A spectroreflectometer/transmissometer is described that permits determination of absolute optical charac-teristics in the 300-2600-nm wavelength region (which is essentially the complete solar spectrum). Theuniqueness of the instrument derives from use of three rapidly interchangeable 20-cm (8-in.) integratingspheres to measure (1) absolute hemispherical spectral reflectance as a function of angles of incidence from-40 to +400 employing an Edwards-type integrating sphere with a center-mounted sample [using small2.5-cm (1-in.) diam specimens], (2) absolute hemispherical and absolute diffuse spectral reflectance at anangle of incidence of 20° employing a sphere with a wall-mounted sample (for large specimens) and ascreened detector, and (3) absolute hemispherical and absolute directional (near-normal exitance) transmit-tance employing a complete integrating sphere with the only ports being for the sample and reference beams.Data are presented that demonstrate the ability to measure the spectral reflectance of nonmirror surfacesto an absolute accuracy of 0.995 (an uncertainty of ±0.005 reflectance units) in both reflectance spheres andof highly specular mirrors to an absolute accuracy of 0.993 (an uncertainty of ±0.007 reflectance units).Spectral transmittance can be measured to an absolute accuracy of better than 0.995 (an uncertainty of±0.005 transmittance units).
1. Introduction
A precise knowledge of the reflectance and trans-mittance characteristics of potential solar device ma-terials is required to make the proper choice of materi-als, to be able to predict the optically dependent ther-modynamic and electrical performance properties ofsolar devices, and to determine the extent of environ-mentally induced changes in such materials as a func-tion of outdoor exposure conditions and duration.
A unique multiple-integrating sphere spectropho-tometer was designed and constructed for spectrallymeasuring these characteristics to an absolute accuracyof -0.995 (an uncertainty of 0.005 measurementunits). The basis of the instrument is a BeckmanDK-2A ratio recording spectrophotometer to which hasbeen adapted three rapidly interchangeable integratingspheres each capable of performing different opticalmeasurements.
The authors are with DSET Laboratories, Inc., Phoenix, Arizona85029.
Because the solid angles of irradiation and collectionare mutually exclusive, in reality it is physically im-possible to measure reflectance, the ratio of reflectedflux to incident flux for true conditions of directionalirradiance and hemispherical reflectance. An inte-grating sphere is used to measure what is commonlycalled directional hemispherical reflectance but whatis actually a directional hemispherical reflectance factor.Reflectance factor is defined as the ratio of the flux re-flected by a sample under specified conditions of irra-diation and collection to that reflected by the idealcompletely reflecting isotropically diffusing surfaceunder identical conditions of irradiation and viewing.In a true reflectance measurement, the incident flux andreflected flux are measured separately. If an inte-grating sphere is used to collect the flux reflected overa hemisphere, there are losses in the reflected flux outof the apertures of the sphere, one or two entrance ap-ertures for the incident beam or beams and one for themeasured beams, and the measured reflectance is al-ways low by the amount of such flux losses. In a re-flectance factor measurement, the flux reflected by thesample and that reflected by a standard are measuredseparately and the ratio computed and multiplied bythe known reflectance of the standard. If a diffusely
reflecting standard of known absolute reflectance isused to measure the reflectance factor of a diffuselyreflecting sample, the flux losses will compensate, andsince both are usually quite small (of the order of 1-2%),the residual error due to flux losses will be very small.If a specularly reflecting standard of known reflectanceis used when measuring a specularly reflecting sample,the error due to flux losses is again very small. Thenumerical value of reflectance and reflectance factorsare identical for conditions of hemispherical collection,and perhaps for this reason, what is actually measuredas a directional hemispherical reflectance factor is fre-quently called a directional hemispherical reflec-tance.
The earliest integrating sphere reflectometers utilizedsubstitution-type spheres where first the referencestandard and subsequently the unknown specimen wereirradiated at the sample port. Large errors occurreddue principally to the fact that the sphere efficiency, asdefined by Jacquez and Kuppenheim,1 is different forthe two sphere geometries-one when the referencecompletes the sphere and another when the sample isin place. This error, which can be > 10%, results fromthe fact that the sphere wall radiance is not a simplefunction of the first reflected flux and results addi-tionally from the multiple-reflected flux hitting in theone case the reference standard and in the other thesample.
Many currently employed spectroreflectometersutilize comparison-type integrating spheres in whichthe sphere efficiency remains essentially unchangedwhen first the reference and then the sample are irra-diated. Nonetheless, small errors exist in those sphereswhere the beam-reference-detector geometries differslightly from the geometry existing when the sample isirradiated. These types of errors tend to be limited to-1% (Ref. 1).
Significant errors can be introduced in comparison-type spheres, however, when the geometry is such thateither or both the reference standard or sample speci-men disproportionately deposits first-reflected flux ontothe detector. This is not uncommon when measuringthe reflectance of mirrors and specular surfaces such asgloss paints. To minimize this error, a not uncommonpractice is to employ a highly reflective reference havingthe general specular characteristics of the sample to bemeasured-mirrors to measure mirrors and flat non-specular surfaces to measure flat nonglossy specimens.Additional errors may be introduced by an uncertaintyin the reflectance value assigned to the reference stan-dard. Furthermore, precise accounting of the referencereflectance requires wavelength-by-wavelength mul-tiplication of the measured reflectance of the specimenby the often-presumed reflectance of the standard.
The absolute measurements described in this paperare defined as being the true values obtained withoutthe necessity of employing reference standards. Inthese cases, the sphere wall is the standard, and thedevelopment of the specific integrating sphere theorycompensates for the absolute reflectance of the spherewall by mathematically treating the wall reflectance as
unity. Hence measurements made with such inte-grating spheres are the absolute value of reflectancesubject to the small errors associated with aperturelosses, small values of nonuniformity of sphere wallreflectance, and stray reflectance from internal baffles,sample mounts, etc.
B. Reflectance
1. Center Sample-Mounted SphereA major step forward in integrating sphere design was
the development of a sphere that provided for a cen-ter-mounted sample. Edwards et al. 2 have describedthe development of such a sphere, work that has led toits employment in commercial instrumentation and inthe laboratory by others including the author. Theirwork was based largely on the perfection of and elimi-nation of principal errors associated with an integratingsphere constructed by Toporets.3 Uncertainties asso-ciated with the Edwards-type sphere have been deter-mined by Newman et al.
4
The principal advantages of integrating spheres withcenter-mounted samples are that the total sphere ra-diance is a function of the first reflected flux only. Ithas been shown by Edwards et al. that the ratio ofsphere wall radiance when first the sample is irradiatedcompared with when the sphere wall is irradiated is, fora uniform reflectance of sphere coating, independentof the reflectance of the sphere wall and hence is de-pendent solely on the reflectance of the center-mountedsample. Other advantages of mounting the sample inthe center are that, because the detector may be locatedso as to not view the incident beam, the total sphereradiance is independent of the directional properties ofthe specimen, and the absolute reflectance may be de-termined through a wide range of incident angles. Theprincipal disadvantage of such a sphere is the require-ment that the specimens be small-usually circular witha diameter of 2.5 cm (1 in.).
A variation of the integrating sphere with wall-mounted samples has been constructed by the authorsbased on discussions in the mid-1960s with Kneissl andRichmond.5 6 While other detector-baffled integratingsphere reflectometers have been described,7 ' 8 the theoryof the DSET sphere is an extension of that presentedby Kneissl and Richmond.6 The design employs asmall internal screen that permits by the techniquesoutlined determination of the entering flux while at thesame time preventing the disproportionate irradiationof the detector by nonisotropically reflecting testspecimens. The principal advantage of the presentsphere and the impetus for its design are its capabilityfor measuring the reflectance of large specimens. In theKneissl and Richmond papers cited, the detector isdescribed as being occluded so that it views a smallconfined area A of the irradiated sphere. In the presentcase, the detectors are mounted flush with the wall of
the sphere and are irradiated over 27r sr (i.e., hemi-spherically). Thus the SNR of the present arrangementis much greater, providing both greater resolution ofspectra and a considerably greater operating wavelengthrange (i.e., in the short wavelength UV to as low as 220nm and IR regions to 2500-nm wavelength).
In developing the theory applicable to the presentdetector-baffled integrating sphere (shown in Fig. 7),the following notations will be employed.
PS = directional-hemispherical reflectance of thesample;
Pw = directional-hemispherical reflectance of thesphere wall;
Pw = average reflectance of sphere wall;F0 = radiant flux entering the sphere; andF3 = radiant flux first reflected from the sample.
For the conditions in which the sample is irradiated, thefirst reflected flux is
F, = p(1)
As shown by Miller and Sant9 and Edwards et al. ,2the successive interreflections at the sphere wall whenthe screened sample is irradiated result in a radiant fluxseen by the detector N8,
N, = FSp + Fpwpw + p +.. . + Fspj#- -'. (2)
Factoring we obtain
s-s pw + p2 +, * * + pa), (3)which is the nth partial sum of the geometric series
i Pwn=O
Multiplication of Eq. (3) by j5, gives
PWN = Fsp'(0 + -w2 +. + ~Subtracting PW. I4) from (3) gives(4)
Subtracting Eq. (4) from Eq. (3) gives
(5)
and solving for N, we obtain
Fspw( _ pfl) (6)
ForipiW < 1,pfl 0, and
N p. sP_ * (7)
Substituting Eq. (1) in Eq. (6), we obtain
N 1 Fop (8).=I .p s.
For the case where the sphere wall is irradiated (withthe irradiated area visible to the detector), the multiplereflected flux N, seen by the detector is
N. = Fop. + Foig+ .. + Fo~r, X(9)
and again for n - [analogous to Eqs. (3)-(9)]
Fop.lim NW = - (10)n-- I PW
Since the detector circuit in the Beckman DK-2Aspectrophotometer is designed to plot the ratio of the
signal from the sample to that from the reference beam,corresponding to the specimen and sphere wall radi-ances, respectively, it can be shown by the ratio of thecorresponding sphere irradiances that the reflectanceis simply the ratio of the detector signals when first thesample is irradiated N, to the signal when the spherewall is illuminated N.:
N, (FopI1 -Pw)psN. (Fop11 - Pv)
Solving for p,, we obtain
p, = NjNW,
(11)
(12)
which is simply the ratio of the radiant flux seen by thedetector (i.e., its signal) when the sample is irradiatedto the flux/signal when the sphere wall is irradiated.
Although the equivalency between the DSET and theEdwards-type spheres is demonstrated both in theoryand by experimental results, it is noted that the sum ofthe error contribution of the baffle and sample arearemoved from the sphere, however small, is larger thanthe error contributed by the interference of the cen-ter-mounted sample in the multiple reflection processassociated with the Edwards-type sphere. Edwards etal. 2 have evaluated the various errors associated withthe center-mounted sample configuration and showthem to be limited to i1% for diffuse and specularsamples. Kneissl and Richmond6 have shown by arigorous error analysis of their detector-baffled spherethat the errors can range from -0.5% for specularly re-flecting samples to +1.5% for diffusely reflecting sam-ples.
A rigorous error analysis of the present detector-baffled sphere is in development. Special emphasis isbeing devoted to the errors associated with the use ofthe absolute light trap, which is a unique feature of thewall-mounted-sample sphere geometry.
C. TransmittanceMany comparison-type integrating sphere reflec-
tometers are used to make hemispherical transmittancemeasurements by placing the specimen at the sampleport (the reference standard is air) and employing eitherflat or preferably spherical caps for the reflectance portsto complete the sphere. While the errors are usuallysmall with this technique, unmatched spherical caps cancreate significant errors when the detector-beam-spherical cap geometry is different for the two positions.As will be shown in our design, we have eliminated thispossibility altogether by designing a separate completesphere for hemispherical and directional transmittancemeasurements.
II. Instrument Design
A. Spectrophotometer
A Beckman model DK-2A ratio recording spectro-photometer was chosen for the multipurpose instru-ment described for the following reasons: (1) itsratio-recording electronics are ideally suited to reflec-tometry and hemispherical transmissometry; (2) itsmechanical design easily permits reconstruction of both
the entrance and exit optics (to the monochromator);and (3) the DK-2A's wavelength range of 200 nm (withhydrogen discharge source) through 2800 nm includesover 99% of terrestrial solar energy available at air mass1.5 (which most likely will become the U.S. ReferenceStandard Spectra). Furthermore, the DK-2A's wave-length resolution is adequate for the measurementsrequired. (The DK-2A is no longer manufactured al-though refurbished instruments are often available.However, the Beckman model 5240 UV-VIS-NIR in-strument is a modern solid-state version of the DK-2A,and it and similar ratio recording spectrophotometerscan easily be adapted for multiple sphere operation withsimple transfer optics.)
The DK-2A's single-beam single-dispersion mono-chromator utilizes a high-dispersion quartz prism anda slit servomechanism that permits a nearly constantreference beam energy. Scanning rates may be selectedfrom seven distinct operating ranges, and four separatetime constants are available.
B. Redesigned Optical Path
The DK-2A has historically been provided with anoptional reflectometer sample compartment as an ac-cessory. The attachment consists of an entrance mir-ror, a motor-chopper assembly, focusing sample andreference beam mirrors, an integrating sphere, and adetector housing (with a 1P28A photomultiplier andlead sulfide cell). Only the motor-chopper assemblyand two beam focusing mirrors were utilized in rede-signing the front-end optics.
The common optical path of the multipurpose spec-trophotometer is dimensionally different but geomet-rically similar to the original DK-2A spectroreflecto-meter design. It is shown in Fig. 1. The light tightintegrating sphere compartment is shown in the pho-tograph of the instrument (Fig. 2) and utilizes a baseplate with two sets of recessed sphere-mounting holes(one for the two reflectance spheres and one for thetransmittance sphere).
C. Integrating Spheres
All three integrating spheres are constructed fromspun aluminum alloy as 20-cm (8-in.) diam, 0.4-cm(0.1875-in.) thick hemispheres. The hemispheres wereprovided with 1.8-cm (0.75-in.) helioarc-welded flangesand mounted on frames with four mounting legs asshown in Fig. 3. The hemispheres were machined to aNo. 8 finish and lapped to mirror brightness, then spraypainted with eight coats of Eastman 6080 barium sulfatepaint.
The detector housing was constructed in accordancewith the original circuitry of the DK-2A spectroreflec-tometer. However, the two-detector housings of theoriginal instrument were redesigned to contain both the1P28A photomultiplier tube (PMT) and the 10- X20-mm lead sulfide cell. As shown in Fig. 4, the PMTis mounted flush with the sphere, and the lead sulfidecell is mounted on the surface of the BaSO 4-coatedspherical segment. (The leads are directed through
Fig. 4. Sphere-wall mounted lead sulfide and 1P28A PMdetectors.
>SPECIMEN -40' TO +10'\ \ INCIDENCE
REFERENCE BEAM
TRANSFER MRRORSSAMPLE BEAM
Fig. 5. Edwards-type integrating sphere with center-mountedsample.
ceramic ferrules.) Because the PMT's active surfaceis removed from the sphere wall, giving it directionalresponsivity, it was lightly smoked with MgO by burn-ing magnesium ribbon to provide a near-Lambertianreceptor. To ensure matched responsivity at 700 nmin both detector regions, the lead sulfide cell was alsolightly smoked. The ratio-recording circuitry is phasedwith the chopper shaft commutator, which is in turnaligned with the oscillating mirror to detect first thesample beam and then the reference beam. The ratioof the two is displayed on the chart (shown in Fig. 2) asa function of wavelength.
D. Edwards-Type Sphere
The design of the Edwards sphere with a center-mounted sample is shown in Fig. 5. The relation amongthe motor-chopper assembly, sphere, and detector isshown in this drawing. The specimen is affixed to themount located at the end of a rigid arm so that it is at theexact center of the sphere. The detector is mounted sothat the entrance port, specimen, and detector form aright triangle with the detector looking only at thespecimen's edge. The sample mount holder is aspherical segment graduated in 5 increments from -40to +40° with respect to the entrance beam. The two
sphere entrance mirrors are matched and mounted soas to focus the sample beam on the center of the sampleat 00 incidence angle and the reference beam on thesphere wall immediately above the center-mountedsample.
The sphere is coated until a 100% line provides thekind of uniformity shown in Fig. 6. This is accom-plished by removing the sample mount (with sample)so that both beams strike the upper sphere region. Thesample is placed in the beam and turned to the desiredincident angle other than 00. (At 00 most of the spec-ularly reflected component of the incident beam wouldbe reflected back out of the sphere.) The reflectanceis determined by following the published operatingprocedures of the DK-2A spectroreflectometer. Sincethe resultant spectra are considered absolute, no ad-justment is necessary to account for a reference stan-dard.
E. DSET Detector-Baffled Sphere
The design of the detector-baffled sphere is shownin Fig. 7. The relation among the light baffle, sampleport, detector, and specular light trap is shown in thisdrawing. The sample port is 2.5 cm (1.0 in.) in diameterand machined to a knife-edge flat [to accommodate
1001-
-80a
I
L60J
40
I 2Is
; 970- LINC (EDWARDS SPHERE)
- DSET DETECTOR-BAFFLED SPHERE
- EDHARDS-TPF SPHERE
FRESNEL LIGHT TRAP
500 1000 1500WAELENTH, NM
Fig. 6. Comparison of integrating spheres in thea flat white surface.
2000
measurement of
A LIGHT AFFLE
LARGESAMPLE
Fig. 7. DSET detector-baffled integrating sphere with wall-mountedsample.
samples of up to 10 X 30 cm (4 X 12 in.)]. The incidenceangle is 200. The light trap port was drilled only aftercompletion of the sphere to account for the exact opticalpath. The light trap is provided with a BaSO4-paintedspherical segment to complete the sphere and a razor-blade Fresnel-reflecting light trap. The light trapconsists of 160 Gillette Blue Blades (cut in half lon-gitudinally and glued side by side in standing position).It is fitted with a BaSO4-painted mask to just admit tothe trap the reflected beam image when a front surfacemirror is placed at the sample port. The field of viewis an angle of 3.6°. The trap and its spectral reflectancecurve, when placed at the sample port, are shown in Fig.8. A sphere-completing plug is used when determiningtotal hemispherical reflectance. (It is essential that itbe in place when the sphere is coated to ensure equiv-alency of the first reflected flux from the sphere wallwhen mirrors are being measured.)
The 100% line may be obtained by a two-positionmirror that directs the sample beam from the sampleport onto the sphere wall but in a position not shieldedfrom the detectors [as defined by theory, Eq. (1-12)].However, we have experienced mechanical difficultieswith this method and instead employ an alternativeindirect method of obtaining the spectrophotometer's100% line when using this sphere. This is accomplishedby using an FEK aluminized mirror (for mirror reflec-tance measurements) and a white enamel (for all othersurfaces) that were measured on the Edwards sphere toset the 100% line for the detector-baffled sphere (FEKis the registered trademark of 3M Co.). The reflectancevalues at 700 nm obtained with the Edwards sphere areset for the baffled sphere with the respective referencemounted in the sample port; this simply sets the proper100% line, and the reference is replaced with the un-known specimen. The light trap or the spherical plug,as appropriate, is inserted, and the reflectance is de-termined as with the Edwards sphere. With thespherical segment in place, absolute hemisphericalspectral reflectance is determined, whereas with thetrap in place, the absolute diffuse spectral reflectanceis determined (with specular component in a 3.60 coneremoved). The computed wavelength-by-wavelengthdifference provides the 3.6° specular spectral reflec-tance.
F. Transmittance Sphere
The optical path and transmittance sphere design areshown in Fig. 9. With the sample mounted at positionS2, the measurement obtained is the near normal-nearnormal (70) spectral transmittance with the trans-mitted flux collected over a conical solid angle havingan included plane angle of 7°. By extending dimensiond to 40 cm (16 in.) (with a bare plate adaptor), the in-cluded plane angle can be reduced to 1.80. With thesample mounted at position SI, absolute hemisphericalspectral transmittance is determined at an incidentangle of 70.
IV. Results
A. Resolution
The resolution of the instrument was determined forall three detector-sphere combinations (each of thesedetector pairs is permanently mounted to a sphere);nominal values of slit width are presented in Table I.
B. Edwards-Type Sphere
The absolute hemispherical spectral reflectances ofa front surface aluminum mirror and a black nickel se-lective surface are presented in Fig. 10 as functions ofangles of incidence. Typically a 97% line (shown) isdetermined instead of a 100% line to measure more ac-curately the maximum deviation existing in the entirewavelength range (Fig. 6).
Fig. 10. Dependence of incident angle on hemispherical spectralreflectance of specularly reflecting surfaces measured on the Ed-
wards-type sphere.
100FRONT SURFACE (FS) MIRROR
D. Transmittance SphereThe absolute spectral transmittances of 3-mm (1/8-in.)
thick Pyrex glass and 3-mm thick textured glass arepresented in Fig. 12. The transmittance spectra wereobtained at an angle of incidence of 70 in all cases. Thedetector geometries employed were hemispherical (27rsr with the specimen mounted against the integratingsphere) and 2.5° included plane angle with the specimenmounted 25 cm (10 in.) from the integrating sphere'ssample port (as described in Fig. 9).
E. Measurements of Standard Reference MaterialsCertified NBS Standard Reference Materials (SRMs)
were obtained and measured to determine the certi-fiable accuracy of the two reflectance spheres. SRM2022 black porcelain enamel and SRM 2020 diffusewhite coating for 6 directional-hemispherical reflec-tance were measured on both spheres. (SRM 2020cannot be measured in the Edwards sphere due to thespecimen size.) The results of these measurements arepresented in Fig. 13.
100
-------- DSET DETECTOR-BAFFLED SPHERE
- EDWARDS-TYPE SPHERE 1/8-IN THICK PYREX GLASS80 , IE , I | . . I , , , , , , I
HEMISPHERICAL EXITANCE ----------…NEAR NORMAL EATRNCE (2,5 F.O.V.)
1001
BLACK CHROME ,,COATING -a
._, _______-_____=____--RR (DIFFUSE ONTY)
9 -- .- - t ~ I ._ __ I t- I 1-~~~~1000 1500
HAVELENGTH, NH2000
Fig. 11. Comparison of integrating spheres in the measurement oftwo specularly reflecting surfaces.
00I
, 0
60
401/3-IN THICK TEXTURED GLASS
l l I I l
500 1000 1500WAELENGTH, NM
I I I I I I I I
2000
C. DSET Detector-Baffled SphereThe spectral reflectances of a white coating, an alu-
minum front-surface glass mirror, and a black chromeselective foil were determined first on the detector-baffled integrating sphere and then in the Edwards-typesphere at an angle of incidence of 200 (the fixed angleof incidence employed in the detector-baffled sphere).Specimens for the Edwards sphere were cut from longerspecimens employed in measurements made on thedetector-baffled sphere.
The reflectance of the flat near-Lambertian whitepaint is presented in Fig. 6 as measured with the twodifferent spheres. Also shown in Fig. 6 is the spectralreflectance of the razor blade Fresnel light trap.
The spectral reflectances of the front surface mirrorand the black chrome foil-backed surface coating arepresented in Fig. 11. The diffuse reflectance of themirror is also shown. It was determined with the 3.60conical component of the first reflected flux from themirror removed with the Fresnel light trap in place.
Fig. 12. Measurement of transmittance of two optically differentglasses in both hemispherical and near normal exitance modes (2.50
FOV).
1000_
:80us
X 402
,6
o20
SNS ERR N. 2020 (HITE)
- D0ET DETECTOR-BAFFLED SPERE
0 CERTIFIED TALUE PRETIDED BY THENATIONAL BUREAU OF STANDARDS
BS SR NO. 2022 (BLACK)
I I I I I I I I I I I I I I
1000 1500HATELENYTH, N1
u 500 2000
Fig. 13. Comparison of measured values of absolute hemisphericalspectral reflectance with certified values furnished by NBS.
V. Discussion and ConclusionsThe unique utility of the multipurpose integrating
sphere spectrophotometer has been shown. Sphereinterchange can be accomplished in <3 min, and, be-cause of common optical alignment, only a perfunctorybeam alignment check is required after any one sphereinterchange.
Agreement between the DSET detector-baffled in-tegrating sphere and the well-known proven accuracyand precision of the Edwards-type reflectance spherehas been demonstrated. The greatest agreement be-tween spheres exists for diffusely reflecting materials.The difference between spheres for front surface mirrormeasurements is only 0.005 from the limit of slit servocontrol of the detector-baffled sphere at the 375-1900-nm wavelength, 0.01 from 1900- to 2300-nmwavelength, and 0.02 from 2300- to 2400-nm wavelength(the long wavelength limit of the instrument). This isconsidered excellent agreement insofar as the greatestsensitivity to differences between sphere-detectorgeometries is exhibited by highly specular samples.Similarly good agreement was observed betweenspheres in the measurement of the black chrome foilspecimen. The difference in absorptance units is 0.005from 375- to 700-nm, 0.01 from 700- to 1700-nm, and-0.015 from 1700- to 2400-nm wavelengths. The dif-ference in the 1700-2400-nm wavelength region isthought to be due to the illumination of different areason the specimen; selective surfaces typically have po-sition-sensitive broad interference characteristics in thenear IR (which are usually small in magnitude).
The agreement between the multisphere spectrore-flectometer and the National Bureau of Standards inmeasuring the NBS SRMs is considered excellent. Themaximum difference measured in the DSET detector-baffled sphere was 0.006 reflectance units (at 650 nm)for the white SRM, and the average difference was only0.004 in the 350-2400-nm wavelength region. Themaximum difference observed for the black SRM was0.005 reflectance units at 2200-nm wavelength, and theaverage difference in the entire region measured wasonly 0.002.
The maximum difference observed for the black SRMwas 0.002 for the Edwards-type sphere (at 1000 nm),and the average difference was <0.001 throughout theinstrument's wavelength range.
The full versatility of the instrument is demonstratedby measurements of near-normal/hemispherical andnear-normal/near-normal transmittance of two opti-cally different transparent materials.
It is noted that the high confidence level of the in-strument is maintained by periodic interlaboratorymeasurements of the various generic types of reflective,absorptive, and transparent materials described. Theseinclude the MCCA/NBS collaborative reference pro-gram for color and appearance, the National Bureau ofStandards in informal intercomparisons, other com-mercial and federal laboratories in specially selectedareas, and, most important, regular and periodic de-termination of the measured reflectance of StandardReference Materials obtained from the National Bureauof Standards.
The authors are indebted to Cedric Currin of DowCorning Corp. whose support during the early devel-opment of the DSET MIS spectrophotometer was in-valuable. We are further indebted to Robert Heiskellof DSET for his painstaking efforts in construction ofthe MIS spectrophotometer. Finally, we are indebtedto Joseph C. Richmond and Jack J. Hsia of the NationalBureau of Standards for their review of the manuscriptand their many helpful suggestions.
References1. J. A. Jacquez and H. F. Kuppenheim, J. Opt. Soc. Am. 45, 460
(1955).2. D. K. Edwards, J. T. Gier, K. E. Nelson, and R. D. Roddick, J. Opt.
Soc. Am. 51, 1279 (1961).3. A. S. Toporets, Opt. Spectrosc. USSR 7, 471 (1959).4. B. E. Newnam, G. L. Brown, and E. E. Luedke, Investigation of
5. G. J. Kneissl and J. C. Richmond, "A Laser Source IntegratingSphere for the Measurement of Directional, Hemispherical Re-flectance at High Temperatures," in Thermophysics of Spacecraftand Planetary Bodies, Vol. 20, Progress in Astronautics andAeronautics, G. Heller, Ed. (Academic, New York, 1967).
6. G. J. Kneissl and J. C. Richmond, Natl. Bur. Stand. U.S. Tech.Note 439, Institute for Basic Standards (Feb. 1968).
7. M. W. Finkel, "Portable Reflectometer," in Thermophysics:Applications to Thermal Design of Spacecraft, Vol. 23, Progressin Astronautics and Aeronautics, J. T. Bevans, Ed. (Academic,New York, 1970).
8. F. G. Sherrell and F. Shakrokki, "Determination of HemisphericalEmittance by Measurement of Infrared Bihemispherical Reflec-tance," in Heat Transfer and Spacecraft Thermal Control, Vol.24, Progress in Astronautics and Aeronautics, J. W. Lucas, Ed.(MIT Press, Cambridge, 1971).
9. 0. E. Miller and A. J. Sant, J. Opt. Soc. Am. 48, 828 (1958).
