Optical Methods for Dimensional Metrology in Production Engineering

Heinrich Schwenke’, Ulrich Neuschaefer-Rube’, Tilo Pfeifer’ (2), Horst Kunzmann’ (1)
1
Physikalisch-Technische Bundesanstalt, Braunschweig, Germany
2
RWTH Aachen, Germany

Abstract
Metrology in production engineering must be fast, accurate, robust and automated, and ideally integrated
into the production line. In many respects, optical methods seem to fulfil these requirements. Although opti-
cal methods have a long tradition in dimensional metrology, the rapid progress in the development of opto-
electronic components and availability of increased computational power makes many new technical ap-
proaches possible. This paper provides a technical overview of the optical methods available for dimen-
sional metrology. Methods for the measurement of length, angle, surface form and spatial co-ordinates are
described. The paper summarises both the metrological characteristics and the technical limitations of the
methods. Furthermore, it presents some new and promising approaches that, in the future, may play an im-
portant role in dimensional metrology for production.

Keywords
Dimensional Metrology, Optical Methods, Production Engineering

Acknowledgements (if the distance is known). By combining measurements

The authors gratefully acknowledge the following, who from different distances, even more than one unknown
have contributed to this paper: length can be determined.
Graham Peg s (1) Important milestones in the application of optics for di-
Gerd Dussle
Jens Flugge’
7 mensional measurement are: the invention of the micro-
scope (Leeuwenhoek 1660), the utilization of light inter-
Klaus Wendt’ ference (Michelson 1892) and the development of gas
Hans Danzebrink’ lasers (Maiman 1960), which replaced other light sources
Markus Bartschec (e.g. mercury vapour lamps) for most interferometry appli-
Ahmed Abou-Zeid’ cations from the 1970s on. Later, the rapid growth of
computational power and the wide application of opto-
1 INTRODUCTION electronic components for consumer products increased
the speed of development of optical metrology [1,2]. In
From its early beginnings dimensional metrology was many cases components of the consumer world are now
closely related to the use of light for measurement. An used for metrology purposes, e.g. CCD sensors (video
early example is the so-called ‘Jacob Bar’ for taking opti- cameras), laser diodes (CD players) and Digital Micro
cal bearings in the fields of civil engineering and naviga- Mirror Devices (video projectors). Photodiodes and CCD
tion (Figure 1). Based on the intercept theorem, such an sensors convert light intensities to electrical signals and
instrument can either determine the distance from an perform the data acquisition for computer systems, which
object (if its lateral dimension is known) or its dimension often execute sophisticated data evaluation. In many
cases, the need for expensive precision mechanics is
reduced or eliminated by optical methods.
A changing production environment with permanently
growing demands on quality, flexibility and speed chal-
lenges traditional metrology. Even for mass products such
as hard disk components or fuel injection systems, the
dimensional precision required for the functional require-
ments come close to the limits for production. This re-
quires processes to be under rigorous metrological con-
trol, both in sense of completeness and accuracy. Com-
pared to mechanical systems, optical methods often can
acquire more data in less time (e.g. CMM versus fringe
projection). Another important advantage is the possibility
to measure without contacting the part. These character-
istics qualify optical methods especially for in-process
measurement. For the next decade, a study of the Opto-
Electronics Industry Development Association projects
Figure 1: The ‘Jacob Bar’ for the determination of dis- rapid growth for the market for opto-electronic sensing
tances and lateral dimensions (14th century). and metrology (Figure 2).
 principle applied, surfaces can be entitled co-operative or
non-co-operative for a specific metrology.

2.2 Wave optics

The wave characteristic of light is made use of by another
category of optical methods. Direct frequency measure-
ment of the electromagnetic field of light (4.1014 Hz to
8.1014 Hz for visible light) is still impossible. Therefore, all
existing methods employ the superposition of light of the
same, or of very similar, wavelengths to detect either
static intensities or changes in intensity in a measurable
frequency range. In addition to the components used for
geometrical optics, instruments based on wave optics
incorporate polarisers, optical retarders, diffraction grat-
ings, or holographic elements. Precision measurements
often have the advantage of the high linearity and the high
60s 70s 80s 90s resolution of interferometric measurements. One impor-
tant characteristic of interferometry is that the traceability
Figure 2: (a) Projected worldwide market for optoelec- to the SI unit can be directly achieved by frequency com-
tronic sensing equipment [ I ] , (b) Number of patents in the parison [5].
field of optical metrology worldwide [4].
2.3 Quantum optics
This paper focuses on the technical applications of optical
The main application of light in dimensional metrology for
methods for dimensional measurement in production
which the quantum theory has to be considered is the
engineering. It does not cover the great number of optical
generation of laser light [6]. The triggered transition to
methods for sensing mechanical stress, temperature,
lower electron energy levels produces light quantum of
chemical composition etc. Neither will this paper deal with
defined energy and phase, which can be used as mono-
metrology in astronomy, remote sensing, geodesy and
chromatic light of high intensity and coherence length for
civil engineering, although it should be noted that many of
many purposes in dimensional metrology. The laser has
the principles described can also be applied in these
revolutionized optics for metrology and is applied in many
areas [3]. In the following sections, the basic optical prin-
of the methods described in this paper. Only for a few
ciples of dimensional metrology are described and techni-
methods quantum effects have to be considered when
cal solutions are reviewed. Special emphasis will be
analysing the measurement process, e.g. for scanning
placed on their application, their metrological characteris-
near field microscopy and sub-nanometre interferometry.
tics and their present technical limits. Furthermore, an
outlook on promising developments is given, which might
expand the application of optical methods or overcome 3 LENGTH MEASUREMENT
today's limitation of some techniques.
3.1 Linear encoders with optical grating scales
2 PRINCIPLES AND CLASSIFICATION The most common length measuring devices for high-
precision applications are linear encoders based on opti-
2.1 Geometrical optics cal grating scales. In comparison with laser interferome-
ters (see section 3.2), linear encoders are less expensive
Geometrical optics is based on the linear propagation of
and more robust even under unstable environmental
light and the laws of refraction and reflection. Typical
conditions. In production technology they are incorporated
examples of the application of geometrical optics for di-
e.g. in transducers, machine tools, or co-ordinate meas-
mensional measurement are microscopy, photogramme-
uring machines. Gratings are typically applied on a steel
try, autocollimators and laser straightness measurement.
or glass substrate. For the conversion of the grating dis-
Typical elements of a measuring instrument based on
placement relative to the read head into an electrical
geometrical optics are light sources (lamps, diodes, la-
signal different types of optical encoders have been de-
sers), lenses, irises, mirrors, prisms, beam splitters, filters
veloped. Rather simple systems are based on the geo-
and opto-electronic components like photodiodes, CCDs
metrical effect of phase-shifted optical relays. For finer
and Digital Micro Mirror Devices (DMD). In most cases,
grating periods, graduations of an appropriate structure
magnification lenses are used as an 'optical gearbox' to
are used to evaluate phase shifts in interference patterns.
transform the scale of measurement. For measurements
Figure 3 illustrates this principle [7].
on the microscopic scale, a physical limit for lateral reso-
lution of microscopes in air is given by approximately half Today, grating pitches range from 40 pm to 0.5 pm, but
the employed wavelength due to diffraction (Ernst Abbe the resolution of such systems is generally increased by
1873). The wavelength of visible light ranges from 400 nm interpolation by a factor of 10 up to 4000. Scales com-
to 800 nm, which implies in a resolution not below 200 mercially available have resolutions down to 1 nm [8]. For
nm. One way to bypass this limit is to use electromagnetic traceable length measurements the scales must be cali-
radiation of shorter wavelengths like ultraviolet light (400 brated over their whole length. Most grating scales are
nm - 3 nm) or X-rays (60 nm - 1 fm). Typical error incorporated into machine tools. Due to bending during
sources for measurements based on geometrical optics mounting the scale assembly, and deviations of the ma-
are lens distortion, spherical aberration, astigmatism and chines. In practice, it is preferable to calibrate the whole
coma on the optical side and non-linearities and geomet- machine and not just the integrated scales. The machine
rical errors of the opto-electronic converters. For meas- calibration can be carried out with laser interferometers or
urements that directly probe the surface of a specimen, a calibrated grating scale system mounted in an appropri-
often the surface characteristic itself dominantly contri b- ate way. The stand-alone calibration of scales is usually
utes to the uncertainty of measurement. Depending in the achieved by comparing the output of the grating scale
 two additional signals with opposite phase are used to
correct for changes in optical intensity. As regards the
heterodyne principle, the frequencies of the light in the
measurement and reference beams are slightly different
and the beat frequency of the interference signal is de-
tected. The phase of the beat frequency changes with the
motion of the mirrors and can be compared with a fixed
reference frequency. The main advantage of the hetero-
dyne principle is the better signal-to-noise ratio, which
allows one HeNe laser to be used for more than ten axes
simultaneously, e.g. for stepperkcanner in semiconductor
production (Figure 4).

Figure 3: Principle of the diffraction grating encoder [8]

system to that of a laser interferometer. Today, calibration

facilities can achieve measurement uncertainties for the
calibration of a 400 mm scale in the order of U = 25 nm
(k=2), but recent developments are targeting U = 5 nm
(k=2) for the same length [9].
Especially in an industrial environment, the thermal ex-
pansion of the scale, the machine structure and the work-
piece dominantly contribute to the measurement uncer-
tainty. To reduce these contributions, different strategies
can be applied: temperature measurement and mathe-
matical compensation of the thermal expansion, use of
materials with a very low expansion coefficients (e.g.
Zerodur@),or the use of scale and workpiece materials
with similar thermal expansion coefficients. Other sources Figure 4: Heterodyne interferometer system for the
positional control of multiple axes in a stepper for semi-
of uncertainty are short periodic errors due to interpolation
conductor manufacture [14].
and the long-term stability of the scale material [lo]. Un-
der good laboratory conditions, relative measurement
uncertainties down to 5.10-7 can be achieved, but in the For applications in a non-laboratory environment, the
production environment the uncertainties may be much refractive index of air is the main source of uncertainty
greater. when interferometers are used. The refractive index can
be determined by measuring temperature, pressure and
3.2 Displacement Interferometry humidity using Edlen's equation [15,16] or by direct in-
terferometric measurements in a refractometer [17]. In
For displacement measurement applications with high both cases the refractive index can only be determined
accuracy requirements, optical interferometers are often outside of the measurement beam so that the measure-
used. Typical applications are the error analysis of preci- ment accuracy strongly depends on temperature gradi-
sion axes in machine tools and co-ordinate measuring ents and air turbulences [18]. In a similar way to grating
machines, and the use as control axes for very high pre- encoders, non-linearities occur in the interpolation proc-
cision machines [ I I ] . ess. The main sources of interpolation errors are different
The principle of length interferometry is based on the work for homodyne and heterodyne interferometers but in both
of Michelson and BenBit some 100 years ago [12]. But the cases they can be minimised to less than 10 nm using
quality of the light sources, the optical components and high-quality optical components and can be further re-
especially the opto-electronic components available have duced by compensation methods [ I g]. Increasingly appli-
made interferometry a very powerful tool in length metrol- cations occur to which even sub-nanometre accuracy is
ogy even for industrial applications. Until the late ~ O ' S , demanded as, for example, in the next-generation semi-
interferometric measurements used krypton or cadmium conductor production systems. In this case, many other
isotope lamps as a light source. Later, the invention of the uncertainty contributors like diffraction effects, photon
helium-neon laser with a wavelength of 633 nm provided bounce, ghost reflections and quantum effects have to be
a light source with an improved coherence length and considered [ I 31. To quantify non-linearities of optical
intensity and effective frequency stabilisation. Laser in- interferometers, X-ray interferometers have been used
terferometers realise the scale of length by a well-defined [13]. They supply a scale with a very high resolution in the
wavelength that can be linked to the definition of the me- range of 10s of picometres.
ter by frequency comparison [5]. Two families of interfer-
ometers for displacement measurement can be distin- 3.3 Triangulation distance sensors
guished by the principle of signal detection [13]: the ho- Triangulation sensors have become frequently employed
modyne and the heterodyne interferometer. In the homo- for in-process metrology and co-ordinate metrology, es-
dyne interferometer only one wavelength is employed. pecially for the measurement of car bodies in the automo-
Two waves phase-shifted by 90" are generated by polari- tive industry. The main components of a triangulation
sation optics to determine the direction of motion. Usually, sensor are a collimated light source (generally a laser
diode) and a detector unit consisting of an imaging lens the position of the surface relative to the focal point, the
and a position-sensitive detector (CCD line or position- outer or inner segments of the focus detector are illumi-
sensitive diode (PSD)). The optical axes of the light nated. Signal processing produces an error signal inde-
source and the imaging lens form a fixed angle, the so- pendent of the reflected intensity. Either the sensor or the
called triangulation angle. The object surface is brought objective is shifted in the direction of the optical axis in
close to the point in which both axes intersect and the closed loop control until the surface is in focus. Then the
diffuse reflection of the light spot on the workpiece surface position of the sensor relative to the surface can be de-
is imaged onto the detector. To achieve sharp imaging, an termined. The specimen is mounted on a translation stage
inclined detector arrangement is necessary (Scheimpflug to determine the surface profile. The measurement
condition). The position of the image on the detector is a ranges of typical autofocus sensors are small (up to
function of the distance between sensor and specimen Q50 pm) compared to the triangulation sensor. On the
(Figure 5). Typical measurement ranges of triangulation other hand, autofocus sensors can provide a significantly
sensors are 2 mm to 200 mm; they provide relative reso- higher accuracy. On co-operative surfaces, relative re-
lutions down to The main uncertainty contributor in peatabilities in the order of to of the measure-
most applications is the optical characteristic of the work- ment range can be achieved. The typical diameter of the
piece surface, for example very smooth surfaces cannot measurement spot (-2 pm) is small compared to conven-
be measured because of insufficient diffusely reflected tional surface profilers. When sharp edges are measured
light. Controlling the laser intensity and the sensitivity of directly or when the reflectance of the surface changes
the detector can moderate this effect. Errors are also significantly within the measured profile, an autofocus
induced by: the slope of the surface (which may produce sensor can produce large errors. The tolerable surface
direct reflections to the detector), volume scattering (e.g. slope is limited by the aperture of the focusing lens to
for plastic material), or an inhomogeneous surface tex- approximately 45 degrees.
ture. Work is described in reference [20] which relates to
a detailed analysis of triangulation sensors on different
surfaces. It was shown, that surprisingly large errors can
be observed on non-co-operative surfaces. Triangulation
sensors using CCD-detectors generally yield a better
accuracy because the intensity profile of the image can be
determined. On the other hand, for this sensor type, the
influence of ambient light cannot be eliminated by a
modulation of the laser intensity. Furthermore, sensors
with CCD detectors are more expensive and their meas-
urement bandwidth is smaller.

Figure 6: Principle of the autofocus sensor [22]

Recent arrangements [23] use a focus detector consisting

of an astigmatic lens system and a four-quadrant diode.
By this symmetrical configuration, the problems men-
tioned previously are reduced.
Figure 5: Principle of the triangulation sensor [21].
3.5 Recent developments
An extension of the triangulation principle is known as
light sectioning (for example as described in reference Multi-wavelength interferometry
[21]). Here a line is projected onto the specimen instead Using light of a single wavelength in an interferometer
of a single spot. A CCD array detects the image of the system limits the range of unambiguousness of length
line. Measured co-ordinates are generated from the im- measurement to only half the optical wavelength. There-
age of the line deformed by the surface topography. fore, single-wavelength interferometers like incremental
scales only can measure translations and not absolute
3.4 Autofocus sensors distances. However, when light with two wavelengths is
The principle of the autofocus sensor is similar to that of used simultaneously, the range of unambiguousness can
the triangulation sensor, but instead of collimated light be extended to half a synthetic wavelength. Only if the
being projected the light is focused onto the surface of the interpolation error of the synthetic wavelength is smaller
specimen. The reflected light is directed to a focus de- than the range of unambiguousness of a single wave-
tector by a special beam splitter (Figure 6). Depending on length, can the resolution of a single-wavelength interfer-
ometer be maintained. Additional wavelengths can be
used to further increase the range of unambiguousness. Nevertheless diode lasers are an interesting alternative to
This means that for typical applications in mechanical gas lasers in diverse applications:
engineering at least three, frequency-stabilised light 0 For measurement of angle, refractive index of air [30],
sources are needed. An alternative is to use a tuneable surface topography and length of incremental scales
laser, which allow interferometric measurements at differ- and tapes, the frequency stability of passively stabilised
ent wavelengths. An example for high-precision absolute diode laser is often sufficient [26].
interferometry is the optical calibration of gauge blocks
0 Due to spectral variety of diode lasers (400nm -
~41.
1500 nm) and their good tunability, diode lasers are an
Another application for two-wavelength interferometry is unique light source in multiple-wavelength interferome-
the compensation of fluctuations in the refractive index of try and phase shift interferometry [31].
air [25]. Two wavelengths with a large frequency offset
can be generated e.g. with the first (1064 nm) and second Applications of femtosecond lasers
harmonic (532 nm) of a YAG laser by a frequency- Femtosecond lasers produce ultra-short light pulses of
doubling crystal. The difference of the refractive indices at some femtoseconds s). In recent years, the femto-
these two wavelengths produces a phase difference be- second laser technology has developed significantly and
tween the two signals from the interferometer. With the some researchers deal with applications in dimensional
aid of Edlen's equation, the refractive index can be cal- metrology. Two main research areas for metrology appli-
culated from this dispersion and corrected. The differential cations have been identified by the authors: the genera-
resolution of the system must be extremely high for the tion of frequency combs for the direct comparison of fre-
detection of the dispersion of the two frequencies to a quencies [32] and the use of ultra-short light pulses for
degree that delivers the effective refractive index with time-of-flight measurements [33]. The first approach uses
sufficient accuracy. Nevertheless, this technique has the the equally spaced comb of frequencies in a femtosecond
potential to overcome the main problem of precision in- pulse to compare optical frequencies to one another or a
terferometry in a production environment by correction for frequency standard. For the national metrology institutes,
the fluctuation of the refractive index in the beam path. this facilitates the traceability of wavelength standards to
Use of laser diodes for interferometric measurement the caesium clock and reduces the uncertainties associ-
ated with frequency splitting. In the second field of re-
Laser diodes have been a promising light source for in- search, ultra-short pulses and non-linear optics are used
terferometry since they emerged about 40 years ago. for the time-of-flight measurement. Within a femtosecond,
They have some obvious advantages compared with gas
light propagates -300 nm in space. By measuring short
lasers with regard to their size, price, life time, efficiency time intervals, short distances can also be resolved. Al-
and optical output power. On the other hand, the wave- though some results have already been presented [33],
len th stability is significantly worse than a gas laser (-
9
10- for an uncontrolled diode) and depends strongly on
these techniques are still far from industrial application. It
should be mentioned, that this technique has the general
temperature and injection current [26]. Further disadvan- potential to directly realise the definition of the SI unit of
-
tages are the spectral ageing of the diode laser of 3.10-5 length: the meter is defined by the length of the path trav-
per year for GaAlAs diodes [26] and the limited beam elled by light in vacuum during a time interval of 1/299
quality. These disadvantages can be compensated by 792 458 of a second [5].
suitable wavelength stabilising methods and special op-
tics. Conoscopic holography
The emission frequency of a diode laser can be passively
or actively stabilised. The passive stabilisation can be Conoscopic holography is an upcoming measurement
realised by controlling of the diode's parameters by principle in the field of 3D measurement. The measure-
means of variations in diode current and/or temperature. ment principle must be thoroughly separated from classi-
So a short-term stability of can be achieved by stabi- cal holography. The sensor contains a laser diode that
lisation of the temperature within 1 mK and the current illuminates the specimen by a quasi-monochromatic light
within 10 PA [26]. Although this is a rather efficient and beam. The light emitted at the spot on the surface passes
practical solution, it does not encompass long-term drift of through a lens and a circular polariser (Figure 7),
the diode. By using an external resonator the long-term
stability and the coherence length can be significantly
enhanced [27]. The active stabilisation of the diode laser
frequency, i.e. spectral ageing-independent, can be real-
ised by coupling the diode laser direct to an external ref-
erence such as a material standard (long-term stability of
- [28], Fabry-Perot Etalon (> lo-@) [26], or an atomic
(Cs, Rb) or a molecular (12, H20) absorption line (c
~91.
The development of laser diodes is mainly stimulated by
mass production applications, e.g. for consumer products
and fibre-optic communication. For these applications the
main criteria are power, reliability and price. In the experi-
ence of the authors the ultimate frequency stability has
not been of major interest so far. This has impeded the
progress of diode laser interferometry over the last few
decades. One of the visions for the application of laser
diodes is the production of miniaturised low-priced inter-
ferometers on a chip for industrial applications. Such
interferometers could become a serious competitor for
high-resolution scales for length metrology and machine
control. Figure 7: Principle of conoscopic holography [34]
then the circularly polarised light passes through a uniax- sical autocollimators in production engineering are the
ial crystal (e.g. calcite). The crystal axis is oriented paral- measurement of angular errors of rotary axes and pitch,
lel to the optical axis of the lens. As a result, the incident yaw and straightness errors of translation axes. The prin-
light hitting the crystal at an angle is split into an ordinary ciple of the autocollimator is the imaging of a target in the
and an extraordinary ray. The ordinary refractive index is focal plane by use of a parallel beam reflected on a flat
constant, while the extraordinary refractive index - and so mirror. The shift of the target in the imaging plane solely
the phase delay between the rays - is a function of the depends on the tilt of the flat mirror and not on the dis-
direction of propagation relative to the crystal axis. In a tance between mirror and lens [37]. This shift can be
first-order approximation, the different refraction angles of determined by a visual method or converted into an elec-
the rays can be neglected because the difference be- trical signal by a lateral photodiode or a CCD array. Mod-
tween the refractive indices is small. The two rays inter- ern opto-electronic autocollimators reach accuracies up to
fere after having traversed the circular analyser (a second 0.01 arc second [38]. The measurement range can be up
circular polariser) and form interference rings. This pattern to some 1000 arc seconds.
is detected by a CCD array. The distance between the
rings depends on the distance between sensor and 4.3 Length difference
specimen [35]. Typical measurement ranges of cono-
For the measurement of small angles, e.g. for the meas-
scopic sensors are between 0.6 mm and 70 mm. The
urement of rotation errors on a linear axis, a differential
relative accuracy is in the order of and the repeat- interferometer can be used. In these interferometers, the
ability in the order of l o 4 of the measurement range. The laser beam is split again into two beams directed to retro-
manufacturer specifies that sharp surface slopes (up to 85 reflectors that have an offset with respect to each other
degrees) may be measured. An improved version of the perpendicular to the axis of rotation measured [39]. The
sensor measures the distance between sensor and reflected beams interfere after recombination. The de-
specimen along a line max. 24 mm on length in one step. tected signal depends only on the path difference of the
The maximum data rate of this sensor is 15000 beams, which is a function of the tilt angle. Due to the
pointskecond [34]. Since the principle is rather new, no small path difference of the beams in a differential set-up,
experimental verification of this sensor has been found in no special care has to be taken as regards the frequency
the literature up to now. stability of the laser source.
Angle measurements with high lateral-resolution on a
4 ANGLE AND STRAIGHTNESS MEASUREMENT surface are achieved using a Nomarsky interferometer
(Figure 8). In this interferometer, the two measurement
4.1 Optical encoders spots separated by a distance of approximately 2 pm are
generated by a Wollaston prism and an objective.
Optical encoders for highly-accurate angle measurement
are operated in many technical systems such as robots,
machine tools, rotary tables, laser trackers and tele-
scopes. Their functional principle is identical to that of
linear encoders. In the same way as for linear axes, a
circular, radially oriented optical grating (e.g. chromium) is
applied to a substrate (e.g. glass), and at least one optical
head reads a sinusoidal intensity signal that is generated
by the grid on the basis of geometrical or interferential
optics (3.2). Generally speaking, precision encoders
measure incrementally. Commercially-available encoders
can have a resolution as small as 0.025 arc second [8].
For the highest-precision applications like the guidance
and control systems for astronomical telescopes or rotary
tables for high-accuracy CMMs, in some cases up to four
reading heads per encoder are used. By averaging the
signals from the reading heads, periodic errors of the
divided circle can be reduced. Interpolation errors due
distortions of the sinusoidal signal can be eliminated by
electronic correction.
Typical uncertainty contributors in angle measurement for
technical applications are: the non-linearity of the encoder
(short and long periods), errors due to the eccentricity in
Figure 8: Principle of the Nomarsky interferometer [40]
relation to the rotary axis, errors in the transmission be-
tween axis and encoder, friction and thermal influences.
The calibration of angular encoders can be carried out at Commercial Nomarsky interferometers reach a lateral
national metrology institutes with an uncertainty of U=0.05 resolution of 0.5 pm. Path differences down to 0.1 nm can
arc second (k=2) [36]. The calibration of angles can be be measured. To determine the sign of the difference, two
performed without reference to a primary standard. The detector signals with a phase difference of 180 degrees
angle of 360 degrees (and its fractions) is intrinsically are generated. The Nomarsky interferometer is applied
defined by a full revolution. Rotary encoders are cali- e.g. in semiconductor industry to measure the surface
brated using either a reference rotary table [36] or a mirror topography of wafers.
polygon and an autocollimator.
4.4 Laser gyroscopes
4.2 Autocollimator
Laser gyroscopes are based on the Sagnac effect. When
A classical instrument for measuring small angles with light propagates along a circular path in both the clock-
high precision is the autocollimator. Applications for clas- wise and in the counter clockvvlse directions, a rotational
movement of the system relative to the inertial space can
 SY
be detected in an apparent path difference between the
two directions. The value of the offset depends linearly on
the angular velocity and can be measured by interfe-
rometry [41]. The absolute angle is determined by inte-
gration of the velocity. There are two technical realisations
of laser gyroscopes: active laser gyros (ring lasers) and
fibre-optic laser gyros. In general, active laser gyros per-
form better.
Laser gyroscopes are used primarily when no external
angular reference is available, especially for 'strap-down'
navigation systems. There are limited applications in
production engineering. For dynamic calibration of optical
polygons and encoders a system based on a ring laser
has been developed, the performance of which agrees
with reference methods within 0.1 arc second [36]. This
system has the advantage that ring lasers incorporate an
almost perfect circular division based on the laser wave-
length. Figure 9: Directional laser straightness measurement by
detection of a laser beam with a CCD array.
A particular disadvantage of laser gyros in practical appli-
cations is the occurrence of the 'lock-in' effect which
arises from imperfections in the optical components. 5 SURFACE MEASUREMENT
Cross-talk between light travelling in both directions can 5.1 Fringe interferometry
synchronise both wavelengths if the angular velocity is
In contrast to the interferometric methods described
small. Therefore quasistatic velocities cannot be meas-
above, in fringe interferometry, a complete surface section
ured. In navigation, this problem is solved by 'dithering'
is illuminated by the measurement beam and the interfer-
the gyro, i.e. introducing a mechanical oscillation about
ence pattern is detected by a CCD camera. The surface
the axis normal to the plane of the ring.
topography is determined with nanometre accuracy using
4.5 Straightness measurement fringe counting and interpolation methods. Optical con-
figurations which are often used are the Michelson inter-
The measurement of straightness deviations is a classical ferometer (Figure l o ) , the Mach-Zehnder interferometer
task of the quality assurance for manufacturing proc- and the Twyman-Green interferometer [39].
esses. In optical straightness metrology a distinction is
made between the inclination method and the height
method. The inclination method is based on the tilting or
inclination of a measuring slide along an axis due to its
deviation from straightness. By angle measurement and
its integration along the travel of the axis (e.g. by an auto-
collimator, 4.2), the straightness of the slideway can be
evaluated. The height method serves to directly measure
the two-dimensional position of the slide. For this pur-
pose, a laser beam is directed parallel to the surface
tested and the mirror on the measuring slide is replaced
by a sensor capable of measuring the lateral and trans-
verse positions of a laser spot. This procedure is som-
times known as the directional laser beam method [42]
(Figure 9). Another height technique is laser straightness
interferometry based on the Doppler shift of two super-
posed laser beams [39]. By the application of a Wollaston
prism a laser beam can be split into two perpendicularly
polarised beams that diverge at a specific angle. The two
beams are reflected by an angular mirror and reunited in
the Wollaston prism mounted on the measuring slide. Any Figure 10: Michelson interferometer and detected inten-
movement of the prism perpendicular to the tested sur- sity distributions [21].
face results in a Doppler frequency shift measured by a (a) lnterferogram of an optical smooth spherical surface.
photodetector, which is a measure for the straightness (b) lnterferogram of a rough surface (speckles).
deviation. (c) Difference between two interferograms of a rough
spherical surface, determined at two wavelengths.
Modern laser-based measurement systems are capable
of detecting straightness deviations of a few micrometres
up to distances of 20 m [43]. Accuracy and robustness of A clear interference pattern (Figure 10a) can only be
straightness measurement systems depend strongly on observed on a smooth surface. On rough surfaces
the particular application. For some technical applications, (R, > U4), the interference fringes are destroyed by
measurement uncertainties of < 5 pm are required. For speckle effects (Figure lob). There are important applica-
these applications, the influence of surrounding light and tions, e.g. in the optical industry [44,45], to which the
refraction effects caused by air turbulence on optical limitation of speckle has no relevance, but unfortunately
straightness measurement systems is one of the major these methods cannot be directly applied to most techni-
problems. An example of the application of laser straight- cal surfaces. Nevertheless, a way to measure rough sur-
ness measurement systems is the in-process compensa- faces is to perform two measurements at different light
tion of processing forces in a high-precision turning ma- wavelengths [46]. The difference between the two result-
chine [43].
ing speckle patterns again contains interference fringes and are suitable for subsequent image processing. For
(speckle interferometry, Figure 1Oc). sampling of the workpiece surface, different techniques
Another interferometric set-up often used in surface have been developed: the lateral displacement of the
measurement is the Mireau interferometer (Figure 11). specimen, pinholes on a rotating disk (the so-called Nip-
Here the measurement beam and the reference beam are kov disk), scanning mirrors, arrays of microlenses and,
not separated geometrically. As a result, a very compact recently, the Digital Micro-Mirror Device (DMD) [53].
set-up can be realised. Mireau interferometers are often
combined with microscope objectives. It is of advantage
that the light paths in the interferometer are nearly the
same length. This feature reduces the required coherence
length of the light source. Mireau interferometers are
therefore suitable for white-light interferometry (see sec-
tion 5.3).

Light Source and Camera

Figure 12: Principle of confocal microscopy [22].

5.3 W hite-light interferometry

In contrast to laser interferometry, white-light interferome-
try uses light with a very short coherence length. Most
Figure 11: Principle of the Mireau interferometer [47]. applications are based on a Michelson or a Mireau inter-
ferometer. When the specimen is moved in the direction
of the optical axis (z direction), the intensity shown in
5.2 Confocal microscopy Figure 13 is observed by the interferometer detector.
Another measurement technique is confocal optical mi-
croscopy [48,49] - this has several advantages over con-
ventional optical microscopes. The most important benefit
is the fact that confocal images do not show out-of-focus
blurs, so that it is possible to generate three-dimensional
images of objects by means of optical sectioning [50,51].
It allows profiling of surfaces, 3D objects and multi-layer
structures, such as integrated circuits deposited on sili-
con, by a non-contacting and non-destructive method.
In conventional epi-illumination light microscopy, the si-
multaneous illumination of the entire field of view of a
specimen will charge reflections throughout the depth of
the specimen, rather than just the focal plane. Much of the Figure 13: Intensity observed in a white-light interferome-
light collected by the objective lens to form the image will ter during z-translation [21].
thus come from regions above and below the focal plane
selected. This light will contribute to the final image as Because of the short coherence length Ic of white light
out-of-focus blur. Also, this blur deteriorates the image (approximately 3 pm), an intensity modulation due to
quality by reducing contrast and sharpness. interference effects (coherent superposition) occurs only
In confocal microscopy, the illumination of the specimen is close to the position Z O . At zo (the so-called reference
restricted to the focal plane, which is scanned to produce plane), the lengths of measurement and reference path
a complete image. Pinholes prevent the light emanating are identical. Here the envelope of the intensity reaches
from regions above and below the focal plane from con- its maximum value. To determine the surface height with
tributing to the image observed. This enables the confocal a white-light interferometer, the reference mirror or the
microscope to achieve high axial resolution. It permits its complete interferometer is displaced continuously. The
use for serial non-invasive optical sectioning and the surface height is concluded from the maximum of the
acquisition of 3D image data. Figure 12 illustrates the envelope. To measure a complete surface section, a
principle of confocal microscopy. Confocal microscopes series of CCD images is evaluated. By determination of
have further benefits over standard optical light micro- the maximum of the envelope pixel by pixel, the topogra-
scopes, including an improvement in lateral resolution phy can be reconstructed. Commercially available white-
[52]. They also eliminate stray light from the optical sys- light interferometers reach resolutions of up to 0.1 nm.
tem itself and thus increasing the contrast and improving The measurement range is small (< 100 pm). So, the
the signal-to-noise ratio of the final image. Confocal mi- main area of application for this method is micro- and
croscopy is also capable of generating high-resolution nanotechnology. The lateral resolution (0,6pm - 10 pm)
digitised data sets for the 3D objects that can be stored depends on the field of measurement.
5.4 Recent developments The collimated beam of a He-Ne laser is split by a circu-
larly symmetric diffractive element, the beam splitting
Holographic form testing diffractive axicon. The undiffracted beam serves as the
Since the invention of holography in 1948 by Dennis Ga- reference beam of the interferometer. The first-order
bor [54] holograms have become a tool used in optical diffracted beam hits the workpiece surface. The angle of
metrology. Due to their capability of recording not only the incidence is small (approximately 5 degrees). After reflec-
amplitude but also the phase of a light wave, holograms tion on the workpiece a second axicon recombines the
can be regarded as one of the most flexible optical tools. beams and the interference pattern is observed. To re-
They are used in several areas of application such as duce speckle effects, the beam passes a rotating frosted
optical imaging, beam deflection, beam splitting, beam glass panel before it is detected by the CCD-camera [57].
shaping and wavefront reconstruction. This panel causes a changing speckle pattern, which is
averaged during the integration time of the camera.
Holograms play an important role in interferometric form
testing as far as complex surfaces such as aspheres are The device shown in Figure 14 can measure workpieces
concerned [44]. For testing a strongly aspherical optical with diameters between approx. 3 mm and 25 mm. The
surface, a hologram can be used as a reference to pro- maximum cylinder length is approx. 125 mm. The accu-
vide the reference wavefront and thus to perform a null racy achieved with this device ?0,05 pm, the specified
test. For this purpose, synthetic or computer-generated reproducibility is ?0,02 pm. The measurement of a com-
holograms are used. A computer-generated hologram plete workpiece takes approximately 1 minute. Due the
consists of a planar substrate covered with a diffractive small angle of incidence, speckle effects caused by sur-
microstructure. This microstructure has to be calculated face roughness are significantly reduced, because the
numerically and is produced by lithography. Due to the effective light wavelength is increased by approximately a
recent progress in lithography, it is possible to produce factor 10. Therefore, workpieces with an R, up to 0.8 pm
these diffractive microstructures with very high accuracy can still be measured [56]. The circularly symmetric dif-
[55]. The phase of the reference wavefront, which is a fracting axicions serve as references containing the ideal
crucial parameter for interferometry, is determined by the cylindrical form. Errors in the cylindrical gratings cause
diffractive structures of the hologram. The accuracy of errors in the measurement result. Special diffraction grat-
interferometers using holograms as references is there- ings must be used, when objects of other than cylindrical
fore mainly influenced by the accuracy of the lithographic shapes are measured [58]. Errors in the alignment be-
manufacturing process. tween the axicons and the workpiece are reduced by a
Holographic measurement techniques can also be applied post-process least-squares fitting of the measured data
for testing objects with diffusely reflecting rough surfaces. [58]. The small angle of incidence limits the measurable
Since a hologram provides the complete optical informa- workpieces. Moreover, recessed and concave surfaces
tion about the current condition of an object, this condition often cannot be measured. Furthermore, the set-up
can be interferometrically compared using holographic shown in Figure 14 is not suited to the measurement of
interferometry techniques [39]. So, contours and dis- blind holes, but special set-ups for this purpose are under
placements of a technical object subjected to stress can development [56].
be measured with interferometric precision. With the help Scanning near-field optical microscopy
of computers and CCD cameras the comparison between
the different conditions of an object can be carried out As mentioned in clause 2.1, the resolution of conventional
digitally directly in the computer. optical microscopy is limited by diffraction to approxi-
mately half the wavelength used. Near-field scanning
An application of holographic form testing particularly optical microscopy (NSOM) can overcome this limit. The
suited to measure cylindrical parts with rough surfaces is basic principle of NSOM is to scan a sub-microscopic
grazing incidence interferometry [56]. Important applica- radiation source (or, alternatively, a sub-microscopic
tion fields of this method are the measurement of fuel radiation detector) as so-called near-field probe across
injection systems, roller bearings and cylindrical lenses. the sample surface at a scan height of only a few
Figure 14 shows the set-up of commercial available de- nanometres [59]. Unlike in conventional microscopy, the
vice, based on this method [57] resolution depends only marginally on the wavelength of
radiation but mainly on the geometry of the probe, i.e.
particularly the aperture diameter and shape of the probes
[50,60]. Probes with a well-defined geometry and high
transmission efficiency (or a high detection efficiency in
the case of active detecting probes) are therefore crucial
where NSOM is expected to yield sensitive and repro-
ducible results [61]. NSOM can be a powerful tool to ana-
lyse sub-microscopic structures - a crucial and challeng-
ing task of semiconductor technology. With NSOM a
lateral resolution of better than 80 nm can be reached
with 1064 nm wavelength light. This is about one order of
magnitude better than in conventional optical microscopy.
Figure 15 shows a scanned picture of a sub-microscopic
structure with different distances between the probe and
the surface.
Figure 14: Set-up of grazing incidence interferometer [57]
 global contrast function of the image. This procedure is
rather slow and the measurement uncertainty strongly
depends on the slope and the texture of the surface.
Some multisensor CMMs additionally have laser-based
autofocus systems (see section 3.4) which are integrated
into the optical system, and often are faster and more
accurate. Considering the advantages and disadvantages
of different sensor types, the trend for CMM manufactur-
ers goes towards multi-sensor machines in which me-
chanical and different optical sensors are combined to
measure in a common co-ordinate system (Figure 16).

Figure 15: Principle of near-field scanning microscopy

(NSOM) [61].

6 CO-ORDINATE METROLOGY

6.1 MuItisensor co-ord inate measuring machines

Conventional CMMs are equipped with tactile probes that
mechanically contact the workpiece surface. Tactile CMM
probes have reached a very high level of accuracy (c
0.5 pm) and reliability. Increasingly different types of opti-
cal sensors are being used. With non-contacting optical
sensors the measurement speed can be increased and
therefore a larger number of measurement points can be Figure 16: Applications of tactile and optical sensors by a
acquired within a shorter time. Flexible parts can be multisensor CMM [62].
probed without being deformed and very small structures
can be measured which could not be resolved with a With the opto-tactile fibre probe [65], characteristics of
conventional tactile probe. Typical optical sensors used optical and mechanical probing are combined in one
by CMMs are triangulation sensors (see section 3.3), sensor system: a fibre probe with a tip diameter down to
autofocus sensors (see section 3.4) and video probes. 25 pm and a probing force down to 1 pN mechanically
A video probe combines an opto-electronic camera with contacts the surface while its position is evaluated by a
microscope optics and appropriate illumination. Typical video camera. Other types of optical sensors may be
lens magnification factors for technical applications are 5 integrated into CMMs to meet future requirements con-
to 20, providing a resolution of 2 pm to 0.5 pm per pixel cerning flexibility, speed and accuracy: users and manu-
on a standard VGA CCD. Due to sub-pixeling, the meas- facturers look for systems based on structured light pro-
urement resolution can be several times higher. Attention jection (see section 6.2), white-light interferometry (see
has to be paid to illumination. Most systems offer at least section 5.3) or confocal microscopy (see section 5.2)
three different types of illumination that can be adapted to integrated into a CMM. The market share of CMMs with
the measurement conditions: bright-field toplight, dark- optical sensors has strongly grown over the last few years
field toplight, and backlight. Both, camera and illumination and today accounts for some 20% of the worldwide CMM
optics should generally be equipped with telecentric optics sales [66].
for high-accuracy measurements. This ensures that all
6.2 Photogrammetry and fringe projection systems
structures within the depth of focus are measured on the
same scale. Nevertheless, for many technical applications Photogrammetry and fringe projection systems are based
diffuse illumination has proven to furnish comparable on the principle of triangulation. They enable the recon-
results. All elements of the measurement process contrib- struction of surfaces by mathematically combining images
ute to the measurement uncertainty of video probes from different viewpoints. In photogrammetry for large-
[63,64]: the illumination, the object itself, the camera op- scale metrology, the measured surface is usually provided
tics, the camera, the image processing system and the with physical markers (e.g. retro-reflective dots) to gener-
mechanical structure of the CMM. ate so-called homologous points. These points are re-
On technical surfaces, especially the interaction between corded by a mobile camera from different perspectives.
illumination and workpiece surface influences the meas- To measure surfaces without physical markers, two ap-
urement quality. Although image processing has signifi- proaches are distinguished [67,68]: the first one is based
cantly improved over the last few years, optical edge onto the projection of a pattern on the surface and at least
detection on real product surfaces generally has not yet two different camera views. In this case, the pattern only
reached the accuracy and reliability of mechanical prob- serves to generate homologous points for triangulation
ing. On co-operative features like chromium structures on within the two or more camera images. The second ap-
glass, uncertainties in the sub-micrometre range can be proach evaluates the deformation of the pattern itself. In
reached, whereas they can be higher by one or two or- this case, the projector of the pattern takes up the role of
ders of magnitude with real workpiece features. Most one camera (Figure 17). The underlying mathematical
video probes on CMM offer an integrated autofocus func- principles are identical and can even be combined [69].
tion. The image-based autofocus records a number of For the calibration of such systems, sophisticated bundle
images in a narrow range around the focal point and adjustment techniques have been developed [70]. The
interpolates the position of the surface by analysing the
mathematical bundle adjustment model takes account of Since the eighties, attempts have been made to techni-
the exterior orientation of the cameras (or the projector) cally implement the interferometric principle of multilatera-
as well as interior geometrical parameters such as princi- tion based on interferometry to realise high-accuracy
pal focal length, centre of image plane, and lens distor- CMMs [74-761. For trilateration in space, at least three
tion. reference points with known co-ordinates are needed. If
the absolute distances of a moving target in relation to
these three points can be measured (e.g. by interferome-
try), its position can be determined. For most applications,
the positions of reference points are not precisely known
from the start. Therefore, at least a fourth reference point
is introduced. This over-constrained system allows a self-
calibration procedure to be applied, which can evaluate
the positions of all reference points by moving the target
through a set of positions in the measuring volume. After
the calibration procedure, positions in space can be de-
termined exclusively by distance measurements if the
reference points are invariant. The accuracy of the target
position depends only on the stability of the reference
points, the accuracy of the distance measurement and on
Figure 17: Use of structured light for co-ordinate the geometrical configuration. Currently, research insti-
measurement [73]. tutes such as the National Physical Laboratory (NPL) [77]
and the National Metrology Institute of Japan (NMIJ) [78]
Measuring systems based on fringe projection provide are developing systems based on conventional CMMs to
fast acquisition of dense point clouds. Vertical changes in which a metrology frame based on multilateration by
the surface topography result in the lateral deformation of interferometric distance measurement is added. Figure 18
the imaged light pattern, which is observed by at least one shows a scheme of the system as set up at the NMIJ.
matrix camera at an angle a, the triangulation angle. The
deformed light pattern can be assessed either by analys-
ing the bright-dark transition of projected lines (light stripe
technique) or by evaluation of a phase map associated
with the projection of a sinusoidal grating (phase-shift
method). If the object topography is not smooth (discon-
tinuous), the evaluation of the light pattern can fail be-
cause of shadowing and non-continuous imaging of the
light pattern. In this case, the evaluation of the pattern is
not unique. To overcome this problem, the coded light
approach [71] can additionally be applied. It is less pre-
cise but produces absolute 3D depth information even if
the surface shape is discontinuous. To calculate absolute
distances, multiple light patterns are needed, which can
be effectively projected by means of Digital Micro Mirror
Devices (DMDs). Fringe projection systems are benefiting
from the development of high-resolution CCD cameras, Figure 18: Scheme of the NMIJ multilateration system
video projectors with DMDs and the steep increase in [781.
computational power. Furthermore, the algorithms for
image processing and bundle adjustment are constantly The technical challenge of such systems is the realisation
improved. A typical measurement volume of structured of the optical and mechanical components for interfe-
light systems is in the range of side between 0.1 m and rometry. Commonly, at least four tracking interferometers
1 m. These systems provide a relative accuracy of up to are used, which point to one common target. The NPL
which depends on the phase measuring errors, the even suggests eight tracking interferometers pointing to
pixel and image co-ordinate measuring errors and the four targets to simultaneously evaluate the orientation of a
lateral structural resolution. Further uncertainty contribu- probe. Simulations have shown that such a system allows
tors are an incomplete geometrical model, inaccurate volumetric uncertainties in the range of U=0.2 pm (k=2) to
determination of geometrical parameters, mechanical and be achieved in most of the measuring volume of side
thermal instability, illumination and the optical properties 600 mm [77]. Another approach is the enhancement of
of the object surface. Secondary reflections, specular the accuracy of a CMM by a single laser tracker [79].
reflections, volumetric scattering, colour transitions, or Another approach which is also based on the multilatera-
ridges left by machining, may lead to gross systematic tion principle, is the so-called 'Laser p-GPS' [80], in anal-
measuring errors. When applied to the microscopic scale, ogy to the Global Positioning System (GPS). In this set-
fringe projection systems can even be used for the meas- up, white light emerges from the end of an optical fibre
urement of micro-shape and roughness [72]. and is reflected and focused back into the fibre by three
'satellites'. The reflected light is brought to interference
6.3 Recent developments with a reference signal, whose optical path length is con-
tinuously varied mechanically by a so-called scanner.
Multilateration, Optical GPS
Figure 19 shows the principle.
Standard kinematics of co-ordinate measuring machines
When the intensity of the white-light interference is con-
is based on a sequential chain of translation axes. In this
tinuously evaluated, three peaks can be observed, which
case, the Abbe principle is inherently violated. Rotational
correlate with the absolute distances from the 'satellites'.
errors of machine axes cause measuring errors that are
If the positions of the three satellites are known, the posi-
proportional to the distance between scale and object.
tion of the fibre end can be determined. This system is still
under development, but a prototype has already reached
an accuracy in the range of 10 pm in a measuring volume
of side 200 mm [80].

Figure 20: Principle of an industrial CT scanning system

with 2D detector.

7 CONCLUSIONS AND OUTLOOK

The ever quickening pace of the development of opto-
electronics and computational power has opened up a
Figure 19: Micro-GPS using white-light interferometry [80].
wide field of technical possibilities. Furthermore, optical
metrology is pushed by the demanding production re-
X-ray tomography quirements. Modern production is characterised by de-
X-ray computer tomography (CT) - so far primarily used creasing product development cycles, shorter innovation
for medical diagnostics - becomes an increasingly prom- intervals, rising automation, demand for highest quality
ising tool for dimensional metrology on engineering parts and shorter times of reaction to quality deficits. This de-
[81,82]. The visible contrast in CT is produced by the X- mands fast and accurate metrology which, ideally, should
ray absorption of the material and therefore is a function be integrated into the production line [83]. For this pur-
of the local electron density of the object under study. So pose, optical sensors have to be robust and reliable even
far, the main aspects of interest for the industrial applica- in harsh production environment. In-line applications of
tion of CT scanning are the non-destructive analysis of optical metrology often are found to be very sensitive to
faults (like cracks, flaws, shrinkholes) and the material contamination, temperature, vibration and e.g. lubricants
composition inside the volume. Increasingly, users want from the production process. An important role is played
to quantitatively measure internal and external geometri- by optical metrology, especially in the rapidly growing
cal features. The measurement chain of industrial CT fields of micro- and nanotechnology [53] for the meas-
starts with the source where X-rays are emitted either by urement of smallest structures. The application for optical
tubes with defined focal points or linear accelerators. The methods and the technical realisations are numerous, and
object to be scanned is located on a rotary table (Figure not all of them could be addressed in this paper.
20). Depending whether a line (ID) or an area (2D) de- In many cases, the obvious advantages of optical meth-
tector is used, CT systems are capable of measuring 2D ods related to speed and accuracy encompass complex
or 3D information with one revolution of the part. Due to metrological characteristics that demand specialised
its complexity, the CT scanning measurement uncertain- knowledge from the user. The characteristic uncertainty
ties have not yet been thoroughly determined. Up to now, contributors for optical methods are often not well known
resolution and repeatability aspects have been studied or documented. Moreover, the underlying algorithms and
and laid down in standards [see e.g. ASTM E 1441-001. filters often are not transparent to the user, which some-
One limit for the spatial resolution in CT is the size and times leads to difficulties in the interpretation of the re-
the shape of the X-ray focus. For X-ray tubes the mini- sults. Therefore, special considerations of the traceability
mum focus size increases with tube power and tube volt- of optical measurements have to be made [84,85]. In this
age. For larger parts, tubes of higher voltage have to be context, national metrology institutes in particular are
employed. Therefore, the achievable resolution decreases required to supply artefacts and methods to make optical
with the size of the object scanned. Typical parameters measurements traceable and comparable with mechani-
are in the order of 250 pm resolution with a 450 kV tube cal measurement. It cannot be denied that most of the
for an aluminium object up to 150 mm, or 10 pm resolu- surfaces of manufactured parts have mechanical func-
tion with a 225 kV micro focus tube for an aluminium tions (e.g. fittings, bearings, threads, gears, and so on).
object up to 10 mm in size. The quantities mainly influ- Mechanical measurement therefore, in most cases is
encing the measurement uncertainty are also the relative 'closer to function.' Furthermore, standardisation is mainly
distances of source, object and detector, the geometry of focussed on mechanical measurement, e.g. for roughness
the object, the lateral resolution of the detector and the measurement or co-ordinate metrology.
geometrical deviation of the mechanical axes. Additional A trend which can be recognised in optical metrology is
attention must be paid to the X-ray source spectrum, the miniaturisation of optical components, which reduces
beam hardening (change of X-ray spectrum inside the costs and benefits the applicability of optical methods in a
measurement object), object composition (material, production environment. In most cases, optical systems
roughness) and the detector properties (energy- are combined with a specialized and sophisticated soft-
dependent sensitivity, signal-to-noise ratio and dynamics). ware for data acquisition, evaluation and documentation.
For dimensional measurements, surface point detection For the user, this may involve difficulties as regards the
by identification of edges inside the 'voxel' stack has to be application of different systems in the same product con-
performed afterwards. For this purpose, algorithms similar text. A current initiative of sensor manufacturers and
to those used for video image processing are employed. users called OSlS (Optical Sensor Interface Standard) is
aimed at overcoming this shortcoming [86].
REFERENCES 21 Koch, W.,et al., 1998, Optische MelJtechnik an
technischen Oberflachen, Expert Verlag.
1. Harnessing light: Optical Science and Engineering 22 Neuschaefer-Rube, U., 2002, Optische Oberflachen-
for the 21st Century, 1998, National Academy Press, messtechnik fur Topografie und Material, VDI
Washington, D.C., ISBN 0-30905991-7. Fortschritt-Berichte, Reihe 8, Nr. 953, VDI-Verlag,
Dusseldorf.
2. Siegel, A,, Lifin, G. (editors), 2000, Deutsche
Agenda Optische Technologien, ISBN 3-00-006083- 23 Fan, K.-C.,. Chu, C.-L, Mou, J.-I., 2001,
9. Development of a low-cost autofocusing probe for
profile measurement, Measurement Science and
3. Estler, W.T., et al., 2002, Large scale metrology - an Technology, 12, 2137-2146.
update, Annals of the ClRP 46/2/2002.
24 Bonsch, G., 2001, Automatic gauge block
4. Market engineering research for the total European measurement by phase stepping interferometry with
industrial vision system market, 2000, Frost 8, three laser wavelengths, Proc. SPIE Vol. 4401.
Sullivan Report, Frost 8, Sullivan.
25 Matsumoto, H, Honda, T., 1992, High-Accuracy
5. Quinn, T. J., 1993, Mise en pratique of the definition length measuring interferometer using two-color
of the metre, 1992, Metrologia 31, 523-541. method of compensating for the refractive index of
6. Silvast, W.T., 1996, Laser Fundamentals, air, Meas. Sci. Technol. 3/1992, 1084-1086.
Cambridge University Press. 26 Abou-Zeid, A,, 1991, Diode Lasers for Industrial
7. Teimel, A, 1991, Technology and application of Length Measurements, Progress in Precision
grating interferometers in high precision Engineering, Springer Verlag, Berlin, 50-53.
measurement, Proc. 6th lPES/2"d UME 1991, 27 Maki, J.J., Cambell, N.S., Grande, C.M., Knorpp,
Springer Berlin, 15-30, R.P., Mclntyre, D.H., 1993, Stabilized diode-laser
8. Dr. Johannes Heidenhain GmbH, Linear encoders, system with grating feedback and frequency-offset
www. heiden hain.corn. locking, Opt. Comm., 102, 251-256.
9. Flugge, J., Dangschat, H., Spies, A,, Tschirnich, J., 28 Abou-Zeid, A,, Erdtmann, B., Kunzmann, H.,
Pieles, H., 1999, Concept of an interferometric length Prellinger, G., 1991, Wellenlangenstabilisierung
comparator with measurement uncertainties on the eines Diodenlasers durch Vergleich mit einer
nanometer scale, Proc. 1. Euspen Conference, MalJverkorperung, PTB Mitt. 101, 333-336.
Bremen. 29 Zarka, A,, Abou-Zeid, A,, Chagniot, D., Chartier, J.-
10 Kunzmann, H., Pfeifer, T., Flugge, J., 1998, Scales M., Cliche, J.-F., Cip, O., Edwards, C.S., Imkenberg,
vs Laserinterferometers - Performance and F., Jedlicka, P., Kabel, B., Lassila, A,, Lazar, J.,
Comparison of two Measuring Systems, Annals of Millerioux, J., Merimaa, M., Simonsen, H., Tetru, M.,
ClRP 42/2/1998, 753-767. Wallerand, J.-P., 2000, International Comparison of
11 Donaldson, R.R, Patterson, S.R., 1983, Design and Eight Semiconductor Lasers Stabilized on 12712 at
construction of a large vertical axis diamond turning 633 nm, Metrologia, 37, 329-339.
machine, Proc. SPIE, 433, 62-67. 30 Abou-Zeid, A. , Wiese, P. 1998, Diodenlaser-
12 Michelson A.A., Benoit J.R, 1895, Determination Refraktometer fur die interferometrische
experementale de la valeur du metre en longueurs LangenmelJtechnik, Technisches Messen 65, 91-95.
d'ondes lumineuses, Trav et Mem., BlPM ,ll,l. 31 Imkenberg, F., Nicolaus, A,, Abou-Zeid, A,, 1999,
13 Yacoot, A. 8, Downs, M. J., 2000, The use of x-ray Tunable 633 nm Diode Lasers and Application for
interferometry to investigate the linearity of the NPL Phase Stepping Interferometry, Proc. of the 1st Inter.
Differential Plane Mirror Optical Interferometer, euspen Conference, Pat McKeown a.o.(Ed.),
Meas. Sci. Technol., 11, 1126-1130. Bremen, Germany, 2, 243-246.
14 Zygo Corp., Heterodyne laser interferometers, 32 Stenger, J., Binnewies, T., Wilpers, G., Riehle, F.,
www.zygo.com. Telle, H. R., Ranka, J. K., , Windeler, R. S., Stentz,
A. J.,,Phase-coherent frequency measurement of a
15 Edlen, B., 1966, Refractive index of air; Metrologia 2,
71-80. Ca intercombination line at 657 nm with a Kerr-lens
mode-locked femtosecond laser, 2001, Phys. Rev. A
16 Birch, K.P., Downs, M.J., 1994, Correction to the 63, 021802-1-021 802-4.
updated Edlen equation for the refractive index of air,
33 Minoshima, K., 1999, Ultrashort pulse application for
Metrologia, 31, 315-316.
dimensional measurement, Femtosecond
17 Renkens, M., Schellekens, P.H.J., 1993, An Technology, Springer, 293-305.
accurate interference refractometer based on a
34 Data sheets Conoprobe, Conoline, Optimet Optical
permanent vacuum chamber, AoC 42/1/1993, 581.
Metrology Ltd, Jerusalem, Internet:
18 Estler, W.T., 1989, High accuracy displacement www.optimet.com.
interferometry in air, Appl. Optics, 24/6/1985, 808-
35 Sirat, G.Y., 1992, Conoscopic holography. I. Basic
815.
principles and physical basics, Journal of the Optical
19 Bobroff, N., 1987, Residual errors in laser Society of America A, 9, 70-83.
interferometry from air turbulence and nonlinearity.
36 Filatov, Y., Loukianov, P., Probst, R, 1997, Dynamic
Appl. Opt., 26, (13), 2676-2682.
angle measurement by means of ring laser,
20 Huser, D., Pfeifer, T., Scharsich, P., Trapet, E., Metrologia, 34, 343-351.
1994, A survey of the errors of optical coordinate
37 Beyer, W. (editor), 1990, lndustrielle Winkel-
measuring probes, Utraprecision in manufacturing
melJtechnik, Expert Verlag, ISBN 3-8169-0321-5.
engineering, Verlag Franz Rhiem, Duisburg, 11-14,
ISBN 3-926832-1 1-8. 38 Moller-Wedel Optical GmbH, Autocollimators, www.
moeller-wedel-opticaI.com.
39. Donges, A, Noll, R., 1993, Lasermesstechnik - 59. Paesler, M.A., Moyer, P.J., 1996, Near-field optics:
Grundlagen und Anwendungen, Huethig Buch Theory, instrumentation and applications, Wiley-
Verlag GmbH, Heidelberg. lnterscience Publication, New York.
40. Whitehouse, D.J., 1994, Handbook of Surface 60. Ohtsu, M., 1998, Near-field nano/atom optics and
Metrology, Institute of Physics Publishing Bristol and technology, Springer-Verlag, Tokyo.
Philadelphia. 61. Dziomba, T., Danzebrink, H.-U., Lehrer, C., Frey, L.,
41. Lawrence, A,, 1993, Modern Inertial Technology, Sulzbach, T., Ohlsson, O., 2001, High-resolution
Springer Verlag, Berlin, New York. constant-height imaging with apertured silicon
42. Pfeifer, T., Fertigungsmesstechnik, 1998, cantilever probes, Journal of Microscopy, Vol. 202,
Oldenbourg, ISBN 3-486-2421 9-9. Pt. 1, January 2001.
43. Pfeifer, T., Glockner C., 1999, Applications for a new 62. Werth Messtechnik GmbH, Giessen, Germany.
laser based straightness metrology system, Proc. 63. De Chiffre, L., Hansen, H.N., 1995, Metrological
SPIE laser metrology, Brazil, 121-132. limitations of optical probing techniques for
44. Malacara, D., 1992, Optical Shop Testing, John dimensional measurements, Annals of the CIRP
Wiley 8, Sons, New York. 44/1/1995, 501.
45. Bruning, J., Herriott, D. R., Gallagher, J. E., 64. De Santo, M., et al., 2000, Uncertainty
Rosenfeld, D. P., White, A. D., Brangaccio, D. J., characterization in image-based measurements: A
1974, Digital wavefront measuring interferometers preliminary discussion, IEEE transactions on
for testing optical surfaces and lenses, Appl. Opt. 13, instrumentation and measurement, Vol. 49, 1001-
2693-2703. 1007.
46. Pfeifer, T., Tutsch, R., Evertz, J., Weres, G., 1995, 65. Schwenke, H., Waldele, F., Weiskirch, C.,
Generalized aspects of multiple-wave length Kunzmann, H., 2001, Opto-tactile sensor for 2D and
techniques in optical metrology, 44/1/1995, 493-496. 3D measurement of small structures on coordinate
measuring machines, Annals of the CIRP,
47. Bodschwinna, H., 1987, Einsatzmoglichkeiten
50/1/2001, 361-364.
optischer Verfahren zur Rauheitsmessung in der
industriellen Fertigung, VDI Berichte 659, VDI- 66. N.N., VDMA-Marktstudie Messtechnik, 2000, VDMA,
Verlag, Dusseldorf. Germany.
48. Minsky, M., 1961, Microscopy apparatus, U.S. 67. Chen, F. et al., 2000, Overview of three-dimensional
Patent 3,013,467. shape measurement using optical methods, Optical
Engineering, 39/1, 10-22.
49. Wilson, T., 1990, Confocal Microscopy. Academic
Press, London. 68. Reich, C., et al., 2000, 3-D shape measurements of
complex objects by combining photogrammetry and
50. Wilson, T., Sheppard, C., 1998, Theory and practice fringe projection, Optical Engineering, Vol. 39, No. 1,
of scanning optical microscopy. Academic Press, 224-231.
London.
69. Schreiber, W., et. al., 1997, Self calibrating optical
51. Corle, T.R., Kino, G.S., 1996, Confocal scanning 3D-shape measurement system, Optical 3-D
optical microscopy and related imaging systems. measurement techniques IV, Eds., Gruen, Kahmen,
Academic Press, San Diego, USA. 243-250.
52. Pfeifer, T., Dussler, G., Brocher, B., 2000, Confocal 70. Atkinson, K.B. (editor), 1996, Close range
laser-scanning-microscope for microsurface photogrammetry, Whittles Publishing Services, UK.
characterisation and micrometrology, Proc.
International Colloquium on Surfaces, Shaker 71. Wahl, F., 1986, A coded light approach for depth
Aachen, Germany, 288-295. map acquisition, Mustererkennung 86, Ed.:
Hartmann, Springer Verlag, 12-17.
53. Pfeifer, T., Bitte, F., Dussler, G., 2000, Optical
Metrology for Microsystem Inspection, Proc. 72. Frankowski, G., et al., 2000, Real-time 3D-shape
International seminar on precision engineering and measurement with digital stripe projection by Texas
micro technology, Aachen, 39-52. Instruments micromirror devices DMD, SPIE-Proc.
3958, 90-105.
54. Gabor, D., 1948, A new microscope principle, Nature
161, 777-778. 73. Graebling, P., Boucher, C., Daul, Ch., Hirsch, E.,
1995, 3D sculptured surface analysis using
55. Dresel, T., 1996, Computergestutzter Entwurf
structured light approach, Videometrics IV, SPIE
diffraktiver optischer Elemente fur technische Proc. 2598, 128-139
Anwendungen, PhD thesis, University Erlangen-
Nurnberg, Germany. 74. Egdall, M., Breidenthal, R.S., 1983, Large surface
measuring machine, Applications of Optical
56. Weckenmann, A,, Bruning, J., Patterson, S., Knight,
Metrology - Techniques and Measurements II, Proc.
P, 2001, Grazing incidence interferometry for high SPIE, 416, 62-68.
precision measurements of cylindrical form
deviations, Annals of the CIRP, Vol. 50/1/2001, 381- 75. Pfeifer, T., Walter, J., 1989, Three-dimensional
384. interferometer system, SPIE, 1121, 258-265.
57. Zeiss, Tropel, 1998, Product information No. 60-20- 76. Nakamura, O., Goto, M., Toyoda, K. , Tanimura, Y.,
801-d. 1991, Development of a coordinate measuring
system with tracking laser interferometers, Annals of
58. Brinkmann, S., Dresel, Th., Schreiner, R., Schwider, the CIRP, 40/1/1991, 523-526.
J., 1999, Testing of rod objects by grazingincidence
interferometry: experiments, Applied Optics, 38, 121- 77. Hughes, E.B., Wilson, A,, Peggs, G.N., 2000, Design
125. of a High-Accuracy CMM Based on Multi-Lateration
Techniques, Annals of the CIRP, Vol. 49/1/2000,
391-394.
78. Takatsuji, T., et al., 1998, The first measurement of a 82. Flisch, A,, Wirth, J., Zanini, R., Breitenstein, M.,
three-dimensional coordinate by use of a laser Rudin, A,, Wendt, F., Mnich, F., Golz, R., 1999,
tracking interferometer system based on trilateration, Industrial Tomography in Reverse Engineering
Meas. Sci. Technol. 9, 38 -41. Applications, DGZfP-Proceedings BB 67-CD, Berlin,
79. Hartig, F., Keck, C., Kniel, K., Schwenke, H., 45-53.
Waldele, F., Wendt, K., 2002, Accuracy 83. Pfeifer, T., 2002, Optische Messtechnik -
enhancement of a co-ordinate measurement Erfolgsfaktor fur die Produktion, Proc. Aachener
machine by flexible integration of a position tracking Werkzeugmaschen Kolloquium, Aachen.
interferometer, Proc. 3. Euspen Conference, 84. Hillmann, W., Kunzmann, H., 1990, Surface profiles
Eindhoven, 541-544. obtained by means of optical methods, are they
80. Weule, H., Plutowsky, A,, Holler, F., Spieweck, M., representations of the real surface?, Annals of the
Werner, J., 2001, A six-degree-of-freedom ClRP 39/1/1990, 581.
measurement system for machine tools, WGP- 85. Bonsch, G., 1998, Gauge blocks as length standards
Annals. measured by interferometry or comparison: Length
81. Berger, U., Goebbels, J., 2001, Rapid tooling and definition, traceability chain and limitations, Proc.
computer tomography for aluminium casting of SPlE 3477, 199-210.
automotive components, Proc. pRapid, Amsterdam, 86. Keferstein, C.P., Zust, R., Marxer, M., Gachter, O.,
315-323. 2001, Optische Sensoren auf Koordinaten-
messgeraten, VDI-Bericht 1618, 181-192.