NIST_method for Absolute Measurement of T, R, And a of Specular Samples

8/10/2019 NIST_method for Absolute Measurement of T, R, And a of Specular Samples

1/9

Integrating-sphere system and method for

absolute measurement of transmittance,

reflectance, and absorptance of specular samples

Leonard Hanssen

An integrating-sphere system has been designed and constructed for multiple optical properties mea-surement in the IR spectral range. In particular, for specular samples, the absolute transmittance andreflectance can be measured directly with high accuracy and the absorptance can be obtained from theseby simple calculation. These properties are measured with a Fourier transform spectrophotometer forseveral samples of both opaque and transmitting materials. The expanded uncertainties of the mea-

surements are shown to be less than 0.003 absoluteover most of the detector-limited working spectralrange of 2 to 18m. The sphere is manipulated by means of two rotation stages that enable the portson the sphere to be rearranged in any orientation relative to the input beam. Although the spheresystem is used for infrared spectral measurements, the measurement method, design principles, andfeatures are generally applicable to other wavelengths as well. 2001 Optical Society of America

OCIS codes: 120.0120, 120.3150, 120.5700, 120.7000, 120.3940, 120.4570.

1. Introduction

Integrating spheres have long been used for the mea-surement of diffuse reflectance and transmittance of

materials in the UV, visible, and near-IR spectralregions, as well as somewhat more recently since the1970s in the mid- to far-IR regions. However, in-tegrating spheres have been used infrequently forspecifically measuring specular materials. This istrue despite the fact that, according to integrating-sphere theory, for an ideal sphere, a simple ratio oftwo measurements should result in the absolute re-flectance of a specular sample. The reason for thelack of use of integrating spheres for specular mea-surements of reflectance is that real integratingspheres are not ideal. The sphere-wall coating isnever a perfect Lambertian diffuser, baffles perturbthe light distribution within the sphere, and all de-tectors exhibit some angular dependence. Theseand other deviations from an ideal sphere can dras-tically affect the accuracy of the sphere equations.1

Because of the deviations, all spheres have somedegree of nonuniformity of throughput. This meansthat the detector signal will vary as the direction of

the reflected or the transmitted light and the regionsupon which the light is incidentwithin the sphere isvaried. The result may be errors in the quantitiesderived from the measurements. For this reason,measurements of reflectance of specular materialsare usually performed relative to a known specularstandard. When this is done, the regions of thesphere wall upon which the reflected light of the sam-ple and the reference falls usually have similarthroughputs.

For absolute regular specular reflectance mea-surements, various methods, including, VW, VN,and goniometer-based methods, are typically used.2,3

These methods typically do not involve an integrating

sphere. They involve an input beam, several direct-ing mirrors, and the detector. Some subsets of themirrors, the sample, and the detector are rotated andtranslated between sample and reference measure-ments. For the VN and the goniometer methods, asimple ratio of the two results produces the absolutesample reflectance, whereas for the VW method, thesquare root is taken.

The primary sources of error in the methods justmentioned are the result of alignment problems andthe spatial nonuniformity of the detector. Theseproblems can easily lead to errors of several percent

L. Hanssen [email protected] is with the Optical TechnologyDivision, National Institute of Standards and Technology, Gaith-ersburg, Maryland 20899.

Received 17 October 2000; revised manuscript received 20March 2001.

0003-693501193196-09$15.000 2001 Optical Society of America

3196 APPLIED OPTICS Vol. 40, No. 19 1 July 2001


2/9

or more.4 With considerable effort to achieve accu-rate alignment, excellent results can be achieved forstandards quality samples. However, even in thesecases, characteristics of the sample surface can limitultimate measurement accuracy.5,6 For transpar-ent materials, dealing properly with the transmittedlight and measuring the backsurface reflection accu-rately pose additional difficulties.

An important application of integrating spheres is

their use as an averaging device for detectors. Be-cause of the useful properties of the sphere, an aver-aging spheres entrance port can be both significantlylarger and much more spatially uniform than a baredetector. The trade-off made for these improve-ments is a degradation of the signal-to-noise ratio.The benefits of using the integrating sphere for moreaccurate detection of light are used in the design ofthe system and development of the method presentedin this paper. The measurement of absolute trans-mittance (), reflectance (), and absorptance () ofspecular samples is described and demonstrated.The inherent problems of sphere spatial nonunifor-mity are overcome through judicious use of the sym-

metries of the sphere design to establish symmetriesin the measurement geometry. After describing thespecifics of the integrating sphere in Section 2, theother components of the sphere system in Section 3,and the absolute measurement method in Section 4,we present the sphere characterization measurementresults for error analysis in Section 5. The achieve-ment of measurement uncertainties of 0.002 to 0.004are demonstrated in Section 6 for several common IRmaterials. Finally, Section 7 contains the discus-sion of the results with conclusions about the useful-ness of the sphere method for specular materials.

2. Description of the Integrating Sphere

The integrating-sphere system has been designedand constructed according to the specifications de-tailed in the following paragraphs. Figure 1 is aphotograph of the integrating sphere. Specific pa-rameters of the sphere, including a description andanalysis of the detectornonimaging-concentratorsystem, have been described previously.7 The insidewall of the sphere is coated with a material that isnearly a Lambertian diffuser and at the same timehas a high directional hemispherical diffusereflec-tance 0.9 for the IR spectral range: plasma-sprayed Cu on a brass substrate, electroplated with

Au.

The sphere has entrance, sample, and referenceports, all centered on a great circle of the sphere, asshown in Fig. 2. There also is a detector port, withits center located along the normal to the great circle.The white Hg:Cd:Te MCT detector Dewar locatedon the port can be seen mounted on the top of thesphere in Figs. 1 and 2. The detectors field of viewis centered on the same normal and corresponds tothe bottom region of the sphere. The sample and thereference ports are located symmetrically with re-spect to the entrance port and can be seen in theforeground of Fig. 1 the sample port has a KRS-5

sample mounted on it. The exact location of thesample and the reference ports is in general deter-mined by the angle of incidence for which the reflec-

tance and the transmittance are to be determined.An arrangement of ports could, in principle, be set upfor any angle of incidence from approximately 2 to28 and from 32 to 75, depending on the input-beamgeometry of the source or spectrophotometer. Forthis sphere, port locations have been selected for 8,which is close to normal incidence, yet for which noportion of the f5 6 half-angle input beam will bereflected back onto itself. For incidence angles inthe neighborhood of 30, a variation of the designwould be required so that the reflected beam from thesample port does not hit the reference port and vice

versa. The entrance port is of sufficient size 3.3-cmdiameter to accept the entire input beam, and the

sample and the reference ports are also sized2.22-cm diameterto accept the entire beamat thefocus, in a focused geometry. All the ports are cir-cular in shape, with the spheres inside and outsidesurfaces forming a knife edge at the port edge wherethey meet. In this sphere, as seen in Fig. 2, themeasurement of reflectance is designed for an inci-dence angle of 8 in general ; the sample and thereference ports are located at 16 in general 2and 16 in general 2, respectively, measuredfrom the center of the sphere and with respect to theline through the sphere center and the sphere wall at

Fig. 1. Photograph of the integrating sphere for absolute IR spec-tral transmittance and reflectance. The Hg:Cd:Te MCT detectorDewarwhiteis mounted on the top of the sphere. We view theback side of the sphere that includes referenceemptyand samplewith a KRS-5 window ports. A pair of rotation stages under-neath the sphere is used to move the sphere into positions for bothreflectance and transmittance measurements.

1 July 2001 Vol. 40, No. 19 APPLIED OPTICS 3197


3/9

a point directly opposite the entrance port. Bafflesseparating the detector port and the detector field-of-

view region from the sample and the reference portsare shown in Fig. 2. The baffles are critical to thesphere performance for characterization of diffusesamples,8 but do not play a significant role for thespecular-sample case.

The arrangement of the ports described above re-sults in the regions of the sphere wall illuminated bythe specularly reflected or transmitted light and thereference beam being centered on the same great circleas the entrance, sample, and reference ports. In ad-

dition, the regions are symmetrically positionedaround the entrance port. The reflected or the trans-mitted light also will be incident at the same angle onthese regions. As a result, the reflected or the trans-mitted light will have throughput to the detector thatis nearly identical. The procedure for orienting thesphere for the reflectance, transmittance, and refer-ence measurements is described in Section 3.

3. Sphere Mounting and Manipulation Hardware

The sample and the reference mounts, a pair of whichcan be seen on the sphere in Fig. 1, are constructed to

hold the sample against and centered on the sampleport from outside of the sphere. During spheremovement, the holders prevent the sample from mov-ing or shifting relative to the sample port. This isdone in such a way as to leave the back of the samplefree and open, so that the beam centered on the sam-ple can proceed through it for a transparent samplewithout obstruction. This is required for performingeither transmittance or reflectance measurements ontransparent samples. This arrangement can also beused to check thin-film mirrors for optical opacity.

The integrating-sphere system includes two motor-

ized rotation stages stacked on top of each other. Thestages are mounted with their axes of rotation parallelto each other. The rotation axes of the stages areidentified in Fig. 3. Stage 1 has its axis of rotationoriented parallel to the normal of the great circleformed by the entrance-, sample-, and detector-portcenters, as well as passing through the edge of thiscircle. This base stage remains fixed to the opticaltable. Its rotation axis is perpendicular to the inputbeam and passes through the beam-focus position.Stage 2 is mounted on the rotation table of the basestage so that its axis of rotation is located a distance

Fig. 2. Diagram of sphere interior and arrangement of its elements. Input and reflected beams are shown for a specular sample in thereflectance measurement geometry. The sample and the reference specular regions of the sphere wall are the first to be illuminated inthe sample and the reference measurements, respectively. The baffles are positioned for measurement of diffuse samples and are notcritical for specular sample measurement.



4/9

away from the base-stage axis exactly equal to thesphere radius. The integrating sphere is mounted tothe rotation table of stage 2 so that the stages axis ofrotation is along the sphere axis that includes the cen-ter of the detector port and the sphere center.

The function of base stage 1 is to vary the angle ofincidence of the input beam on the sphere surface andto switch between reflectance and transmittance

measurement geometries. The function of stage 2 isto select upon which port, the entrance, the sample,or the reference port, the beam will be incident.

4. Measurement Geometry and Method for

Reflectance and Transmittance

The arrangement of the input beam and the integrat-ing sphere for absolute transmittance and reflectancemeasurements is shown in Fig. 3. The sample re-flectance measurement setup is shown in Fig. 3a aswell as in Fig. 2, the reference measurement in Fig.3c, and the sample transmittance measurement in

Fig. 3e. In each diagram, the rotation required forreaching the following diagram is shown as a curvedarrow around the appropriate rotation axis. In thereflectance measurement geometry of Fig. 3a, theinput beam passes from the spectrometer through thesphere entrance port and onto the sample surfacefacing the sphere. This is the typical reflectance ge-ometry for directional-hemispherical sample reflec-tance in most sphere systems. The only difference

in Fig. 3ais the empty reference portas opposed toone occupied with a standard for a relative measure-ment. On reflection off the sample, the beam trans-

verses the sphere and is incident upon a region wedenote as the sample specular region see Fig. 2.From this point, the reflected flux is distributedthroughout the sphere in an even fashion because ofthe Lambertian coating and the integrating nature ofthe sphere. In Fig. 3a a clockwise rotation aboutaxis 1 turns the back of the sample to the beam in Fig.3b. An additional clockwise rotation about axis 2places the empty reference port at the input beamfocus in Fig. 3c, where it continues on to strike thesphere wall at the reference specular region labeled

in Fig. 2, producing the reference measurement.Another counterclockwise rotation about axis 2 re-sults in Fig. 3d, a repeat of Fig. 3b. A final coun-terclockwise rotation about axis 1 positions thesphere in Fig. 3e for the transmittance measure-ment, with the same angle of incidence as that of Fig.3a on the sample and incidence region on the spherewallthe sample specular region.

The somewhat unusual geometry for the referencemeasurementFig. 3cis chosen in order to achievethe highest degree of symmetry between the reflec-tance and the reference measurements. The sampleand the reference specular regions are symmetrically

located on either side of the entrance port. Becauseof the symmetry of the sphere design, the throughputis nearly equal for these two regions. Because theonly other difference between the sample reflectanceand reference measurements is the initial reflectionoff the sample, the ratio of sample reflectance andreference measurements is equal to the absolute sam-ple reflectance for specular samples. Varioussources of error, including the difference in the sam-ple and the reference specular region throughputs,can be included in the expanded measurement un-certainty or can be corrected for.

The ratio of the sample reflectance measurement ofFig. 3ato the reference measurement of Fig. 3cis

equal to the absolute sample reflectance for specularsamples. The ratio of the transmittance measure-ment of Fig. 3e to the reference measurement of Fig.3c is equal to the absolute sample transmittance forspecular samples.

The absolute absorptance is indirectly obtainedwhen the sum of the absolute reflectance and trans-mittance is subtracted from unity. Kirchoff s lawapplies because the reflectance and the transmit-tance measurements are made under identical con-ditions of geometry and wavelengths. The inputbeam is incident upon opposite surfaces of a sample

Fig. 3. Sphere measurement geometries for reflectance andtransmittance and rotation steps used to orient the sphere for eacha reflectance measurement geometry, c reference measurement,

and e transmittance measurement geometry. b and d areintermediate steps. Two rotation stages, stage 1 centered at theinput-beam focus and sphere wall and stage 2 centered at thesphere center, are used to change geometries.



5/9

for the reflectance and the transmittance measure-ments. For a uniform sample with identical frontand back surfaces, the side of incidence is immaterial.For samples with some asymmetry that is due to, forexample, a coating on one or both sides, the samplecan be reversed to make a pair of reflectance mea-surements and to obtain the corresponding pair ofabsorptance results.

5. Measurement Conditions and Sphere

Characterization

The integrating-sphere system is a measurementcomponent of a larger Fourier transform FT spec-trophotometerFTSsystem described in more detailelsewhere.9 The incident-beam geometry for allmeasurements described in Sections 5 and 6 is anf5cone with an 8 central angle of incidence. The FTSwas configured with a Whalogen lamp and coated-quartz beam splitter for the near-IR spectral region of1 t o 3 m 10,000 to 3300 cm1, and a SiC source andcoated-KBr beam splitter for the mid-IR region of 2 to18 m 5000 to 550 cm1. The spectral resolution iseither 4 or 8 cm1 for all results shown.

Every plot of transmittance and reflectance is ob-tained by the following procedure. A number of al-ternating measurements of reference, samplereflectance, and sample transmittancewhere appro-priatesingle-beam spectra are performed accordingto Fig. 3 and are repeated between 8 and 24 times.For each repetition, we calculate the transmittanceand the reflectance by taking ratios of the correspond-ing single-beam spectra and a reference single-beamspectrum obtained by interpolation to reduce theerror that is due to instrumental drift. From theresulting series of individual transmittance and re-flectance spectra, the mean transmittance and re-

flectance , along with the standard deviation andstandard error, spectra are calculated. Finally, theabsorptance spectrum is obtained from 1 .

We obtain each single-beam spectrum by coadding512 or 1024 scans of the FTS. The total measure-ment time for most of the results shown is severalhours from 2 to 10. The long measurement timesare required for obtaining the lowest noise level in thespectra. For less stringent requirements of 1% un-certainty, shorter measurement times of the order of10 to 20 min will suffice.

In addition to the various potential sources of errorthat are due to the FT spectrometer,10,11 several othersources of error may play a role in the sphere system

measurements. These are1spatial nonuniformityof the sphere-wall regions directly illuminated by theinput beam reference specular region or sample firstreflection sample specular region, 2 nonuniformityof the throughput of the sphere-wall region directlyilluminated by the input beam compared with that ofthe region illuminated by the sample first reflection;3overfilling the entrance port in the sample reflec-tance measurement, and 4 overfilling the sampleport in any of the measurements.

The spatial nonuniformity of the integrating-sphere throughput has been evaluated at the

10.6-m wavelength by use of a CO2 laser system.The local spatial variation across the region illumi-nated directly by the reference beam or indirectly bya specularly reflected or transmitted beam is approx-imately 0.4%.8 A translation of the light incidentupon the sphere wall of 0.5 cm through deflection ordeviation of 2 should result in a 0.1% relativechange in throughput. Thus a transmittance mea-surement of a sample with an effective wedge of 1

could lead to a 0.1% relative error in transmittance.Because of the decrease in wall reflectance and cor-responding decrease in throughput with decreasingwavelength, the error that is due to spatial nonuni-formity is greater at shorter wavelengths, especiallyin the near-IR region, approaching 1m.9 Spectralevaluation of this error is currently in progress.

The difference in throughput between sample- andreference-port measurement geometries can lead to asmall relative error, varying between 0% and 0.5%,depending on wavelength. The direct measurementof transmittance or reflectance for specular sampleswill include this error, but an additional measure-ment of the transmittance ratio of empty sample and

reference ports can be used to correct for the error.This has been done with good reproducibility andneed not be repeated unless the optical-input-systemgeometry or alignment is altered. A plot of such athroughput ratio is shown in Fig. 4.

Besides the generally featureless spectral curve, asharp structure occurs at approximately 8.5 m.This structure occurs at the only wavelength at whichthe incident beam from the FTIR beam is signifi-cantly polarized. It is anticipated that a future testof the sphere in which polarized light is used willshow increased throughput variation with polariza-tion. At the same time, averagings- andp-polarized

beam measurements is expected to eliminate thestructure seen at 8.5 m.The extent of overfilling the entrance, sample, and

reference ports can be examined by a measurement ofan empty sample or reference port in the reflectancemode. Any light coming through the entrance portand overfilling the sampleor referenceport will bemeasured as a reflectance component with near-unity reflectance of the sphere-wall region surround-ing the port. In addition, any overfilling of theentrance port in that measurement will result insome light scattering off the rim of the entrance portinto the sphere, resulting also in a reflectance com-ponent with high effective reflectance. A knife-edge

design could be used to reduce the result of entrance-port overfilling, but it is preferable to be sensitive toit in order to quantify it set an upper limit to it. Anexample of the combined overfilling-error reflectancemeasurement is shown in Fig. 5. This was obtainedafter careful alignment of the sphere system and theinput FTIR beam. The level of this measurement islower than has been reported previously.9 The pre-

vious empty-port measurement was in error becauseof a subtlety of the FT processing that produced arectification of the noise in the measurement. Thiserror was eliminated when the phase-error spectrum



6/9

obtained in the reference measurement was used forcorrection of the sample empty-port measurement.

The remaining important sources of error are re-lated to the FTIR spectrometer, detector, electronics,and FT processing, which are not discussed in detailhere. The combined measurement error for thetransmittance and the reflectance measurements in-clude both FTIR-related errors and the integrating-sphere system errors. A straightforward method ofevaluating the measurement accuracy of the systemis to compare measurement results with calculated

results for the optical properties of common IR opticalmaterials. Measurements of a few common materi-als are presented in Section 6, and the results areused to perform this comparison.

6. Transmittance, Reflectance, and Absorptance

Results

A number of optical components have been charac-terized by the integrating-sphere system, includingwindows, filters, and mirrors. Several examples ofwindow materials are shown in Figs. 6 and 7. Bothtransmittance and reflectance are measured with thesame geometry. From these two quantities the ab-sorptance is simply determined by subtraction of

their sum from 1. Comparisons can be made withcalculated values from handbook indexes of refrac-tion data.12,13 This can be done in two ways: 1Acomparison can be made with calculated transmit-tance and reflectance, and 2 a comparison can bemade with calculated absorptance.

The calculated and values from the n and kvalues in handbooks have finite uncertainty values.These are based on the uncertainties of the originaldata and the mathematical processes with which theindex valuesn and k were obtained. Other sourcesof error in this comparison include variations in thematerial itself, such as the method of growth and

processing.For specific spectral regions for many materials,however, the calculated absorptance can be deter-mined, with insignificant104error, to be 0. Be-cause of this, the indirectly measured absorptance inthese spectral regions can be used as an accurateevaluation of the total measurement error, not just asan estimate of uncertainty. If, in addition, a numberof materials with transmittance and reflectance val-ues spanning a significant fraction of their range 0 to1are measured, the measurement error can then beused with reasonable confidence for all transmittanceand reflectance results.

Four common IR window materials were charac-terized with the sphere system, with the results plot-ted in Figs. 6a 6d. They are Si 0.5 mm thickFig. 6a, ZnSe 3 mm thick Fig. 6b, KRS-5 5mm thick Fig. 6c, and MgF2 5 mm thick Fig.6d. For each sample, the transmittance, reflec-tance, and absorptance are plotted over a spectralrange of 218 m 118 m for Fig. 6a. Eachmaterial has a nonabsorbing spectral region oversome portion of that range. The reflectance valuesin the nonabsorbing range for the selected materialsrange from 0.05 to 0.45, and the corresponding trans-mittance values range from 0.55 to 0.95, in the non-absorbing regions. The MgF2 spectrum exhibits a

wide range of values for , , and within the spectralrange. At 12.5 m, both the reflectance and thetransmittance are 0, at which the absorption coeffi-cient is substantial 30and the indexnis close to1 3 0.

A closer examination of the indirectly measuredabsorptance in the nonabsorbing regions is shownin Figs. 7a7d. For each material, an ab-sorptance close to 0 is observed in the spectral re-gions with a nearly 0 k value. These are 1.25.5mFig. 7a, 213.5mFig. 7b, 218mFig.7c, and 24.5 m Fig. 7d. In the results,

Fig. 4. Sample- and reference-port throughput comparison, theresult of an empty-sample-port transmittance measurement.The curve represents the difference in detector signal for specu-larly transmitted and reflected light from the sample comparedwith light in the reference case. This spectrum can be used ineither of two ways: 1as a component to the systematic uncer-tainty of or , or 2as a correction spectrum to divide into theinitially obtained spectra of or to eliminate the error.

Fig. 5. Sample-port overfill measurement, the result of an empty-sample-port reflectance measurement. This is a characterizationof the baseline measurement capability of the integrating-spheresystem. The result can be used to apply corrections to a blacksample measurement.



7/9

spectral features are seen that can be attributed tocontaminants, primarily on the surface, such as wa-ter and hydrocarbon modes. Features from 6 to 7m, 9 to 10 m, and 12.5 m appear in the ZnSeand the KRS-5 spectra. These features could also,in part, be associated with the input-beam nonuni-formitys interacting with residual sphere spatialnonuniformity. In this way, the spectral structure

that is due to the beam splitter and the detector canbe observed in the results, thus comprising a com-ponent of measurement error. Conclusive resultswill require further characterization.

The indirectly measured absorptance levels inFigs. 7a7d over the spectral ranges cited abovecan be interpreted as arising from cumulative mea-surement error. The results exhibit an absolutelevel of error ranging from 0 to 0.002 for the struc-tureless spectral regions up to a maximum of 0.004,at which a structure is observed.

The evaluation of measurement error by use of

the zero-absorptance level for reflectance of thetransparent materials can be transferred with con-fidence to the opaque sample casefor which a zero-absorptance test is not feasible. An example is a

Au mirror reflectance measurement, shown in Fig.8. In the opaque sample case, there is only a singlereflection, whereas for the transparent sample case,multiple reflections contribute to the reflectance re-

sult. The higher-order reflected beams will be dis-placed because of the angle of incidence, enlargedbecause of focus shift, and perhaps deviated be-cause of sample wedge. Hence the effects of spa-tial nonuniformity of the sphere throughput will besmaller for the opaque mirror measurement, result-ing in a smaller relative measurement uncertaintyfor the sample reflectance. This is corroborated bythe relative lack of spectral structure below 16 min the Au mirror reflectance in Fig. 8 the structureabove 16 m is due to the increased noise level atthe extreme end of the detectors spectral range.

Fig. 6. Transmittance, reflectance, and absorptanceobtained from 1 of several common IR window materials, ranging in indexfrom 3.4 to 1.3: aSi, bZnSe,cKRS-5, dMgF2.



8/9

7. Discussion and Conclusions

The benefits of using the integrating sphere for moreaccurate detection of light are used in the design andthe development of a device for absolute measure-ment of transmittance, reflectance, and absorptanceof specular samples. The method is demonstrated inthe case of IR windows and mirror characterization.The ability to measure both transmittance and reflec-

tance in the same geometry is used to quantify di-rectly the total measurement error for nonabsorbingspectral regions, thereby also obtaining reliable un-certainty values for results outside these regions.

On careful study and consideration, it can beobserved that use of the integrating sphere sig-nificantly reduces several important sources ofmeasurement error, enabling the levels of accur-acy demonstrated in this paper. These errorsources include sample detector and detectorinterferometer interreflections, detector nonlinear-ity, detector spatial nonuniformity, and sample

Fig. 8. Au electroplated mirror reflectance. The spectrum is acombination of near-IR23.5m, circlesand mid-IR3.518m,squares spectra taken with different sourcebeam-splitter com-binations of the FTIR. This accounts for the reduced noise near 2m in comparison with Fig. 6.

Fig. 7. Expanded plot of spectra shown in Fig. 6, highlighting regions with absorptance near zero: aSi,bZnSe,cKRS-5,dMgF2

.The spectra, in regions where kshould be negligible,12 result from a combination of1cumulative measurement error from all sources intransmittance and reflectance, and 2 additional absorption that is due to volume or surface contaminants such as hydrocarbons andwater. The MgF2spectrum also shows regions of near-zero transmittance and reflectance at longer wavelengths.



9/9

beam geometry interaction beam deviation,deflection, and focus shift.11

The integrating-sphere system is not suitable forhigh-sample-throughput applications. For these ap-plications other instrumentation designed for fast rel-ative measurements can be used. Such instruments,in turn, can be calibrated with transfer standard sam-ples that are characterized with the sphere system.This approach has allowed us to improve the accuracy

of measurements made on all our FTS instrumenta-tion,9 including those designed for variable sampletemperature and variable incidence-angle character-ization, not directly feasible with the sphere system.

For accurate characterization of specular samples,direct mounting onto the sphere is not an absoluterequirement. Effective systems have been built, es-pecially for the UVvisiblenear-IR spectral region,that incorporate integrating spheres with detectorsin the standard averaging mode. However, there areat least two important advantages of mounting thesample directly onto the sphere as shown in this paper.Both of these relate to the characterization of nonidealsamples. The first is that this design can better han-

dle the worst samples and allow for the greatestamount of beam deflection, deviation, distortion, andfocus shift. The second is that one can measure sam-ples that are not perfectly specular, but that alsoexhibit some degree of scatter. The sphere measure-ment is a hemispherical measurement and will collectall or most of the scattered light in addition to thespecular component.14 Through use of a compensat-ing wedge to achieve normal incidence on the sample,for reflectance, and the sphere oriented in the positionof Fig. 3b for transmittance, supplementary mea-surements can be made to sort out the scattered com-ponent from the specular. For important IR window

materials, such as ZnS and chemical-vapor-depositeddiamond, some scatter is unavoidable and typically iswavelength dependent. The ability to detect andevaluate scattered light is important in understandingIR materials, in quantifying their behavior, and indetermining how they can be used appropriately inoptical systems.

The sphere system and method presented in thispaper is one that can be reasonably easily reproducedand implemented on, or adapted to, most FTIR spec-trophotometers in near- to far-IR regions. In addi-tion, the method can be readily implemented in theUVvisiblenear-IR regions for use with monochro-mator instrumentation. A duplicate system of the

one described herein would entail a moderate cost.A less expensive version with a simpler rotation stagemechanism and detector arrangement, smallersphere, etc., although perhaps not yielding the high-

est accuracy resultsapproaching 0.10.3%, should,if well designed, be expected to produce consistentlymeasurements with uncertainties of the order of 1%.This expectation is reasonable because many of theinherent benefits of using the sphere would remain,even for a simpler version. Such uncertainty levelswould compare favorably with most, if not all, com-mercial accessories currently available. At thesame time, improvements to the current sphere sys-

tem, including the development and the applicationof a more spatially uniform coating, should result inbetter performance and potentially higher-accuracydata.

References

1. D. Sheffer, U. P. Oppenheim, and A. D. Devir, Absolute ref-lectometer in the mid-infrared region, Appl. Opt.29,1291321990.

2. Absolute methods for reflection measurement, CIE Tech.Rep. 44 CIE, Vienna, 1979.

3. J. M. Palmer, The measurement of transmission, absorption,emission, and reflection, in Handbook of Optics AmericanInstitute of Physics, New York, 1997, Vol. II.

4. K. A. Snail, A. A. Morrish, and L. M. Hanssen, Absolute

specular reflectance measurements in the infrared, in Mate-rials and Optics for Solar Energy Conversion and Advanced

Lighting Technology, S. Holly and C. M. Lampert, eds., Proc.SPIE692,1431501986.

5. T. M. Wang, K. L. Eckerle, and J. J. Hsia, Absolute specularreflectometer with an autocollimator telescope and auxiliarymirrors, NIST Tech. Note 1280 U.S. Government PrintingOffice, Washington, D.C., 1990.

6. F. J. J. Clarke, Infrared regular reflectance standards fromNPL, inDevelopments in Optical Coatings, I. Ried, ed., Proc.SPIE2776,1841951996.

7. D. B. Chenault, K. A. Snail, and L. M. Hanssen, Improvedintegrating-sphere throughput with a lens and nonimagingconcentrator, Appl. Opt. 34, 795979641995.

8. L. M. Hanssen and S. G. Kaplan, Infrared diffuse reflectanceinstrumentation and standards at NIST, Anal. Chim. Acta380, 289302 1999.

9. S. G. Kaplan and L. M. Hanssen, Infrared regular reflectanceand transmittance instrumentation and standards at NIST,Anal. Chim. Acta380, 3033101999.

10. J. R.Birch and F. J.J. Clarke, Fiftysources of error in Fouriertransform spectroscopy, Spectrosc. Eur. 7, 1622 1995.

11. S. Kaplan, R. U. Datla, and L. M. Hanssen, Testing the ra-diometric accuracy of Fourier transform transmittance mea-surements, Appl. Opt. 36, 88968908 1997.

12. E. O. Palik,Handbook of Optical Constants of SolidsAcadem-ic, San Diego, Calif., 1985.

13. P. Klocek, Handbook of Infrared Optical Materials Wiley,New York, 1991.

14. L. M. Hanssen and S. G. Kaplan, Problems posed by scatter-ing transmissive materials for accurate transmittance and re-flectance measurements, in Optical Diagnostic Methods for

Inorganic Transmissive Materials, R. U. Datla and L. M. Han-ssen, eds., Proc. SPIE 3425,2836 1998.


Documents

NIST_method for Absolute Measurement of T, R, And a of Specular Samples