If this symbol - “ω” - is not a lower case omega inside quotation marks you should go the Browser Selection Considerations page.

Volume 104, Number 3, May – June 1999

Journal of Research of the National Institute of Standards and Technology

[J. Res. Natl. Inst. Stand. Technol. 104, 225 (1999)

The NIST Length Scale Interferometer

John S. Beers

William B. Penzes

National Institute of Standards and Technology,
Gaithersburg, MD 20899-0001

Table of Contents

1. Introduction

1.1. Importance of Length Measurement

Major changes have occurred in dimensional metrology in response to rapidly increasing scientific and industrial accuracy requirements over the past several decades. Accurate measurements of all kinds are vitally important to many scientific and industrial operations, and measurement of length is fundamental. Measurement of physical objects for quality control and process control is part of this picture. Semiconductor manufacturers, for example, have achieved remarkable improvements in quality control on ever smaller circuitry as a result of improved metrology in mask fabrication and production processes. Linear rules in many forms from encoded scales to photo lithographic grids are often used as check standards for such purposes and must be calibrated, when usage requires, to relate them to the international standard of length. For more than thirty years the evolutionary development and performance of the NIST length scale interferometer has influenced improvement in high accuracy length measurements.

1.2. Changes in the Meter Definition

In 1960 the Conférence Générale des Poids et Mésures (CGPM) adopted a new definition of the meter specifying it as an exact number of vacuum wave lengths of a particular spectral line of the ⁸⁶Kr isotope.This replaced the prototype platinum-iridium meter bar kept at the Bureau International des Poids et Mesures (BIPM) in Sévres, France as the international definition since 1889. Again in 1983, the CGPM redefined the meter, this time as the path length traveled by light in vacuum in 1/299792458 second [1]. In both redefinitions great pains were taken to make as little change as possible in the actual length of the meter. Along with the 1983 action, the CGPM sanctioned several vacuum wavelengths of iodine stabilized helium neon lasers as working length standards for relatively short lengths.Time-of-flight measurements are generally impracticable for lengths less than 50 m.

1.3. Changes in Length Measurement Methods

From 1889 to 1960, the international prototype length standard and the national length standards were graduated platinum-iridium meter bars. Intercomparisons of national standards with the prototype bar and calibrations of all other scales were accomplished incomparators employing filar microscopes for measuring length differences between a standard and an unknown [2]. Subinterval lengths were determined in the same comparators by a lengthy process of intercomparison by subdivision [3].

In 1958, NIST, then the National Bureau of Standards (NBS), began designing an interferometric linescale comparator in preparation for the proposed change in the definition of the meter. The design was based on the development of high speed automatic interference fringe counters in the preceding decade, opening a new era of length measurement by interferometry.

Prior to 1960, wavelengths from selected atomic light sources were sanctioned by CGPM as secondary realizations of the meter. These sources were used in measuring end standards of length. End standards are bars, usually metal, with optically finished, plane, parallel end faces defining a length. Measured by static interferometry¹ since early in the 20th century, their gaging faces are ideally suited to this method because they can serve as mirrors in the measuring legs of interferometers. Line scales cannot be measured efficiently by the static method for a number of reasons including the multiplicity of subintervals. They are best measured by dynamic (displacement) interferometry.

1.4. Varieties of Scales

Graduated length scales come in many forms, and are made in lengths from a few micrometers to over a meter. Those longer than a meter or two are usually classified as measuring tapes or rods. Many materials are used including steel, Invar, glass, glass-ceramics, silicon, and fused silica. Cross sectional shape can be rectangular, “H,”modified “U” (flat bottom), or a modified “X” (Tresca). At present, the line scale interferometer is limited to graduation widths ranging from submicrometer to 100 µm, and spacings ranging from less than 1 µm up to 1025 mm. Spacings are generally measured from center to center of the graduations, but can also be measured from edge to edge.

Some devices that are not strictly linear scales are measured in the line scale interferometer. These include end standards in a size range (250 mm to 1000 mm) that can present measurement problems with static interferometry. They are converted into line scales with ruled auxiliary blocks wrung to each end [4]. Once converted, the distance between the graduations is measured and then the auxiliary block line spacing is subtracted to arrive at the length of the end standard. Grid plates are measured by treating each row and column of graduations as an independent scale and, when possible, an estimate of orthogonality can be made by measuring the diagonals.

¹A static interferometer is defined as one where none of the optical components are intended to move during a measurement and the interference pattern is stationary. A dynamic interferometer is one where an optical component, usually a retroreflector, moves and the displacement of the moving element is measured by interference fringe counting or its equivalent.

2. The NIST Length Scale Interferometer

2.1. Design and Evolution

By 1961 an experimental Michelson type fringe counting interferometer using a ¹⁹⁸ Hg light source as a length standard was tested [5], but restricted coherence in atomic sources limited the range to about a decimeter. Development of a final version capable of measuring length scales proceeded, by good fortune, in parallel with the development of lasers [6]. The remarkable coherence of laser light made it ideal for fringe counting interferometry over long distances. In 1965, the two activities converged and a successful measurement of a 1 m scale with helium-neon laser wavelengths was accomplished [7]. Further refinements were then made in the line scale interferometer and laser wavelength stability was improved. By 1966, a semi-automated length scale measurement system was in operation [8].The mechanical and optical structures as they existed at that time are shown in Fig. 1.

Fig. 1. The length scale interferometer mechanical and optical components prior to 1970.

Development has continued on both the line scale interferometer [9] and on lasers [10, 11]. The major changes were: (1) rigid coupling and kinematic mounting of the microscope and interferometer fixed leg in 1970, (2) replacement of the Michelson-type plane mirror interferometer with a commercial laser interferometer in 1979, (3) modernization of the microscope electronics together with the addition of a computer/ controller in 1981 to completely modernize and automate the instrument [12].

Many minor modifications in the mechanical structure were also made as well as important changes in measurement procedures, data processing, and computational methods. These will be discussed under measurement assurance (Sec.3).

2.2. Description

The major components of the instrument are:

1. A 2 m waybed with a lead-screw driven carriage for translating the scale.
2. A fixed position photoelectric microscope with a servo system for centering scale graduations in the microscope field.
3. An interferometer using a stabilized laser light source as a length standard for measuring scales.
4. Measuring devices for temperature, barometric pressure, and air moisture content for use in computing laser wavelengths at ambient conditions.
5. A computer/controller for data recording and processing, and for automated instrument operation.
6. A temperature controlled enclosure to provide astable environment for the instrument.

2.3. Operation

A scale to be measured is first mounted on the carriage with its zero graduation at the left (called normal scale orientation). As the carriage is manually moved back and forth the scale is aligned and focussed relative to the microscope, then the interferometer is aligned coincident with the scale graduation axis. At the end of this procedure the carriage is positioned so the zero scale graduation is in the microscope field. After the temperature stabilizes at or very near 20 °C, and temperature gradients are minimized (usually overnight), the scale is ready for measurement. A computer program governs the measurement process. Preliminary data that must be entered by the computer operator includes scale identification, scale thermal expansion coefficient, data file name, the intervals to be measured, and the number of measurement passes to be made. Interferometer path air temperature and relative humidity, scale temperature,and barometric pressure can be entered by the operator or read automatically. Although set to a nominal zero at this point, the interferometer always displays a fractional count which is read by the computer at the start of the measurement.

The computer starts the measuring process as soon as preliminary data are entered. It verifies that a graduation is present and centered in the photoelectric microscope field, and reads the initial interferometer count. It also calculates the steps required of the stepping motor to reach a point just short of the next programmed graduation, and starts the stepping motor in its high speed mode. When the required steps are reached the motor shifts into its low speed mode and the microscope electronic circuits are activated. As the graduation enters the microscope field the line detection circuit stops the stepping motor and activates the hydraulic line centering servo system. When the line is centered, the interferometer is read. This process is repeated for each succeeding equally spaced graduation until the terminal graduation is reached. The operator enters a reduced list of preliminary data for the return pass and the process reverses until it closes at the zero graduation. More passes are performed as needed.

The scale is then remounted on the carriage with the zero graduation at the right end (reversed orientation), and the process is repeated as for the normal orientation.Two passes in each orientation is the minimum to provide valid statistical information. Increasing the number of passes generally reduces standard deviation and, consequently, uncertainty.

2.4. Length Computations

Data from the measuring system is converted into scale interval lengths at 20 ° C by the computer program.The count recorded at the zero graduation is subtracted from each succeeding recorded count in the pass to give the interval lengths in quarter wavelength units. A fringe multiplier then converts these into scale length units at 20 ° C.

3. Measurement Assurance

3.1. Definition and Purpose

Measurement assurance is a systematic standards for continually monitoring a measurement process and determining its performance parameters.

Treating regularly performed measurements as a production process is the basis of the program, while statistical analysis and control charts provide means for characterizing the process. Any organized measurement procedure that generates statistically significant amounts of data can be treated as a production process, monitored with controls, and analyzed to reveal characteristics such as measurement variability and uncertainty.

This approach to measurements was pioneered by Churchill Eisenhart [17] in the 1960s. Its applications were developed and expanded by Paul Pontius [18], Joseph Cameron [19] and Carol Croarkin [20] among others.

In a complex measurement system, such as the linescale interferometer, it is essential to know that all components are working properly. Substantial uncertainties can enter the process if, for example, the interferometer is misaligned, the barometer is out of calibration, or the distance between the remote interferometer and microscope changes during a measurement. A measurement assurance program (MAP) with its control standards, control charts, and statistical analysis is indispensable for detecting process uncertainties and malfunctions. It is also very useful for measuring effects of planned changes made for process improvement.

3.2. Control Standards

Two primary control standards are used to monitor this process. The first control, introduced in 1966, is a SIP Invar meter bar, M5727. It has an H shaped cross section with its neutral plane, located on the upper surface of the H cross piece, graduated in millimeters. The second control, introduced in 1982, is an 0.508 m (20 in) steel scale, No. 6495, graduated each 1.27 mm (0.05 in). A measurement process such as this one, which runs for decades, should have dimensionally stable control standards but completely stable materials are very rare, perhaps even non-existent. This can make it difficult to distinguish real from apparent secular length changes in the controls.

3.3. Process Changes and Control Charts

Figure 10 is the control chart showing the measurement history of M5727 (0 m to 1 m interval) from 1965 to 1985. Individual measurement values are plotted. A change in the measurement process was made at each vertical line, dividing the plot into periods. The mean value for each period is shown as a horizontal solid line, and the limits shown with dotted lines are ± 3 σ where σis the standard deviation of a single value.³ The ± 3 σ control limits are used to judge individual measurements. They are a statistical prediction, based on past performance, that 99.7 % of future measured values will fall within them. Values falling outside these control limits are judged to result from an out-of-control condition and require an investigation into the cause and a correction of the fault. The 3 σlevel is a generally accepted value for judging performance in measurement assurance programs.

Fig. 10. Meter M5727 control chart, 0 m to 1 m, 1971 to 1991.

Each process change and the mean value for its period is designated by M1 through M4. The change associated with each mean is described in Table 1. Data taken and changes made after 1985, designated as M5 through M8, are also described in the table but will be dealt with later in the text.

An individual measurement is defined here as the mean of two data sets of two passes each. One set is taken with the bar in the normal orientation (zero graduation at the left) and consists of one pass up the scale from zero to the terminal graduation (one meter in this case), and one pass down the scale, closing at the zero graduation. Data stops are made at each decimeter in both directions. The second set consists of two passes taken in a similar manner but with the bar in the reversed orientation (zero graduation at the right).

Two critical interferometer dimensions are given special attention in making process changes. These are: (1) distance between microscope and interferometer beam splitter, and (2) distance between retroreflector and scale. Changes in these relationships during a scale measurement will cause uncertainties.

Table 1. Sequence and descriptions of process changes

³ These conventions will be followed throughout for control charts.

3.4. M5727 Control Data, 1965 to 1985

Table 2 shows the mean value and long term measurement variability, as represented by control chart limits, for each period shown graphically in Fig. 10.

Table 2. Mean value and control chart limits for periods M1 through M4

The M1 period measurements were made in 1964 – 65 by comparing M5727 with NBS laboratory reference meter bars in a comparator employing micrometer microscopes to measure the length differences between reference bar and unknown in the traditional way [2, 3].The mean value is plotted in Fig. 10 with its 3 σ control limits. The first interferometric measurement period, M2, extended from 1966 to 1971. The M2 mean agreed with M1 within 0.2 µm. This was considered to be good agreement for that time when uncertainties for lengths up to 1 m were in the range of 0.2 µm to 0.5 µm.

A major structural change in the interferometer was completed in 1971. The photoelectric microscope was originally attached to the side of the waybed at the midpoint, and the beamsplitter was attached to the waybed at one end (see Fig. 1). A steel I-bar (not shown in Fig. 1) extending from the microscope to the top of the beam splitter assembly was installed very early to connect more rigidly these two components when it became apparent that the microscope angle changed when the carriage moved on the ways. The beam reduced this distortion, but a more effective structure was designed and installed in 1970–71.

In this new structure the microscope and beam splitter are mounted at opposite ends of a 1 m steel tube and the assembly kinematically mounted on the waybed. The beamsplitter end is fixed to the end of the waybed and the microscope end is supported on two 25.4 mm ball bearings free to roll only in a direction parallel to the ways (see Fig. 2). This decoupled the assembly from waybed distortions. Period M3 shows resulting measurements. There was an improvement in variability and a change in apparent length.

By 1978 the original fringe counting and microscope electronics were failing from age. In 1979 the electronics were replaced with modern components and the original NBS-made, Michelson-type, interferometer was replaced with a commercial model. All this was done while preserving the principles of the original NBS design. Commercial laser interferometers were by then quite reliable so a Hewlett-Packard (HP) model was installed. HP interferometer alignment was easier and more precise than with the Michelson, and the resulting improved measurement variability can be seen in period M4.

In the 1986 evaluation [20] of M1 through M4 data, a possibility of real secular lengthening of M5727, as suggested by the slope of a fitted line, was considered. It was rejected for several reasons. Most important was that the first interferometric measurement period, M2, was the least reliable, having been taken before the fixed optical components of the interferometer and the microscope were rigidly coupled and kinematically mounted. The change in mean value from M2 to M3 was, therefore, discounted as evidence for dimensional change in M5727. The relatively small change in M5727 from M3 to M4 was not statistically significant, but it might be attributable to measurement process change M4 (replacement of the Michelson type with an HP interferometer). However, the two interferometers were operated simultaneously, from opposite ends of the waybed, for a short period in 1979 to verify continuity. They agreed to within 0.01 µm in the measurement of M5727.

3.5. Control Data, 1971 to 1991, A New Process

3.6. Evaluation of Measurement Uncertainties

The 1986 uncertainty evaluation (see column labeled Process relative standard uncertainty, 1986 in Table 3) was based on data up to 1985. In 1987, a re-evaluation of measurement process uncertainties was undertaken. Although control standards are indispensable tools for monitoring the measurement process, they have the disadvantage of undergoing long-term length changes, leaving some doubt about process performance. Consequently, three goals were set for the new uncertainty study: (1) reduce measurement uncertainties, (2) establish a new measurement uncertainty value by approved methods [21], and (3) determine rates and directions of control bar length changes.

3.7. International Measurements: BIPM Meter Bar No. 12924

The second method for evaluating uncertainty from systematic effects mentioned in Sec. 3.6.2 is to compare measurements of a scale as performed by several independent and equally valid methods. This was done through the good offices and efforts of the Bureau International des Poids et Measures (BIPM). Starting in 1976, BIPM sent steel meter bar, SIP No. 12924, to the national standards laboratories of most of the industrialized nations in a successful and useful international standardization effort.

The following laboratories measured this bar: (Note: Country and laboratory names are those at the time measurements were made.)

1. Bureau International des Poids et Mesures (BIPM), Sévres, France
2. National Measurement Laboratory (NML), Lindfield, Australia
3. National Research Council (NRC), Ottawa, Canada
4. National Bureau of Standards (NBS), later NIST,Washington, DC, USA
5. National Research Laboratory of Metrology (NRLM), Ibaraki, Japan
6. National Physical Laboratory of (NPL), Tedding ton, UK
7. Amt fur Standardisierung, Messwesen und Waren prufung (ASMW), Berlin, East Germany
8. Institut Metrologie D. I. Mendeleev (INM), Leningrad, USSR
9. Physikalisch-Technische Bundesanstalt (PTB), Braunschweig, West German
10. Instituto di Metrologia G. Colonnetti (IMGC),Turino, Italy
11. National Institute of Metrology (NIM), Beijing, PRC
12. Federal Office of Metrology (OFMET), Wabern, Switzerland

BIPM measured the bar six times, NBS/NIST three times, NML and IMM twice each, and the remaining laboratories once each for a total of 21 measurements over a period of 12 years. Values for the one meter length are shown graphically in Fig. 18.

Fig. 18. International measurement data on BIPM Meter No. 12924, 0 m to 1 m.

In evaluating these data [27], BIPM concluded from its own and other measurements that the bar experienced a sudden lengthening of over 0.1 µm early in 1978. Such a change was probably caused by a severe mechanical shock during shipment. Based on this conclusion the data were divided into two groups; one before and one after the change. BIPM further concluded that the bar has a long term linear growth trend. This is shown in Fig. 19 where linear fits are made to the two groups of points. The first group lacks sufficient data and time span to establish a slope so it was given the same slope as the second group. Three points, indicated on the graph by diamonds, were deleted from the analysis.When they were included in the data the ± 3 σ limit becomes much larger and they could not be rejected by the accepted statistical test. In this case they were rejected judgementally because they did not fit the pattern convincingly established by the other 12 points in the second group.

Fig. 19. Lines fitted to the data on No. 12924, 0 m to 1 m.

NBS/NIST measurements are indicated in Fig. 19 by squares. The 1987 and 1988 points agree with the international average, as represented by the line fitted to all the accepted points, within 0.04 µm and 0.03 µm (0.04 µm/m and 0.03 µm/m), respectively. Control bar M5727 was measured a number of times during this same period so the validity of these M5727 measurements is greatly enhanced, the stability of the NIST measurement process is verified, and the case for growth in M5727 is strengthened. Table 5 shows these values.

Table 5. Measurements of control bar M5727 made during same period as measurements of meter No. 12924

3.8. Interpreting Control Charts

3.9. Treating Period M8 Data as a New Process

Data taken after coupling and kinematic mounting of the microscope and interferometer components in 1971 was treated as a new process (see Sec. 3.5.1). Now there is compelling evidence justifying that data taken since 1991, when the revised Edlén equation was adopted, should be treated as a new measurement process. Measurement values and variability have changed on both controls, as shown in Figs. 28 and 29, but it is less apparent in No. 6495 because it is influenced less by length dependent uncertainties except temperature. The rate of change derived from a linear fit to the data on M5727 is – 0.0059 (µm/m)/a. For No. 6495 the rate is – 0.0086 (µm/m)/a.

Fig. 28. M5727 data, 1991 to 1999, establishes new measurement process.

Fig. 29. No. 6495 data from new process, 1991 to 1999.

The slope of the fit is fairly uniform for the various data sets on the 0.508 m bar. However, it is not uniform for M5727 which now shows a negative slope for the first time. Secular dimensional behavior of materials varies widely so this is probably attributable to the metallurgy of the Invar alloy.

For the last seven years the new process has performed consistently and with less variability than before. Fig. 23 shows the old process with the ± 3 σ control limits at ± 0.13 µm while the new process is shown in Fig. 28 with the control limits at ± 0.044 µm. See Table 7 for the long term history of variability on both controls.

3.10. Measurement of Short Intervals

Demand for short interval measurements has risen with the development of ever smaller electronic circuitry. Scales ranging from a few micrometers to a few millimeters with line spacings from 1 µm to 10 µm are in this category. Measurement results on a 4 µm interval on the short-interval control standard, SRM474, are shown in Fig. 30 for the period from 1984 into 1997.

Fig. 30. SRM474, 0 µm to 4 µm control, 1984 to 1998.

Each point is the mean of at least 10 measurements.Expanded uncertainty for each point is shown by a vertical line through the point. All measurements are corrected for zero-shift. Based on these data, the ± 3 σ control limits are ± 8 nm, but starting in 1996 the data shows marked improvement in variability. This short- interval process, like that for longer lengths, has improved and warrants being treated as a new process.Length dependent uncertainties drop out except for the zero-shift correction, and zero-shift can be minimized by keeping the deadpath as short as possible and making measurements during periods of steady atmospheric pressure. Expanded uncertainties of one nanometer are now justified when the statistical analysis confirms them. This has been verified by several short interval calibrations for customers where one nanometer expanded uncertainty was achieved.

3.11. Long Term Variability

Table 7 shows long term variability as represented by control chart limits taken from Figs. 23, 28, and 29. Secular change is eliminated as a factor and the water vapor correction uncertainty is partially reduced in Fig. 23, and greatly reduced in Figs. 28 and 29. The most likely causes of the limits prior to 1991 are variations in instrument calibrations (particularly the barometer and hygrometer), variations in alignment of the interferometer, unstable barometric pressure caused by weather conditions, gradual or sudden changes in laser vacuum wavelength, or a temporary mechanical distortion in the apparatus. These problems are mitigated in measurements since 1991 by the attention given to factors causing them.

Table 7. Long term control chart limits ( ± 3 σ ) Time period M5727 No. 6945

3.12. Measurement Uncertainty

Until recently there was no universally accepted method for evaluating and stating measurement uncertainty. Various international organizations, led by the Comité International des Poids et Mesures (CIPM), have addressed this problem. The outcome is an International Standards Organization (ISO) document, Guide to the Expression of Uncertainty in Measurement [28]. NIST policy based on the ISO document is set forth in NIST Technical Note 1297, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results [21].

This protocol divides uncertainties into two categories:

A Those uncertainties evaluated by statistical methods;

B. Those uncertainties which are evaluated by other means. In the line scale measurement process, for lengths greater than 10 mm, Type B uncertainties tend to predominate and each one is evaluated independently. They are combined and summarized in Table 3 ,and verified by the BIPM meter bar measurements (see Sec. 3.7). The value of 5 × 10^{– 8} is length dependent and is called standard uncertainty arising from systematic effects, u_s(Type B).

When a high quality scale is calibrated it is measured with at least eight passes from which a standard deviation is computed for each interval. Variability represented by this standard deviation is Type A uncertainty and it comes from mostly nonlength dependent uncertainty sources. These sources are in the microscope and its line centering servo system, the interferometer system, humidity and barometric pressure instruments, graduation quality, and in critical distance changes between retroreflector and scale, and between microscope and remote interferometer. These components cannot be evaluated separately during a scale calibration procedure. Their combined value, contained in the standard deviation, is called standard uncertainty arising from random effects, u_r (Type A).

For line scales a typical Report of Calibration uncertainty statement is: “Each length value is the mean of eight measurements and the expanded uncertainty in each value is

where L is the nominal interval length in meters and 2 is the coverage factor.”

3.13. Conclusions

This effort to reduce uncertainties has improved measurement control chart limits (variability), both short and long term, and it lowered measurement uncertainty. Improvements were significant or noteworthy in the following cases:

1. The revised air refractivity equation produced the most dramatic improvements in variability and uncertainty.
2. Improved hygrometric measurements contributed substantially to the improvements.
3. Barometric pressure measurements are now more reliable but an expanded uncertainty of 8 Pa (0.06 mm of Hg) remains. We believe this can be improved with better instrumentation and calibration.
4. Correcting for interferometric zero-shift (dead path correction) has little effect on long term values but improves short term and long term variability.
5. The improved interferometer and scale axis alignment method, although not yet proven, should reduce these uncertainties.
6. Because laser vacuum wavelength measurements are now more readily available, uncertainties from this source are reduced.

Finally, interchanges of the BIPM meter bar verify that the NIST line scale measurement process is consistent and well within the control limits of the world's premier national laboratories. Simulation of control bar stability further demonstrates process stability. Process relative standard uncertainty is now estimated to be 5 × 10^{– 8} by the uncertainty budget method (see Table 3) and 3.5 × 10^{– 8} by the independent measurements method (international round robin). This is a marked improvement over the 9 × 10^{– 8} value in 1986.

4. Future Plans

4.1. Increase Line Detection Resolution

Although nanometer resolution is achievable with the present optical system under ideal conditions, there is the potential for further improvement with other detectors. One of these is the line scanning capacitance detector now under development. It offers the most promising approach to improving line detecting resolution.

4.2. Upgrade the Displacement Interferometers

A new high resolution interferometer system is being installed at the opposite end of the waybed from the present system. When complete, there will be two interferometer systems in place on a permanent basis. This was done on a temporary basis at various times in the past. Two systems can be used simultaneously, separately, or differentially for various applications. An opportunity exists for offering an intercomparison service for interferometer systems with the dual operation feature.

4.3. Upgrade the Data Acquisition and Contro Computer Systems and Their Interfaces

Along with a new interferometer system comes the installation of a state-of-the-art computer/controller. This will provide faster, more flexible, and more automated operation. For example, temperature, barometric pressure, and humidity data will be acquired at every measurement interval on a scale. This data will be used to compute and correct each interval for the exact conditions at the moment of measurement and provide automatic zero-shift corrections.

4.4. Improve the Mechanical Structure

Many structural changes were made over the years, and more are in the planning stage. Foremost among these is a plan to use low thermal expansion Invar in the beams coupling the interferometers to the line detector. These beams, one in each direction from the line detector, and each about 1 m long are steel at present. Using low thermal expansion material will reduce changes in critical dimensions between the line detector and the two remote interferometers located at the ends of the beams. Temperature changes occur slowly in the beams but can be significant in lengthy scale calibrations.

5. Summary

The NIST Length Scale Interferometer has served well, and is still serving well, as both a research tool and a calibration instrument. By systematic performance monitoring with a measurement assurance program, and incremental improvements through experimentation, this instrument keeps abreast of accelerating national accuracy requirements. It proved itself in international intercomparisons to be on a par with the best national laboratories. There will continue to be a need for high accuracy line spacing measurements and it is likely that the line scale interferometer with its proposed improvements will fill that need.

5.1. Acknowledgment

The authors wish to thank Ralph Veale and Clayton Teague for their support and encouragement in this project. We are also indebted to Ted Doiron for many useful discussions and contributions, and for co-authoring the paper on testing the revised air refractivity index; to Jack Stone and Steven Phillips for their valuable work on the zero-shift correction; and to the late Churchill Eisenhart for helping us through some knotty statistical problems.

References

1. Documents Concerning the New Definition of the Meter, Metrologia 19, 163 – 178 (1984).
2. B. L. Page, Calibration of Meter Line Standards of Length at the National Bureau of Standards, J. Res. Natl.Bur.Stand. (U.S.) 54 (1), 1 – 14 (1955).
3. L. V. Judson, Calibration of a Divided Scale, Circular of the Natl.Bur. Stand. No.329 (1927).
4. J. S. Beers, Measurement of Long Gage Blocks with the NIST Line Scale Interferometer, Natl. Inst. Stand.Technol.Technical Note 1410 (1995).
5. H. D.Cook and L. A.Marzetta, An Automated Fringe Counting Interferometer for use in the Calibration of Line Scales, J. Res. Natl.Bur.Stand.(U.S.) 65C (2), 129 – 140 (1961).
6. A. G. McNish, Lasers for Length Measurement, Science 146, 3641 (1964).
7. K. E.Gilliland, H. D.Cook, and K. D. Mielenz, Use of a Laser for Length Measurement by Fringe Counting, Metrologia 2, 95 – 98 (1966).
8. Line Standard Interferometer, Natl. Bur. Stand.Technical News Bulletin, March 1967
9. J. S. Beers and K Lee, Interferometric Measurement of Length Scales at the National Bureau of Standards, Precision Engin.4 (4), 205 – 214 (1982).
10. R. J. Hocken and H. P.Layer, Lasers for Dimensional Measure ment, CIRP Annals 28 (1), 305 (1979).
11. H. P. Layer, A Portable Iodine Stabilized Helium-Neon Laser, IEEE Trans. Instrum. Meas. IM-29(4) (1980).
12. W. B. Penzes and J. S. Beers, Evolution of Automatic Line Scale Measurement at the National Institute of Standards and Technology, Proc.of the Symposium on Measurement and Inspection in Industry by Computer Aided Laser Metrology, IMEKO TC Series No. 28, 41 – 58 (1990).
13. B.Edlén, Equation for the Refractive Index of Air, Metrologia 2 (2), 71 – 80 (1966).
14. K. P. Birch and M. J. Downs, An Updated Edlén Equation for the Refractive Index of Air, Metrologia 30, 155 – 162 (1993).
15. J. Beers and T. Doiron, Verification of Revised Water Vapour Correction to the Index of Refraction of Air, Metrologia 29, 315 (1992).
16. J. S. Beers and W. P. Penzes, NIST Length Scale Interferometer Measurement Assurance, Natl. Inst. Stand.Technol. NISTIR 4998 (1992).
17. C. Eisenhart, Realistic Evaluation of the Precision and Accuracy of Instrument Calibration Systems, J. Res. Natl.Bur. Stand. (U.S.) 67C (2), 161 – 187 (1963).
18. P. E. Pontius, Notes on the Fundamentals of Measurement and Measurement as a Production Process, Natl. Bur. Stand. NBSIR74-545 (1974).
19. J. M. Cameron, Measurement Assurance, Natl. Bur. Stand. NBSIR 77-1240 (1977).
20. C .Croarkin, Measurement Assurance Programs at the National Bureau of Standards, NBS Special Publications 676-I and 676-II (1984).
21. B. N.Taylor and C. E. Kuyatt, Guidelines for Evaluating and Expressing the Uncertainty of NIST Measurement Results, Natl. Inst. Stand.Technol.Tech.Note 1297, September 1994.
22. H. Preston-Thomas, Metrologia 27, 3 – 10 (1990).
23. BIPM Com. Cons. Thermometrie 8 , Annexe 18 (1967).
24. F. E. Jones, The Refractivity of Air, J. Res. Natl. Bur. Stand. (U.S.) 86 (1), 27 – 30 (1980).
25. (U.S.) Intergovernmental Panel (IPCC) Report on Climate Change (1990).
26. J. A. Stone, S. D. Phillips, and G. A. Mandolfo, Corrections for Wavelength Variations in Precision Interferometric Displacement Measurements, J. Res. Natl. Inst. Std. and Technol. (U.S.) 101 (5) (1996).
27. BIPM Meter bar round robin reports (1987).
28. Guide to the Expression of Uncertainty in Measurement, ISO/TAG4/WG3 (1993).

About the authors: John S. Beers is a physicist and Guest Researcher in the Precision Engineering Division of the Manufacturing Engineering Laboratory at NIST and William B. Penzes is an electronics engineer in the same organization. The National Institute of Standards and Technology is an agency of the Technology Administration, U.S. Department of Commerce.