed-throughs and welding it intothe housing are similar to packaging a transis-tor. All the processes are automated, utilizingmachines from the semiconductor industrysuch as Die Bonding and automatic WireBonding, and is improving year by year. Thefinal assembly is by welding beneath the oil,and has been proven for over 15 years.

Finally the silicon sensor lies protected fromall adverse influences, in a sealed oil cham-ber. The only materials in contact with thepressure media are metals and braze.

Conversely, the ceramicsensors are sealed into thehousing with an O-ring,which has to be selected tosuit the pressure media. Thesame O-ring cannot be usedfor both hot water and petrol.Furthermore, the O-ring isalways a potential cause offailure. Envec (Endress +Hauser) attempted to reassurecustomers by means of anadvertising campaign with thefollowing statement:"Pressure transmitters mayfail after 5 years; sure, andBayern Munchen may berelegated next year."To be accurate, Envec shouldqualify this statement andsay: "Pressure transmitterswith O-ring sealed ceramicmeasuring cells fail after 5years."

With the Envec statement, a problem specificto their technology is generalised. But this isonly an aside…

Currently, a large part of the material costs ofa transmitter is taken by the associatedelectronics. However, a further reduction incosts is foreseeable with piezoresistivetechnology. Today, there are already piezo-resistive sensors from Bosch and Fuji, withadjustment electronics integrated into thepiezoresistive silicon chip. Adjustment isdone by laser trimming on the surface of thechip. Work is now being done on similarcircuits where the adjustment to the siliconchip takes place via interface leadouts frominside the oil chamber itself.

The component density of IC technology isincreasing so fast that according to ourestimate, in 10 years a pressure sensor,amplifier, and digital compensation with A/Dand D/A converters can be integrated ontothe piezoresistive silicon chip at unit costs of$1. Then the complete transmitter withelectronics is optimally protected in the oilchamber and the capacitance of the glassfeed-throughs will automatically guaranteeexcellent EMC protection.

30 years ago we aimed for the $10 / 2%transmitter.Within the next 10 years it is the $5 / 0.1%transmitter.

H.W. Keller is the owner of

KELLER AG für Druckmesstechnik

(Figure 1)

2/2