IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 50, NO. 6, DECEMBER 2001 1519

A Silicon-Etched Probe for 3-D CoordinateMeasurements With an Uncertainty Below 0.1�m

Han Haitjema, Wouter Pril, and P. H. J. Schellekens

Abstract—The increasing demand for accurate coordinate mea-surements on products demands new concepts of probe design.Results of some realized designs will be given. One of the mostpromising utilizes microtechnology and etching in silicon in orderto realize the necessary dimensional design with flexure hinges. Mi-crotechnology is also used for the detection system; strain gages areintegrated in the probe. Results for two probes will be given andpossible future developments will be discussed.

Index Terms—CMM-probe, coordinate metrology, MEMS.

I. INTRODUCTION

I N BOTH manufacturing and measuring technology, an on-going trend for higher accuracies can be seen, as is reflected

in Fig. 1. Taniguchi noticed this trend [1] and extrapolated it tothe future, well predicting the present state of the art in preci-sion engineering [2]. Especially during the last three decades,advances in IC-technology pushed the state of the art forward.Also, in consumer electronics, precision techniques found theirway, e.g., in video players, CD, and its successor digital versa-tile disc (DVD), and hard disks.

Together with the increasing precision of products, the needfor highly accurate dimensional inspection increases. A coordi-nate measuring machine (CMM) is often used to accomplish thistask. Recently, the uncertainty of these machines, excluding theprobe, has entered the submicrometer regime and further effortis done to decrease it further [3]. Probe systems with three-di-mensional (3-D) accuracy substantially below one micrometer,however, are not available. There are two different probe types:1) the touch-trigger probe and 2) the measuring probe.

The touch trigger probe was introduced by Renishaw in theearly seventies. As soon as the stylus is moved from its zeroposition, the resistance of an electrical circuit is altered. Atthat moment, the scales of the CMM are read. An exampleof such a probe system is shown in Fig. 2. Measuring oranalog probe systems are measuring the probe ball positioncontinuously. After a surface detection, the CMM is stoppedand controlled by signals of the probe system to reach a presetprobing force. This technique leads to a significant decrease ofthe uncertainty down to 0.5m. In most commercial analogprobe systems, both the measuring system as the suspensionof the stylus have a larger mass compared to the touch triggerprobe. Therefore, the probing speed must be lowered in order to

Manuscript received March 1, 2001; revised July 30, 2001. This work wassupported by Mitutoyo Nederland B.V., the Dutch Technology FoundationSTW, and the NMi-Van Swinden Laboratorium (NMi-VSL).

The authors are with Eindhoven University of Technology, Section PrecisionEngineering, 5600 MB Eindhoven, The Netherlands.

Publisher Item Identifier S 0018-9456(01)10932-0.

prevent intolerable high forces at the moment of probing. Alsothe controlling sequence takes some time, so probing is moretime-consuming than in the touch trigger case. An example ofan often used analog probe system is shown in Fig. 3.

The goal of this research is to develop a new probe systemwhich is lighter and more accurate than such probe systems.

II. BASIC SUSPENSION

From Fig. 3, it is evident that the three stacked guide waystake a lot of mass and space. Looking for an alternative, a sus-pension was designed based on slender rods. A slender rod ishere defined as a rod-shaped elastic element, which fixes onedegree of freedom.

In Fig. 4, an alternative design is given based on elastic ele-ments. The probe is connected to an intermediate body which issuspended to the probe house by three slender rods tangentiallytouching the intermediate body. This suspension has an equalstiffness for all horizontal probing directions. Because of sym-metry, the thermal center is at the z-axis.

Disadvantages are that due to parasitic translations in thelength direction of the rods when they are translated out of thex–y-plane, the probe will rotate around the z-axis when movedin vertical direction. Besides this, when the tip is not on thez-axis, it will move over a small distance in x- or y-directionwhen moved in the z-direction.

III. REALIZATION OF A TWO-DIMENSIONAL PROBE

A probe was designed based on the design in Fig. 4. As a firsttrial for the detection of the displacements, a laser diode gratingunit (LGDU) was used. As this system can only measure two di-mensions: the angle of the suspension and the z-displacement,it was designed as a two-dimensional (2-D) probe; the two di-mensions being the z- and y-direction, as indicated in Fig. 4.Designed on this base, the experimental probe set-up will com-prise a probe housing, a detector, the elastic suspension, and theprobe tip attached to it.

A. Angle/Displacement Measurement Using an LGDU

For the detection of the z-displacement and the rotationaround the x-axis (giving an y-displacement we used an LGDUprobe as this is known to be the most accurate noncontact probeprinciple. The LDGU was designed for use in CD playerswhere it fulfills a threefold task: it generates two feedbacksignals to focus the laser spot on the disk and to follow thetrack, and it reads the digital data.

It can be used outside the CD player as a distance and anglemeasuring device This is commercially explored by Sensor 95.

0018–9456/01$10.00 © 2001 IEEE

1520 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 50, NO. 6, DECEMBER 2001

Fig. 1. Prediction of accuracy limits in the course of time [1].

Fig. 2. Schematics of a touch trigger probe.

Experimental results of set-ups using LDGUs are given in [4].The LDGU is used in the 2-D optical probing system to measurethe z-displacement and one of the lateral displacement of theprobe.

The frame is taken such that the probe is sensitive for z andx-directions. The set-up is depicted in Fig. 5. The LGDU detectsindependent the z-displacement of the mirror and its rotationaround the y-axis.

B. Mechanical Design

A probe house was designed as to connect mechanicallythe LDGU, the lenses, the suspension, and the electronics to

a CMM. As the LGDU produces some heat, a drain for theproduced heat was provided. The cross section of the finaldesign is given in Fig. 6.

C. Results

The probe was calibrated for displacements in the z-directionby a calibration set-up based on a laser interferometer system.The output from the LGDU-electronics (as a voltage) was com-pared to the displacement measurement by a laser interferom-eter. The capabilities of the LGDU become clear when we con-sider the residuals after subtracting a first-order or higher order

HAITJEMA et al.: SILICON-ETCHED PROBE FOR 3-D COORDINATE MEASUREMENTS 1521

Fig. 3. Example of an analog probe system.

Fig. 4. Suspension of the probe to its housing by three slender rods:E: elasticelements,B: frame bars (stiff), andS: position of piezoresistive sensors (ifapplied).

polynomial from the calibration curve. The residuals are shownin Fig. 7.

The figure shows a smooth nonlinearity which can well be de-scribed by a third-order polynomial. After removal of this poly-nomial sharp, spiky deviations of a few nanometers occur. Theseare caused by multiple reflections of the laser light inside the op-tical system. Introducing a quarter-wave plate in the optical pathprobably will reduce these effects.

IV. REALIZATION OF A THREE-DIMENSIONAL PROBE SYSTEM

USING PIEZO RESISTIVE STRAIN GAUGES

The design was extended to 3-D and essentially miniatur-ized by the idea of integrating the displacement-sensors in theslender rods itself. This lead to the idea to make the whole probebasically out of silicon and to use silicon-etching techniques tointegrate the strain gages. This silicon-device then still has to bemounted in some probe housing and with a probe rod and ball.This is described in this section.

Fig. 5. Basic set-up for the LGDU as a distance sensor.

Fig. 6. Cross section of the LDGU probe.

A. Displacement Measurements Using Strain Gages

Usually stain gages are connected to the stressed surfaceby gluing. Due to this hysteresis may arise which results inuncertainty. By using several evaporation, lithography andetching steps the slender rods, the strain gages, and theirelectrical connections can manufactured in one setup. Thistechnology belongs to the field of microsystem technology(MST). It enabled integrated fabrication of mechanics, sensors,actuators, and electronics on a micrometer scale. As a spin-offof integrated circuit technology, silicon is used as a substrate.

Calculations showed that the strain gages are best placed atthe two ends of the slender rods. This is indicated at the left-handside of Fig. 8; the bending due to a force F is maximal at theedges. The readout can best be done by a Wheatstone-bridge. Atthe right-hand side of Fig. 8, the schematic of this is depicted,where temperature-effects are compensated.

The probe design with these strain gages becomes then asdepicted in Fig. 9. Each of the three strain-gage assemblies are

1522 IEEE TRANSACTIONS ON INSTRUMENTATION AND MEASUREMENT, VOL. 50, NO. 6, DECEMBER 2001

Fig. 7. Left: residual of the calibration in the z-direction of the 2-D probe after a linear fit. Right: the same after fitting a third-order polynomial.

Fig. 8. Left: strain gages on a rod in a top and a side view. Right: TheWheatstone-bridge configuration in which the strain gage deformation ismeasured.

Fig. 9. Probe design with strain gages in the slender rodsP: stylus platform,E: elastic element, andS: piezoresistive sensor.

wired into a Wheatstone-bridge, as depicted in Fig. 8. The x, y,and z-coordinates of the probe can be calculated from the strains(displacements) in each rod after calibration.

B. Producing the Probe by MEMS

Manufacture of Chip Containing Wiring, Slender Rods andSensors:Chips are manufactured in the usual way on 3-in sil-icon wafers. In successive steps of illumination, etching, and

Fig. 10. Picture of the realized probe.

adding insulation layers by chemical vapor deposition (CVD),the chips are realized. The chip size is mm. The basicstructure, after being cut from the wafer, is shown in Fig. 10.

Mechanical Parts: The ball and the stylus are mounted on analuminum tripod. The tripod and the stylus are very stiff com-pared to the slender rods.

Gluing: The chips are mounted (glued) in an aluminumhousing (see Fig. 10) and the tripod with the stylus and theball are mounted on the chip by gluing. Finally, a flex cable ismounted to the chip wiring and it is connected to the aluminumhousing.

Electronic Testing of Wiring and Sensors:The wiring andthe sensors on the chip are tested before and after mounting thestylus with the tripod and the flex cable.

C. Results of Calibrations of the Probe

The ac impedance of the strain gage resistors has been testedas a function of the frequency of the applied voltage. This re-

HAITJEMA et al.: SILICON-ETCHED PROBE FOR 3-D COORDINATE MEASUREMENTS 1523

Fig. 11. Residuals of the excitation of the bridges as function of the imposeddisplacement in the z-direction after subtracting of best fitting straight lines. Therod number is indicated on the right side of the plot. Results of rod 2 and 3 areshifted.

Fig. 12. Residuals in the measured direction of the excitation of the bridgesas function of the imposed displacement in the x–y plane after subtracting bestfitting straight lines. The rod number is indicated on the right side of the plot.Results of rod 2 and 3 are shifted.

vealed a purely resistive behavior up to about 40 kHz. The probewas calibrated in a similar way as described in Section III-C.

We give two examples here: the calibration in the z-directionand a calibration a direction in the x–y plane.

The result of the calibration in the z-direction is depictedin Fig. 11. This figure shows for each rod the deviation fromlinearity for several repeated measurements in an up–down se-quence. The figure shows that the output of the bridge is verylinear with the displacement; the residuals being in the 10-nmrange. As no difference between the up and down direction areobserved, hysteresis is virtually absent due to the integration ofthe strain gages.

A similar figure giving the linearity deviations for a displace-ment in the x–y plane is given in Fig. 12. Here, we observe resid-

uals of the order of 30 nm at maximum. Although some inter-dependence in the three directions, which reveals itself whena deliberate direction is calibrated using the calibration resultsfrom three perpendicular directions, still has to be figured out,it is evident that we have achieved a unique probe with unprece-dented low uncertainties.

V. CONCLUSIONS

Using MST-technology, we have developed a 3-D probewith a resolution in the nanometer-range and an uncertainty inthe range of some 10 nm. For a 3-D probe, this is better thanwhat has been achieved so far. This is due to the applicationof MST-technology, proper mechanical design, and keepingproper metrological principles. This result will encouragefurther use of MST techniques in dimensional metrology.

REFERENCES

[1] N. Taniguchi, “Current status in and future trends of ultraprecision ma-chining and ultra fine materials processing,”Ann. CIRP, vol. 32/2, pp.573–580, 1993.

[2] P. Schellekens, N. Rosielle, H. Vermeulen, M. Vermeulen, S. Wetzels,and W. Pril, “Design for precision: Current status and trends,”KeynoteCIRP, pp. 557–586, 1998.

[3] M. Vermeulen, Ph.D. dissertation, Eindhoven Univ. Technol., Eind-hoven, The Netherlands, 1999.

[4] M. Visscher and K. G. Struik, “Optical profilometry and its applicationto mechanically inaccessible surfaces—Part I: Principles of focus errordetection,”Precision Eng., vol. 16, no. 3, pp. 192–198, 1994.

Han Haitjema received the M.Sc. degree in experimental physics from theUtrecht University, Utrecht, The Netherlands, in 1985, and the Ph.D. degreefrom the Delft University of Technology, Delft, The Netherlands, in 1989. Histhesis was on “Spectrally Selective Tinoxide and Indiumoxide Coatings.”

From 1989 to 1996, he was with NMi-Van Swinden Laboratorium, wherehe was involved in dimensional metrology. He is now a Lecturer in the Preci-sion Metrology Group, Eindhoven University of Technology, Eindhoven, TheNetherlands, covering the field of dimensional metrology.

Wouter Pril received the M.S. degree in physics from the University of Ni-jmegen, Nijmegen, The Netherlands, in 1994, and is currently pursuing thePh.D. degree at Eindhoven University of Technology.

He is now with ASM-Litography, Veldhoven, The Netherlands.

P. H. J. Schellekensreceived the M.S. degree in physics and the Ph.D. degreein the technical sciences from Eindhoven Technical University, Eindhoven, TheNetherlands, in 1978 and 1986, respectively.

Currently, he is the Leader of the Precision Engineering Group, EindhovenUniversity of Technology.

Dr. Schellekens is a member of many technical institutions, such as VDI,ASPE, CIRP, EUSPEN, and is Secretary of CIRP-STC-P.