radiometric calibrations of portable sources in the vacuum ... · fusion research, ultraviolet...

Volume 93, Number 1, January-February 1988

Journal of Research of the National Bureau of Standards

Radiometric Calibrations of Portable Sourcesin the Vacuum Ultraviolet

Volume 93 Number 1 January-February 1988

Jules Z. Kiose, J. Mervin The radiometric calibration program their uncertainties. Finally, the calibra-

Bridges, and William R. Ott carried out by the vacuum ultraviolet tion services are delineated in an ap-radiometry group in the Atomic and pendix.

National Bureau of Standards Plasma Radiation Division of the Na-

Gaithersburg, MD 20899 tional Bureau of Standards is presentedin brief. Descriptions are given of the Key words: arc (argCn); arc (blackbody);primary standards, which are the hydro- arc (hydrogen); irradiance; lamp (deu-gen arc and the blackbody line arc, and terium); radiance; radiometry; Standardsthe secondary standards, which are the (radiometric); ultraviolet; vacuum ultra-argon mini- and maxi-arcs and the deu- violet.terium arc lamp. The calibration meth-ods involving both spectral radiance andirradiance are then discussed along with Accepted: October 2, 1987

1. Introduction

The vacuum ultraviolet (VUV) region of thespectrum has become important in several areas ofresearch and development. These include space-based astronomy and astrophysics, thermonuclearfusion research, ultraviolet laser development, andgeneral atomic physics research. Applications ofVUV radiation in chemical, biophysical, and medi-cal fields are widespread. Many applications re-quire knowledge of not only the wavelength of theradiation involved but also the intensity or flux ofradiation. This implies a calibration of some type.The calibration may be based upon a standardsource, i.e., one whose output is known, or a stan-dard detector, i.e., one whose response to a givenradiation level is known. Two general cases can bedistinguished. In the first case one wishes to deter-mine how much radiation a source such as the sunor a plasma device is emitting at a given wave-length. Usually, the source is not monochromatic,so a monochromator must be used to select the

desired wavelength. In this case the most directprocedure is to employ a standard source. Thesource to be investigated as well as the standardsource are set up in turn so that radiation from eachsource passes through the same monochromatorand optical elements. The calibration is performedessentially by a direct substitution of the standardsource for the one to be calibrated.

The second case occurs when one wishes toknow the flux in a monochromatic beam of radia-tion, such as the flux emerging from the exit slit ofa monochromator. For this determination a stan-dard detector is more appropriate; the flux is deter-mined by simply measuring the signal when thedetector is irradiated with the beam to be cali-brated. If one were to attempt to perform the cali-bration in the first case above using a standarddetector or in the second case using a standardsource, one would in both cases need to know thespectral efficiency of the monochromator used inthe measurement. This would require an additionalmeasurement using a second monochromator and

21



would introduce additional uncertainties and com-plexities. Therefore, a need exists for both standardsources and standard detectors.

Standard sources may be divided into primarystandards and secondary or transfer standards. Pri-mary standards are ones whose output is knownfrom basic principles. The primary standards ofVUV radiation include plasma sources, especiallythe wall-stabilized hydrogen and blackbody linearcs, and electron storage rings emitting synchro-tron radiation.

There are storage rings at several laboratories,including NBS, which are used as primary VUVradiation standard sources. They produce highly-polarized continuum radiation for wavelengths

0.03 anm. Limited access to storage rings andproblems with the emitted polarized light, how-ever, make it desirable to have available other stan-dard VUV sources. The wall-stabilized hydrogenand blackbody line arcs were developed to providealternative primary standard sources. Thesesources, however, are also not able to be easilyused for calibrations. Hence, secondary or transferstandards which are relatively easy to apply havebeen developed. These include the deuterium lampand the argon mini-arc. These sources are morereadily available and make possible relatively inex-pensive and convenient calibrations. Also for re-searchers having access to a storage ring, thesecondary standards are useful in making possiblemore frequent calibrations. Finally, the secondarystandards possess some useful properties not char-acteristic of the available primary standards suchas, for example, emission over a relatively largesolid angle.

The two principal radiometric quantities whichare measured and calibrated are radiance and irra-diance. For an object or source which emits radia-tion, the radiance is the radiant power emitted perarea per solid angle, L =(W cm-2 sr-'). If thesource emits a continuum, i.e., emits radiation at allwavelengths near a particular wavelength, thespectral radiance is the radiance per wavelength in-terval or bandpass, LA=(W cm2- sr-1 nm-'). Theradiance will in general vary over the source area,the direction, and the wavelength. The definitionassumes that the area, solid angle, and wavelengthband are small enough that the radiance does notvary greatly within these quantities. Irradiance isthe radiant power incident upon a target per area,E = (W cm-2 ), and spectral irradiance is the radiantpower incident upon a target per area per wave-length band, Ex=(W cm-2 nm'). A source of radi-ation may serve as a standard source of irradiance

by operating it at a given distance from the targetarea. Some sources may be used as either standardradiance or standard irradiance sources. A separatecalibration must be performed, however, for eachquantity.

The services performed by the Atomic andPlasma Radiation Division of the National Bureauof Standards include tests and calibrations ofportable secondary VUV standard radiance and ir-radiance sources. These are usually rare-gas dimerlamps, which emit continuum radiation over lim-ited wavelength ranges, and hollow cathode lamps,which emit spectral lines in the wavelength rangefrom the VUV through the visible. All sources aregenerally supplied by customers. The main groupsof customers have included those in the fields ofspace-based astronomy and solar physics who haveused standard sources to calibrate satellite, rocket,or balloon-borne spectrometers. Other customershave needed calibrations in the 100-300 nm rangefor plasma radiation studies.

2. Apparatus2.1 Primary Standards

2.1.1 The Hydrogen Arc A high temperature wall-stabilized steady-state hydrogen arc has been de-veloped as our primary standard of spectral radi-ance [1]. This type of arc lends itself to such a usebecause at sufficiently high temperatures it yieldsabsolute intensities independent of other radiomet-ric standards or of the accuracy of any plasma di-agnostics. Previous efforts at lower powers werehindered by large uncertainties in plasma diagnos-tics, a difficulty that has been overcome in the hightemperature arc. Figure I illustrates the UV spec-trum of several of the more common standardsources, including that part of the hydrogen arcspectrum that is the subject of this paper.

The method depends upon the phenomenon thatthe continuum emission coefficient for a stronglyionized hydrogen plasma which is in or close to thecondition of local thermodynamic equilibrium(LTE) is calculable to within one percent [2], Thisfollows from the fact that the essential spectro-scopic constants, i.e., the continuum absorption co-efficients and transition probabilities, are exactlyknown for atomic hydrogen. The continuum inten-sities emitted from a typical pure hydrogen wall-stabilized arc discharge in the spectral regionabove 91.5 nm are optically thin and a function ofthe electron density and temperature [3,4]. In thelow power hydrogen arc the electron density

22

VoLurne 93, Number 1, Jantuary-February 1985


lD' 1-

10-14

go-2 -

lC -3 -

L

I B L AB - 12 5 Y00 K,Iff BLACKBODY

BLACKBODY LIMITEDFROM 12500K ARC

LINES

Lyax

1cm HYDROGEN ARC

CARBON ARC -

30 WATT . /D2 LAMP I/

ITUNG~STENSTRI

i LAMPI IIlL

100 200)

xfnml300

plete, and any further increase in arc temperatureresults only in a gradual decrease in intensity be-cause of the decrease in the number density. Theweak wavelength dependence of this maximumseen in figure 2 occurs because of the change in theenergy distribution of the equilibrated electrons asthe electron temperature is varied.

7E

z

di

U-

2

LS-

4.0-

3.0- _ x

124 nm2.0-

JoF,09

OS-37 _

0-

400

o l-6 c00

Figure 1. Cormpadamn9 of spectral radiances fEr far t:V soures.The ogtput of tie Lydroger arc is given fSo she temrersatre ofmairmu coninuum emnti on; the Obtputs of the other sources

are at typical operating conditions.

and temperature are determined from plasma diag-nostics in the visible spectral region using availableradiometric standards [3]. These quantities are thenused to calculate the continuum intensities in theVUV. There are known to be significant uncertain-ties in the plasma diagnostics, and as a result wehave adopted as our primary standard a higherpower hydrogen arc which operates such that a2-mm diameter wall-stabilized discharge reaches atemperature of about 20,000 K. For these conadi-tions the continuum emission coefficient as a func-ticn of temperature shows abroad maximum whichshifts with wavelength as is shown in figure 2 ThisoptimumV condition is brought about essentially bythe compensating effects of an increase in the ion-ization Fraction and a decrease in the total numberdensity as the temperature is increased n a constantI-atm pressure operating environment. Beyond thismaximum [5,6] the ionization is practically com-

l" am

eo000Ti K)

20000 22030

Figure 2. The emission coefficient for a I-atm hydrogen plasmain LTE as a function of tenperature for several wave-lengths: 250amr (-); 200nr, (Rae); 150am (----2; and124 nn (- - >.

Figure 3 shows the measured radial dependenceof the hydrogen continuum emission coefficient at190 run and the corresponding calculated LTEtemperature (2] for an arc current of 80A Becausethe maximum in the coefficient is broad and theabsolute magnitude of the peak intensity is not verysensitive to the electron temperature, the emissioncharacteristics of the plasma are nearly homoge-neous over its certral region which extends toabout 0.6 mm from the arc axis. This is especiallysignificant if only a small sample of the region isobserved as indicated in the figure. First, it meansthat the alignment precision. which was iesponsi-ble for much of the uncertainly in the low powerarc method, is not critical here. Second, it meansthat the arc current necessary to obtain the maxi-mum emission coefficient is also not critical.

21

0,

EaOr

E;

3:

I-4

102

Volume 93, Number i, 3anuary-February 1985


I

I

o0o00 K

0.4 0.2 2 0o 2 I OR nl I

Figare 3. Radial dependence of the continuum intensity and arctemperature for a 2-mm wall-stabilized hydrogen arc operatingat 80 A. The dashed lines define the dimensions of the plasmathat is observed by spectrometers.

Additional advantages of operating the hydro-gen arc under such conditions are (1) the HILyman band molecular emission, which limited thelow power hydrogen arc to wavelengths longerthan 165 nm, is negligible; (2) the assumption ofLTE appears to be very closely fulfilled as shownby experimental consistency checks and theoreticalvalidity criteria [7]; and (3) the hydrogen arc as it isdescribed here constitutes a true primary standardof spectral radiance in that except for the minoruncertainties associated with the theoretical modelfor the plasma, knowledge of the absolute contin-uum intensities does not depend on any other stan-dards of calibration or sophisticated plasmadiagnostics and requires only a measurement of theambient pressure (z I atm) and the arc length. Theprocedure for applying the arc as a radiometricstandard involves only adjusting the arc current formaximum signal at the wavelength of interest. Theoverall system efficiency is given then by the ratioof the detector response to the hydrogen arc spec-tral radiance calculated using the maximum LTEemission coefficient at 1-atm pressure and the ac-tual physical length of the discharge.

Figure 1 shows the wavelength dependence ofthe maximum emission coefficient of the hydrogenarc plasma, and table 1 lists the values of the samequantity obtained through a calculation of the hy-drogen continuum [2]. The Stark broadened wingof the Lyman a line at 122 nm has been included inthese calculations and, as can be seen from thetable, becomes significant at wavelengths below140 rim [8]. The uncertainty in the calculated con-

tinuum emission coefficient is estimated to be about2%, due mostly to uncertainties in the high densityplasma corrections [2]. Combining this uncertaintyin quadrature with the uncertainties in the Starkbroadening calculations, the uncertainty in the to-tal emission coefficient comes out to be 2% above140 nrm, 8% at 130 rnm, and 13% at 124 rin. Theseuncertainties and all those succeeding unless statedotherwise are taken to be 2 standard deviations(2o-), i.e., as having a confidence limit of about95%.

Table 1. Maximum emission coefficient from a 1-atm hydrogenplasma in LTE calculated as a function of wavelength'

X(nm) c(W cm-' nmn< sr-') T(K) e

350 0.73 17,000 0300 0.68 17,000 0250 0.59 17,500 0.01200 0.424 18,500 0.07190 0386 18,800 0.11180 0.347 19,200 0.20170 0308 19,500 0.34160 0.269 20,000 0.65150 0.232 21,000 1.3140 0.200 21,700 4130 0.245 17,500 54128 0.372 16,500 77126 0.79 16,500 90124 3,22 16,000 98

'The percentage of this emission due to Ly a and she arc tem-perature required for maximum emission are also listed.

Three factors are of special concern in applyingthe high power hydrogen arc standard: the arclength, spatial resolution, and off-axis H2 emission.Figure 4 is a schematic of the hydrogen arc (1,4,9].The arc is struck between a set of four tungstenanodes and four tungsten cathodes that are sym-metrically located and share the current loadequally. At each end of the 2-mm diameter channelformed by the 20 water-cooled stacked copperplates, the discharge expands and separates intofour smaller arcs terminating at each of the elec-trode tips. The length of this inhomogeneous re-gion is made small compared to the total length ofthe arc, and the emission coefficient and tempera-tures are lower than in the channel. Therefore, itscontribution to the total arc signal obtained in anend-on measurement along the axis of the dis-charge is small but not precisely known. The mini-mum possible length of the arc is 5.05 cm, the totallength of the channel formed by the 20 arc plates,and the maximum length is 5.5 cm, the total dis-tance between the electrodes. The estimated length

24



of the homogeneous hydrogen arc plasma was thustaken to be 5.3 cm with an uncertainty of 5% (2(r).

Spatial resolution is not much of a problem forthe high power arc since the emission characteris-tics are nearly uniform in the vicinity of the arcaxis. The regions of the arc that fall into the coneof observation determined by the f/200 apertureand the 0.30 mm field stop of the observing opticsare illustrated in figure 3. Thus, knowledge of theemission coefficient as a function of radial positionallows the calculation of the effect of spatial aver-

aging. The main result is that the integrated emis-sion coefficient & has a maximum about 1% lowerthan Ema from a truly homogeneous plasma with anaxis temperature T(&) slightly higher than T(EcD.The arc plasma is cylindrically symmetric, and be-cause the area of the differential shells increases as2lTrAr, the off axis regions are weighted more thanthe axis region. Therefore, when one obtains maxi-mum signal, the current has been adjusted so thatthe temperature maximum occurs slightly off-axisas seen in figure 3.

VACUUM

ARCPLATES

ELECTRODE

water

water

gas 4

AIR

Figure 4. Schematic of the hydrogen wall-stabiiized arc. A cutaway top view drawing of one of the arcplates illustrates the flow of cooling water through specialiy machined channels within each plate. Thesymbols in the four corners show the direction of water flow.

25

Volume 93, Number

Journal of Research of the1, January-February 1988

National Bureau of Standards

Below 165 nm the Lyman bands of H2 are emit-ted from low temperature off-axis regions. Figure 5shows a radial scan of the arc at a wavelengthwhere one of the stronger H2 features can be ob-served. The off-axis line peak is about 100 timeslarger than the on-axis hydrogen continuum, butclearly if the observation column is restricted to nomore than the central 0.5 mm diameter plasma, H2lines should not be observed. Molecular emissionfrom the inhomogeneous end regions also is a pos-sibility. However, none is observed since the arcterminates at points that are out of the line of obser-vation (the cooler plasma regions are off-axis) andthe end-layer on-axis temperature gradient is quitesharp (the length of the cool plasma is thereforevery short).

SPATIAL RESOLUTION

GAIN |X 10

-1.0 0 5 0 0.5 +1.0

R(mm)Figure 5. The radial dependence of the hydrogen arc spectralradiance at 160.6 nm for an arc axis temperature of about19,500 K. The intense off-axis radiation is due to Lyman bandmolecular emission. The nearly uniform radiation in the vicinityof the axis contains no molecular contributions, but is due to thehydrogen continuum that is being used as a spectral radiancestandard.

In figure 6 the calculated spectral radiance of thehigh power hydrogen arc is compared with cali-brations of that arc made with several other stan-dard sources [1]. The theoretical spectral radianceis the product of the theoretical maximum hydro-gen plasma emission coefficient and the arc length.This quantity is graphed as a shaded area whichrepresents the rms uncertainty due to the previ-ously described uncertainties in both the calculatedemission coefficient and the arc length. The solidand open circles represent the measured continuumintensities using, respectively, a calibrated tungstenstrip lamp [10] and blackbody limited lines fromanother arc for spectral radiance normalization.The crosses represent the measured intensities us-ing the low power hydrogen-arc standard. The er-ror bars attached to the data points representuncertainties associated not with the hydrogen arcbut with the various standards used to calibrate thespectral radiance. The discontinuity in the wave-length scale is due to the interruption of the hydro-gen spectrum by the Balmer line series of atomichydrogen, which dominates the high temperaturearc spectrum above 360 nm except for a small re-gion around 560 nm between Ha and Hp. Inspec-tion of figure 6 leads us to conclude that thevarious standards are consistent with one anotherover the entire range of comparison.

In summary, a high power atmospheric pressurehydrogen arc capable of operating at temperatureson the order of 20,000 K has been examined theo-retically and experimentally and found to be suit-able as a primary radiometric standard in the VUV.The experimental investigations have shown that atsuch temperatures molecular hydrogen emission atwavelengths shorter than 165 nm is negligible, andthe hydrogen plasma continuum emission coeffi-cient can be measured throughout the 124-360 nmspectral range. By operating the arc at currentssuch that the maximum emission coefficient isreached, the uncertainties associated with variousplasma diagnostics and alignment imprecisions areminimized to the one-percent range. The results ofa comparison with other available standard sourcesare consistent with the estimated 2oa uncertainty of±+5% in the hydrogen arc spectral radiance be-tween 140 and 360 nm. This uncertainty is duemainly to the uncertainty in the measurement ofthe arc length. Below 140 nm the Lyman a Starkbroadened wing becomes significant, and the esti-mated uncertainty in the hydrogen arc intensitiesincreases to about ±9% at 130nm and ±14% at124 nm due mainly to uncertainties in the plasmaline broadening theory. These results are also

26

Votume 93, Number

Journal of Research of the1, January-February 1988

National Bureau of Standards

I I I I I I VI _

4

-i1; I lil Tlrifil

it

i F'il I I I I , I I

IO 140 I1O ,20 200 220 24C 260 220 300 320 340V 550 60 570

A [nml

Figure 6. Maximum spectral radiance vs wavelength calculated for a 5.3-cm length hydrogen arcplasma in LTE at I arm pressure. The calculation is shown as a grid and represents the rms uncertaintydue to uncertainties in the arc length (±5%) and the calculated emission coefficient (between 2% and13%). The points are actual measurements of the spectral radiance calibrated either with a tungsten striplamp (.), blackbody limited lines (o), or a low power hydrogen arc (x).

consistent with a comparison standard. Finally, allexperimental comparison data confirm that the hy-drogen arc spectral radiance is calculable through-out the 124-360 nm range without dependence onany other radioimetric standard or plasma diagnos-tic technique. Therefore, the high power hydrogenarc is a true primary standard. A detailed descrip-tion of the hydrogen arc is given elsewhere [4].2.1.2 Blackbody Line Arc The short wavelengthlimit of the hydrogen arc at about 124 nm is deter-mined by the onset of the Stark-broadened, opti-cally thick Lyman line series of atomic hydrogenthat dominates the spectrum between 94 and124 mm. Since the short wavelength cutoff of mag-nesium fluoride windows which are used in manyVUV instruments is at about I 15 nm, it would beconvenient to have a primary standard which cov-ers the range 115 to 124 nm. This need has beenmet with the development of the blackbody linearc [11-13].

Blackbody radiances for a number of prominentultraviolet spectral lines are determined in the fol-lowing manner. A wall-stabilized steady-state arcsource is operated at about 12,000 K in argon withsmall admixtures of oxygen, nitrogen, and carbondioxide. The admixtures produce very strong emis-sion for their principal resonance lines, which allhappen to lie conveniently in the wavelength rangefrom 115 to 250 am. These extremely strong reso-

nance lines reach the blackbody intensity limitseven at very small concentrations of the elementsin the plasma, i.e., they become "optically thick."More precisely, the central regions of these lineprofiles reach, for a small wavelength range of afew hundreths of a nanometer, the intensity a black-body would have at the arc temperature. Thus, thearc represents, for a few narrow wavelength bandsin the UV spectrum, a very high temperatureblackbody. To utilize this arc as a stan-dard source, the temperature must be accuratelydetermined using standard plasma spectroscopictechniques including absolute radiometry in thevisible region of the spectrum. Subsequently, usingPlanck's radiation law, absolute intensities are es-tablished for these narrow spectral bands and areutilized as calibration points. Calibrations at otherwavelengths are usually found by interpolation.

This technique has been used in a number of ex-periments and has been found to be reliable, withuncertainties ranging from about 10% at 250 no toabout 25% at 115 nm. For general use, however, ithas the significant deficiency that large areas of thefar UV have to be covered by interpolation sincevery few blackbody limited lines are found outsidethe ranges 115 to 126 nm and 146 to 175 nm. Sincewe are primarily interested in the range 115 to124 nm, this is no great handicap here.

27

-10

Er

Sa

3

Cc

WIa Ox

I � f � i f



The accuracy of the blackbody line method de-pends critically on the measurement of the arc tem-perature, which is applied in the Planck function todetermine the intensity of the blackbody ceilings ofthe optically thick resonance lines. It has been esti-mated that no single spectroscopic method is suffi-cient to determine the temperature of such aplasma with small 02, N2, and CO, additions towithin an uncertainty of ±2% [11]. This representsan uncertainty in the blackbody ceiling of about10% at 250nm and about 25% at 115 am as wasmentioned above. To reduce this uncertainty, itwas recommended that several methods be applied,and the results averaged. However, a singlemethod, which is thought to be superior, wasadopted here to determine the blackbody tempera-ture.

The blackbody ceiling of the C i emission line at247.9 nm is measured with a UV-calibrated tung-sten strip lamp. The uncertainty in the spectral ra-diance of the strip lamps calibrated by NBS at thiswavelength is about 2%. The peak of the247.9 nm C I line is actually calibrated in twosteps: first, the continuum emitted by the arc mix-ture at 250 nm is measured and calibrated at lowspectral resolution using the tungsten strip lampand the air-path predisperser and monochromator;and second, the peak of the carbon line is cali-brated with respect to this continuum using theVUV instrument at high resolution. The differencein the VUV system efficiencies between 247.9 and250 nm is minimal but is measured with the hydro-gen arc and taken into account. Care is also takento ensure that the contribution of the line wing at250 nm in the low resolution air system is negligi-ble. For a 4.7 nm diameter arc at 60 A the axis tem-perature of the argon plasma with admixtures of N,C, 0, and H was measured to be 11,900 K. A tem-perature uncertainty of ±160 K is estimated due toan estimated rms uncertainty of about 6% in theblackbody intensity determination at 247.9 nm.This results in an uncertainty of 10% in the black-body line radiances between 115 and 140 nm.

Besides the temperature of the plasma, the char-acter of the blackbody limited lines is also critical.For example, if the boundary layers between themixed plasma and the pure argon buffer are toothick, the radiation transfer may be complicated bythe density and temperature gradients through thelayers. Such conditions are more the rule than ex-ception and are easily detected at high resolution asstructure in the line peaks. However, by appropri-ate flow rate adjustments, this effect can be mini-mized so that the blackbody plateaus are essentially

flat-topped. Referring to figure 6, we see that for ahigh power hydrogen arc the calibrated spectralradiances obtained using blackbody limited linesare consistent with the corresponding calculatedquantities, as was noted in section 2.1.1.

2.2 Secondary Standards2.2.1 Argon Mini-Arc

Physical Principles After development of the hy-drogen arc as a primary standard of radiance in theVUV, it was realized that a simpler, portable, moreeasily operated, and lower powered transfer stan-dard would be of great value. The hydrogen arc iscomplicated and difficult to operate, and requires amassive highly stable dc power source rated at1200V and 110A. A secondary standard whichcould be compared with our primary standard andthen transported to the user's laboratory was rec-ognized as being essential.

The most widely used transfer standard at thattime was the commercially available deuteriumlamp. Although the deuterium lamp has attractiveproperties such as its relatively strong continuum,low power requirements, and small size, it also haslimitations. First, its region of applicability as a ra-diometric standard is restricted to wavelengthsabove 165 nm due to the presence of a many-linemolecular deuterium band system below 165 nm.Second, the lamp exhibits a variability related tothe positioning of the discharge on the electrodesfollowing each ignition. Finally, the deuteriumlamp has aging characteristics which are not wellknown. At the least, this means that the lamp mustbe frequently recalibrated to ensure accuracy.

To meet the lack of adequate transfer standardsin the far UV, the argon mini-arc was developed atNBS [14]. This source can be conveniently appliedas a radiometric transfer standard without the limi-tations of the deuterium lamp. The wavelengthrange of the argon mini-arc overlaps the lowerrange of the tungsten strip lamp in the near UV andextends beyond the short wavelength limit of thelow pressure deuterium lamp at 165 nm. The mini-arc was designed to fulfill, insofar as possible, thefollowing goals: 1) an intense line-free continuousspectrum between 115 and 330 nm; 2) stability andreproducibility over many hours of operation;3) light source and power supply both portable;4) uniform output over a large solid angle;5) radiant power output adjustable over a range ofseveral decades; and 6) simple alignment and oper-ation. We now proceed to discussions of the arcconstruction and operating characteristics.

28



Description of the Arc Figure 7 is a photograph ofthe arc source. The model described here was de-signed specifically for use as a secondary radiancestandard and is different in some respects from arcsources previously designed for other purposes. Itwas developed after experimentation as the sim-plest model which meets the requirements listedabove. Argon was chosen as the operating gas be-cause of its suitability in providing a stable dis-charge and an intense, line-free UV continuumwith minimal power requirements. In the pho-tograph the arc is mated to a monochromator witha stainless steel bellows. Water cooling connectionsand a triple-tube manifold for supplying argon arealso shown. The two vertical threaded posts on thetop of the arc are the electrical connections to theanode and cathode. The arc is mounted on an ad-justable table which allows precise translation androtation about two axes. All sources utilized in theNBS VUV radiometry program are mounted onthis type of table.

As shown in figure 8, which is drawn to scale,the arc source is constructed essentially of fivecopper plates separated by silicone rubber insulat-ing rings and clamped together to form the device.The central plate forms a channel which guides

and constricts the discharge. The plates adjacent tothe central plate contain the anode, a 3.2-mm di-ameter thoriated tungsten rod pressed into its plate,and the cathode, also of tungsten but mounted sothat it can be moved into and out of the arc chan-nel. The cathode was made adjustable for ease inigniting the arc. All five plates are water-cooledwith the water flowing inside each piece throughholes drilled in the sides. These holes are seen inthe photograph but not in the diagram. The arcconstricting section is 6.3 mm thick with a 4.0 mmdiameter hole. Diameters smaller than this wereconsidered unacceptable since they caused a higherplasma temperature and significant Ar Ii line emis-sion in the wavelength region of interest. Largerdiameters were rejected since the radiant powerfor a given current decreased considerably. Thearc plasma is observed end-on through the holes inthe electrode pieces, and the radiation calibrated isthat emitted along the arc axis and emergingthrough the magnesium fluoride window on thecathode side. This window is set back to avoid pos-sible contamination from the discharge. The gaspurity necessary to maintain arc stability and re-producibility of the continuum emission is ensuredby operating with a continuous flow of argon.

Figure 7. A photograph of the argon mini-arc mated to a monochromator.

29



ARGON GASINPUTS(1 ATM)

Figure 8. A schematic of the argon mini-arc light source.

Maintaining a high degree of argon purity withinthe arc chamber also minimizes the radiant powerof atomic resonance lines in the spectrum whicharise from small (ppm) concentrations of oxygen,nitrogen, carbon, and hydrogen. These elementsare present due to air and water vapor in the arcchamber and gas handling system. The positions ofthe gas inlet and outlet ports are chosen to keep allsections of the arc chamber thoroughly purged.The argon is admitted to the inlet ports throughplastic tubing connected to a flowmeter. Under theflow rate used here, the pressure in the chamber isunchanged from the ambient atmospheric pressure.The arc may be operated most conveniently withthe outlet ports open and the data given belowused to account for local and temporal differencesin atmospheric pressure, if necessary.

Optical alignment can be performed by sightingwith a telescope or laser beam down the arc axis.The distance for proper focus should be measuredto the center of the middle copper piece.

The arc discharge is initiated by applyingvoltage between the electrodes and then inserting atungsten rod, which is externally connected to theanode potential, until it touches the cathode pro-truding into the arc channel. The discharge istransferred from the tip of the rod to the anode asthe rod is withdrawn. Finally, the cathode is with-drawn from the channel as shown in figure 8.Power for the discharge is furnished by a currentregulated supply, the size of which is determinedby the current required. For example, the powersupply used in some of our experiments was a1.2 kW, high efficiency switching regulator weigh-ing only 10 kg. For starting the arc without diffi-culty, it was found necessary to use a ballast

resistor of about 0.5 fl in series and to apply a po-tential of at least 40 V. The arc can be started atcurrents above 10 A, and after ignition the resistormay be shorted out if desired. For currents above10 A the voltage drop across the arc is nearly con-stant at 28 V.

Wall-stabilized arcs employed for end-on mea-surements usually are constructed of many con-stricting plates or disks, in order to obtain a nearlyhomogeneous plasma extending along the axis tothe boundary regions near the electrodes at eachend. These boundary regions emit a negligible partof the radiation emerging along the arc axis. Thehydrogen and blackbody arcs which we use forprimary radiance calibrations are necessarily of thistype, since the lengths of the emitting plasmacolumns must be well defined. In such cases, how-ever, the solid angle over which uniform radiationcan be observed is small (f/200 is used for calibra-tions with the hydrogen and blackbody arcs) andnot convenient for the calibration of many opticalsystems. For a transfer standard, on the other hand,the arc length does not need to be precisely knownas long as it is a reproducible quantity. Thus, inorder to provide a much greater solid angle, theargon mini-arc was made with only one constrict-ing plate. Measurements show that this construc-tion provides a rather large angular beam ofradiation (f/10) having a practically uniform inten-sity distribution [14]. This should be considered anadvantage even if one does not require such a largesolid angle, since having the larger available anglemakes the angular alignment of the mini-arc lesscritical.Spectrum Figure 9 illustrates the spectral radi-ance of the argon mini-arc light source for two arccurrents: 25.0 and 50.0 A. The continuous spec-trum of the argon arc arises primarily from atomicrecombination radiation. Near the short wave-length end the molecular continuum mentionedabove contributes to some degree, and below125 nm the increased output is from the wings ofthe argon resonance lines at 106.7 and 104.8 nm.Below the MgF2 window cutoff, one should expectto have strong emission on the wings of the reso-nance lines with complete absorption at their cen-ters. Below 80 nm there should be completeresonance continuum absorption by ground stateargon atoms.

The spectral radiance was determined by directcomparison to the NBS wall-stabilized hydrogenarc [1] between 130 and 330 nm and to a plasmablackbody line radiator below 130 nm [1,3,11].Data were taken with a spectral resolution of

30

Volume 93, Number 1, January-February 19S8


0.01 rm. The temperature of the blackbody line ra-diator was determined to be 11,800±100 K by di-rectly measuring the blackbody ceilings at 193, 174,165, 149, and 146 nm, using the hydrogen arc as aprimary standard of spectral radiance. With thistemperature the radiant power of other blackbodylimited lines emitted below 130 nm could then becalculated. For the calibration of longer wave-lengths (>210 nm), the second-order spectrumfrom shorter wavelengths was eliminated by purg-ing the small volume between the arc and the spec-trometer with oxygen. Below 210 nm, argon wasused as the purge gas.

10-1

_ 1027

Ea

C

j<10-.

1i- 4 L_110 150 190 230

X [ni

270 310 350

Figure 9. Spectral radiance as a function of wavelength for amini-arc with an arc plate diameter of 4,0 mm operated at twodifferent currents: 50.0 and 25.0 A.

The spectral radiance values illustrated in figure9 and listed in table 2 apply to the plasma radiationemitted along the axis of the mini-arc which is im-aged with a focusing mirror on a 0.30 mm diameteraperture (magnification= 1) at a solid angle V200.The basic uncertainty in the spectral radiance is+5% above 140 nm and + 10% below 140 nm dueto uncertainties in the primary radiometric stan-dards. Also included in table 2 for reference pur-poses is the measured transmission of the MgF2

window used on the mini-arc. If one wishes toknow the spectral radiance of the plasma itself, thefigures in the table must be divided by the transmis-sion. Known systematic deviations or uncertaintiesin the radiant power due to different operating orimaging conditions are described elsewhere [14].

Table 2. Spectral radiance as a function of wavelength for a50.0-A argon mini-arc light source with an arc plate diameter of4.0 mm'

Whnm) Lx(mW cMm T X(nm) Lx(mW cm-2 Tnm-' sr'1) nin' sr-')

330.5 70 0.945 198.0 26.6 0.900325.5 70 0.943 192.5 24.0 0,900320.5 69 0.942 191.0 23.5 0.894315.5 69 0.942 186.0 20.8 0.891310.5 68 0.941 181.0 18.4 0.883305.5 67 0.940 176.0 16.1 0.873300.5 66 0.939 173.5 15.0 0.870295.5 64 0.939 170.0 13.6 0.860290.5 63 O.938 166.5 1 1.8 0.850285.5 62 0.938 165.0 11.1 0.848280.5 60 0.938 162.0 9.9 0.840275.5 58 &.938 159.5 9.3 0.826270.5 57 0.937 157.0 8.3 0.810265.5 55 0.937 155.0 7.8 0.798260.5 54 0.937 153.5 7.3 0.785255.5 52 0.935 151.0 6.5 0.765250.5 50 0.933 i48.5 5.9 0.740245.5 48.1 0.932 147.0 5.4 0.724240.5 45.2 0.931 144.0 4.73 0.700235.5 43.3 0.928 139.5 3.81 0.672230.5 40.9 0.925 135.0 3.38 0.655225.5 39.1 0.922 129.5 2.77 0.662220.5 36.3 0.920 127.0 2.65 0.642215.5 34.6 0.917 123.5 2.54 0.626210.5 32.4 0.914 118.5 2.77 0.597205.5 30.3 0.910 116.0 3.34 0.530200.5 27.9 0.900 114.5 3.74 0.439

-The measured transmission T of the are window is also listed.The radiance of the plasma itself is obtained by calculating thequantity LAT<.

There are no Ar r lines in the spectrum between114 and 330mm. Below 200 nm there are severalnarrow Ar iti lines and atomic nitrogen, carbon, andoxygen lines in the mini-arc spectrum due to airimpurities in the argon gas (99.999% pure). In addi-tion the hydrogen Lyman a line at 121.6 nm ispresent due to trace quantities of water vaporthroughout the system. These impurity lines can beseen in figure 10 which is a photoelectric scan ofthe spectrum between 115 and 320 m. Thehalfwidtbs of the lines are on the order of 0.01 urn,and the lines are well separated. Their presencehas no significant influence on the ability to makeabsolute continuum measurements except when

31



33 MD 2 230

RO~UNGTH (;-)

n0o IOS

as ,,, us - as Ie In I= IsII'OGIOMO (==)

Figure 10. A photoelectric scan of the spectrum of the argon mini-arc between 115 and 320 nm, takenwith a 0.01 nm spectral resolution and with a solar blind phototube detector.

extremely coarse wavelength resolution is used, aswould be the case, for example, if the monochro-mator were replaced by relatively wideband filters.All lines observed in the spectrum are listed intable 3. From this list one may readily determine if,for a given wavelength and bandpass, one is free oflines. This precaution is necessary since the amountof trace impurities in the arc and the correspondingline intensities may vary somewhat from one sys-tem to another.

Figure 11 compares the spectrum of the mini-arcwith several other light sources which have beenapplied as spectral radiance standards. Included inthis figure is a representation of the spectra of thetwo primary standards, the hydrogen wall-stabilized arc and the blackbody line radiator,which were used to calibrate the mini-arc spec-trum. The figure clearly illustrates the main advan-tages of the new transfer source. The mini-arc hassubstantially larger output and longer UV wave-length range than either the deuterium lamp or thetungsten strip lamp. Also, scattered visible light isnot a significant factor, as it is when the tungstenlamp is used as a UV light source. The visible lightfrom the argon arc is of the same order of magni-tude as the near UV light, and there is less than afactor of 100 difference between the radiance at110 and at 330 nm.

2.2.2 Argon Maxi Arc A recent advance in thearc standards program at NBS is the developmentof the argon maxi-arc. This source is essentially ahigh powered version of the mini-arc with three orfour arc confining plates instead of one and powerratings of 5-10 kW instead of 1.5 kW. The maxi-arcwith three plates has an irradiance approximately30 times that of a mini-arc and was designed to

- enable calibrations to be performed at a level com-parable to the solar irradiance in the near UV (250-350 nm). The discussion of the mini-arc givenabove generally applies also to maxi-arcs except forthe powers and radiances. Figure 11 gives a com-parison of the spectral radiance of the maxi-arcwith that of several other UV primary and transferstandard sources. Several of these arcs have beensupplied for use in calibration of space experiments.

2.2.3 Deuterium Arc Lamp

Introduction A source which has advantageousproperties as a radiometric standard and is quiteconvenient to use is the molecular deuterium arclamp [15]. This lamp has considerable UV radiantpower, is light and compact, is low-powered(30 W), and with maximum radiant power at190 nm has a very favorable ratio of UV to visibleradiant power. This last fact is important in

32

DO

a

A0ata

I I 1 I I I I Igo

a11


Journal of Research of the National Bureau of

emission lines appearing in spectrum of

Radiance relativeX(mn) Identification to continuum'

115.22 o I 1116.79 Ni 1117.69 N i 0.3118.94 CI 0.3119.38 C I119.99 NE 30121.57 HI 100124.33 N i 5126.13 C I 2127.75 C I 5128.98 C I 1130.22 oI 15130.55 0o 10131.07 N I X

131.95 Ni 0.5132.93 C I 4133.53 C II 8135.58 C I 0.2141.19 N i 0.7143.19 C I 0.1145.91 CI 0.x 146.33 C I t

146.75 c I o.i148.18 C 1 0.2149.26 No 4149.47 Ni 6156.10 Ci 4157.50 Ar i1 0.2160.04 Ar i 0.2

160.35 Ar i 0.1160.65 Ar i1 0.05165.72 C I 2174.27 Ni 1174.53 N i 0.5175.19 C I 0.1187.31 Ar i 0.05188.90 Art 0.1193.09 C I 2247.86 Cl 0.05

*In column three are listed line radiances relative to the contin-uum, measured with 0.25-nm spectral resolution

avoiding systematic errors due to visible scatteredlight effects when making measurements in theUV. Its disadvantages are that it is restricted towavelengths above 165 nm due to the presence of amany-line molecular band system below 165 nm; itmay exhibit a variability in the positioning of thedischarge on the electrodes following each igni-tion; and its aging characteristics are not wellknown. On balance, however, the advantages ofthe deuteriurn lamp outweigh its disadvantagesmaking it an essential component of any programof UV radiometry. The following discussion willdescribe the calibration of the deuterium lamp forspectral irradiance in the range 167-350 nm.

_.Y ARGON MINI ARC

/2D2LAMP1o-3~ -

10-4 TUNGSTEN STRIP LAMP

110 150 190 230 270 310 350A [rnm]

Figure 11. Comparison of the spectral radiance of several UVprimary and transfer standard sources.

Discussion of the spectral radiance calibration ofthe deuterium lamp is not presented here since it isnot a standard calibration service offered by NBS.However, this type of calibration can be requestedunder the, category "Special Tests of RadiometricDevices in the Near and Vacuum Ultraviolet."Description and Operation of the Lamp A sche-matic illustrating the operation of the deuteriumlamp is shown in figure 12. In order to start thelamp, the cathode coil is first heated for 5 s by a dcpower supply (10 V at 0.8 A) in order to providesome free electrons which facilitate initiating thedischarge. When a voltage of about 400 V is ap-plied to the lamp, an arc forms between cathodeand anode in the general form of an L. Most of theUV light is generated at the constricting aperture(1-mm diameter in our case) located in front of theanode. The main dc power supply is a 500 V,

33

Table 3. Survey ofmini-arc

Standards



300-mA constant current supply, with 0.1% cur-rent regulation. A ballast resistor (I kfI, 100 W) isused in the anode circuit because most power sup-plies cannot react fast enough to maintain a stablearc with only the lamp in the circuit. After the arcis started, the voltage across the arc drops to about100 V. At this point the heater current is switchedoff, and the lamp output stabilizes in 20 minutes orless. If the lamp is switched off, it should be al-lowed to cool back to room temperature beforerestarting.

5(A) =E /(w)A ( A 2 s nm} A (cm2)-.f(sr)-AX(nm),

(1)

where A is the area, Q the solid angle, AN thewavelength bandpass, LA the spectral radiance, andLE the efficiency of the detection system. If sub-script R refers to the standard source and subscriptx refers to the one to be calibrated, then

(2)SR ELXR * AR, * * &AXR

If A, Q, and AX are identical for each source, thenSUPRASILWINDOW

1 K Q, 100 WBALLAST RESISTOR

- 500 V, 300 mAPOWER SUPPLY

10 V, 0.8 A FILAMENTPOWER SUPPLY

Figure 12. Schematic illustrating the operation of a deuteriumlamp. The radiation is measured through the Suprasil windowsealed to the quartz lamp envelope.

3. Calibration Methods3.1 Radiance Calibrations

3.1.1 Introduction Spectral radiance, La, is de-fined in the Introduction as the radiant power emit-ted by a source per area per solid angle perwavelength interval: Lx=[Wcm-2 sr' nm-']. Acalibration of radiance is performed by directlycomparing the source to be calibrated with a stan-dard radiance source. An optical system includinglenses and/or mirrors is used to limit the geometri-cal quantities which are area and solid angle and todirect the radiation to a detector. A monochroma-tor selects the wavelength band desired. First, onesource is placed in position and signals are mea-sured at the various wavelengths. This source isthen replaced by the second source and the mea-surements are repeated. The radiance of the sourceto be calibrated is then obtained from the ratio ofthe signals times the radiance of the standardsource. In general the signal obtained from asource is

LxA= (S) -L R (3)

3.1.2 Argon Arc The hydrogen arc primary stan-dard is generally used to calibrate an argon arc.After measuring its signals, the hydrogen arc is re-moved and replaced with the arc to be calibrated.A telescope is used to align the argon arc, and onceit is operating, a fine adjustment in position is ac-complished by translating the arc both horizontallyand vertically to maximize the signal. As with thehydrogen arc, there is a volume between the arcsource and the vacuum window. This volume ispurged with argon at a flow rate of 5 L/min. Thepurging must continue for at least 10 min beforemeasurements are begun in order to sufficientlydisplace the air from this region. This flow rate iscontinued during the entire calibration. By moni-toring the signal at 170 nm where air absorption isstrong, the rise in signal during purging may beobserved, and the time at which essentially all air isdisplaced can be determined. There are two casesto consider: 1) if the arc to be calibrated containsa window, the purge region is isolated from thearc, and any purge rate which is sufficient to main-tain the region free of air may be used; and 2) if thearc to be calibrated is one without a window, theflow in the "purge" region is not independent ofthe arc, and the same flow arrangement must beused whenever the arc is used as a secondary stan-dard. This is necessary because the flow rate in thepurge region can have an effect on the arc opera-tion and hence on the radiance of the arc. Once theargon arc is operating and the purge region is clearof all air, signals are measured for 124<X<360 nm,the range of the hydrogen arc. In addition to thewavelengths covered by the hydrogen arc, signalsare also measured at shorter wavelengths extending

34

Volume 93, Number 1, January-Febroary 1955


to 115 un. In this region, however, there are nu-merous lines in the spectrum from minute air impu-rities. These lines are typically resonance lines andare extremely strong even at very low concentra-tions. The wavelengths at which the continuum ismeasured are selected to avoid interference fromthese relatively strong lines. Also, in order to avoidinterference from the lines, the wavelength band-pass must be no greater than 0.2 nm (700 rm slitson the 3 m monochromator). The continuum fromthis short wavelength region is calibrated using asecond primary standard, the blackbody line arc.3.1.3 Blackbody Line Arc Another primary stan-dard source, the blackbody line arc, is used to ex-tend the calibration to wavelengths below 124 nm.This is also a wall-stabilized arc, operated with ar-gon but including small admixtures of H, 0, C, N,and Kr. These admixture elements furnish spectrallines which can be made to be optically thick, i.e,they reach a maximum radiance value at the wave-length of the line.3.1.4 Tungsten Lamp Standard In the calibrationof radiance one additional standard source, thetungsten strip lamp [10], is applied. The tungstenlamp is used for two reasons: 1) to increase theaccuracy of the calibration, and 2) to extend thecalibration to longer wavelengths. The wavelengthranges of the hydrogen arc and tungsten lampoverlap in the near UV region, and although theuncertainty in the radiance of the hydrogen arc is±+5%, tungsten lamps are calibrated for radiancewith an uncertainty of ±2.3% in the near UV.Therefore, the calibration of a source extending tothe near UV may be increased in accuracy by com-paring it directly to the tungsten strip lamp. If thisis done, a calibration using the hydrogen and black-body line arcs is considered to furnish only the rel-ative spectral distribution of the radiance of theunknown source. A calibration with the tungstenlamp at the long wavelength end of the range ofthe hydrogen arc then serves to set the absolutescale of the radiance.3.1.5 Calibration of Argon Arcs Relative to an NBSArgon Arc For most calibrations the hydrogenarc and blackbody line arc are not employed, butrather calibrations are performed using a mini-arc(the NBS argon arc) as a transfer standard. Thetransfer standard is employed since calibrationswith the hydrogen and blackbody arcs are muchmore difficult and time consuming than calibra-tions performed relative to an argon arc. More-over, the comparison of two similar sources (twoargon arcs) can be accomplished in reasonable timewith less error than is involved in a calibration in-

volving the hydrogen and blackbody line arcs.Therefore only a slight decrease in accuracy resultsfrom employing the transfer standard. The calibra-tion of one argon arc relative to another proceedssimilarly to the calibration of an argon arc relativeto the hydrogen arc. The NBS argon arc is oper-ated without a window using a purge rate of5 L/min. The arc to be calibrated usually con-tains a window, and thus the calibration of the arcmust include a separate measurement of its trans-mission.3.1.6 Argon Mini-Arc Spectral Radiance Stan-dard An argon mini-arc has been designed, tested,and operated as a transfer source of spectral radi-ance for the wavelength range from 114 to 330 nm.Calibration has been performed using two primarystandard sources: the hydrogen arc from 130 to330 unm and the blackbody line radiator from 114 to130 nn. The mini-arc has the following principalfeatures: 1) a steady-state reproducible sourcewith dc power requirements of less than 1.5 kW;2) an intense continuum which is line-free between194 and 330 nm and which has only a few narrowand widely spaced impurity lines between 114 and194 nm; 3) a uniform output over a solid angle aslarge as f/9; 4) negligible aging effects over at leasta 24 hour period; and 5) uncertainties (2o) in theabsolute spectral radiance of 5.3% above 140 nmand 10. 1% between 114 and 140 nm. The radiantpower emitted by the mini-arc is influenced pri-marily by the arc diameter, the arc current, and thetransmission of the UV window material.

3.2 Irradiance Calibrations

3.2.1 Establishment of Irradiance Scale: GeneralMethod Spectral irradiance is the radiant powerincident upon a target area per area per wave-length band: E =[W cm-2 nm-']. A source of radi-ation may serve as a standard source of irradianceby operating it at a given distance from the targetarea. If one had a radiance source with uniformradiance over its emitting area, the irradiance at agiven distance from this source could be easilycomputed. In general however, radiance sourcesare non-homogeneous, and the radiance is knownonly for a small portion of the source, i.e., an areaelement which can be approximated to be homoge-neous. Thus, the irradiance at some distance fromthe source can be determined only if radiation fromoutside the calibrated area is prevented from reach-ing the measurement system. This can be done bythe use of optical imaging and/or collimating aper-tures. For applications in the VUV, the method

35

Volune 93, Number I, January-Febrary 1988


using collimating apertures [16] has proved to bemore practical and is the basis for the measure-ments described here.

The collimating aperture method is as follows. Aradiation source that is homogeneous over a cer-tain emitting area and whose spectral radiance hasbeen previously determined is situated a given dis-tance from a monochromator. A pair of apertures,one at the entrance slit of the monochromator (thefield aperture) and the other as close to the sourceas possible (the source aperture), is chosen so thatonly the radiation from the homogeneous portionof the source is measured. The spectral irradianceat the field aperture is given by the product of theknown spectral radiance of the source and a geo-metric factor dependent on the aperture dimen-sions and their locations relative to the source. Thisgeometric factor contains the information on theeffective area of the emitting source, the solid an-gle of the radiation beam incident upon the fieldaperture, and the irradiated area.

In principle, these quantities can be determinedby measurement. However, a direct determinationof the geometric factor is unnecessary, as is shownby the following discussion. The response of aspectroradiometer as a function of wavelength to asuitably collimated radiance standard is essentiallya measure of the system detection efficiency on arelative scale. If the same spectroradionmeter is irra-diated with an unknown source and the response ismeasured again, the spectral irradiance of the un-known source can be determined, also on a relativescale. Then, provided that the wavelength range ofcalibration extends to the visible or near-UV regionin which standard sources of irradiance are avail-able, the absolute spectral irradiance of the un-known source can be determined at onewavelength within the calibrated wavelengthrange. This absolute value can then be used to nor-malize the relative scale of irradiance to an abso-lute scale.3.2.2 Argon Arc Calibrations Two types ofsources are calibrated as irradiance standards, theargon mini-arc and the deuterium lamp. Calibrationof an argon arc will be discussed first.

The light source used here as a standard of spec-tral radiance in the VUV and near UV is an argonmini-arc previously calibrated by the hydrogenarc. This source is used according to the methoddescribed above to calibrate a second mini-arc asan irradiance standard. The measurements in theVUV spectral range are performed using a SEYAmonochromator. The above measurements deter-mine the irradiance of the unknown source only on

a relative scale. To obtain an absolute scale, theirradiance of the unknown source is compared inthe near UV region to that of a calibrated tungsten-quartz-halogen irradiance standard [17]. This mea-surement is carried out on a double mono-chromator, with an integrating sphere used as itsentrance aperture. For this measurement this setuphas two advantages. First, the double monochro-mator greatly reduces scattered light which other-wise would be a serious problem for thequartz-halogen source. Second, the integratingsphere serves as a better method of rendering thesignal insensitive to the direction of radiation fromthe source. This is important in comparing thesmall sized arc source with the much larger quartz-halogen source area.

The determinations of spectral irradiance arebased on the assumptions that the system efficien-cies of the measuring spectroradiometer are inde-pendent of the angle at which the radiation entersand that diffraction effects are not significant.Since the monochromator grating used must be as-sumed to have a nonuniform reflection efficiency, adiffuser located directly behind the field aperture isused to ensure that the first assumption is met. Amagnesium fluoride window, ground on one side,was used as the diffusing element. Although itcannot be expected to be as good a diffuser asan integrating sphere, it was shown to be suitable,at least over a relatively small range of angles.For the argon mini-arc the uncertainty in thespectral irradiance that is due to the nonidealproperties of the diffusing window can be as highas 6%.

The possible effects of diffraction must also beconsidered. For an extended homogeneous radia-tion source whose dimensions are large comparedwith the source aperture, it can be shown thatFrauenhofer diffraction effects introduce no wave-length dependence and that the geometric factorsare wavelength independent. This can be under-stood qualitatively by realizing that some percent-age of the radiation from each radiating point doesnot pass through the field aperture because of dif-fraction. However, a complementary point can al-ways be found in 'the domain of the extendedhomogeneous source that exactly makes up forsuch a loss. By substituting a different-sized aper-ture we have shown that the inhomogeneity of themini-arc is insufficient to cause any measurable ef-fects due to diffraction.

The vacuum spectroradiometer consists of a so-lar blind photonmultiplier and a spectrometer with a2-mm diameter field aperture. This aperture is

36



mounted on a MgF2 diffusing window located50 mm in front of the entrance slit. The diffuserwas located some distance in front of the entranceslit so as not to overfill the grating and therebyincrease the scattered light, which was less than2% of the weakest signal. The mini-arc was located50 cm from the field stop, and a 0.3 mm-diameteraperture placed 5 cm from the mini-arc center wasused to restrict the size of the radiating area toabout 0.5mm in diameter. The 2 o uncertainty inthe absolute spectral radiance of the mini-arc hasbeen given above to be 5.3% between 140 and330 nm.

The total uncertainty in the irradiance calibra-tion of an argon mini-arc, including uncertainties inthe primary source calibration and the transfer pro-cedure, is estimated to be < 10%. Although the arcemits radiation at shortertwavelengths than thatshown in figure 13, the short wavelength limit istaken to be 140 nm because of the low signal ob-tained from the stopped-down mini-arc radiancestandard at shorter wavelengths. The bandpass forthe radiance-to-irradiance transfer here was set atI Pm. Not shown in figure 13 are several lines in theargon arc spectrum that are due to residual gas im-purities [14]. The irradiance of the mini-arc was puton an absolute scale by determining its absolute spec-tral irradiance at 280 nm, using a tungsten-quartz-halogen lamp as an irradiance standard. A separatespectroradiometer utilizing a BaSO 4 -coated inte-grating sphere, a predispersing monochromator,and an analyzing spectrometer was used for thismeasurement. One measurement of this type is suf-ficient to determine the absolute irradiance at allwavelengths 116].3.2.3 Deuterium Lamp Calibration Deuteriumlamps are calibrated as spectral irradiance stan-dards in the near UV range 200<CX<350 nm by an-other calibration group in the Radiometric PhysicsDivision of NBS. The lamps calibrated are of theside-on type. Prior to calibration they are potted ina base identical to that of the tungsten-quartz-halogen lamps. Our calibrations of such lamps usu-ally consist of extending the spectral range in thevacuum UV down to 165 rn. The measurementsare performed on the SEYA monochromatorsetup, using the same method as was described inthe previous section for calibrating an argon arc asan irradiance source.

Briefly, a set of commercially available deu-terium lamps was calibrated for spectral irradiancein the 167-350 nm spectral range. At 250 nm andabove the spectral irradiance values were obtainedusing a tungsten-quartz-halogen lamp whose cali-

bration is based on an NBS blackbody. Below250 nm the relative spectral distribution of the deu-terium lamps was determined through the use of anargon mini-arc spectral radiance transfer standardwhose calibration is based upon the NBS wall-sta-bilized hydrogen arc. The absolute values assignedto the deuterium lamps were then obtained by nor-malizing to the spectral irradiance values at 250 nmand above. Confidence in this procedure was estab-lished by comparing the absolute spectral irradi-ance of the deuterium lamps as determinedindependently by the argon mini-arc radiance stan-dard and the tungsten-quartz-halogen irradiancestandard in the 250-330 nm range. The agreementwas within 3%. The large UV flux and high stabil-ity of the mini-arc are the properties which makepossible the calibration below 250 nm,

100

10

a1E

'atA

il.o

10.2

103

A (im)

Figure 13. Absolute spectral irradiance m easured as a functionof wavelength at a distance of 50 cm from the field aperture forfive different continuum sources with the indicated power re-quirements. The spectrum of the D, lamp with MjgF2 windowbelow 170 nm, measured for a I nm bandpass, is a pseudo-continuum made up of blended lines. Shown for comparisonpurposes are spectra of the 250 MeV National Bureau of Stan-dards synchrotron radiation facility, for a beam current of10 mA and a field aperture distance of 19 m, and the tungsten-quartz-halogen lamp, for the standard 50-cm distance.

37



Since the deuterium lamps are aligned and pot-ted in bipost bases identical to those used in mount-ing tungsten-quartz-halogen lamps, the twostandards are interchangeable in a given opticalsystem. Because the shapes of the spectral distribu-tions of these lamps are so different, the presence ofsystematic errors in a measurement system may bedetected by intercomparing the two sources. Thespectral irradiance of the 30 W deuterium lamp isequal to that of the 1000 W quartz-halogen lamp atabout 260 nm, 100 times stronger at 200 nm, and100 times weaker at 350 nm.

The uncertainties in the absolute values of thedeuterium lamp spectral irradiance are: 200 nmand above, 6%; 171 to 195 nm, 7%; and 170 nm andbelow, 10%. A major contribution to each uncer-tainty is the variability associated with the strikingof the deuterium lamps. As a result all calibrateddeuterium lamps supplied to customers by NBS arepreselected for variabilities of 4% or less. Forhigher accuracy and confidence, one may take ad-vantage of the result that the relative spectral dis-tribution of the deuterium lamps is reproducible towithin about 1%. Thus, the uncertainty may be re-duced by renormalizing the absolute scale with aquartz-halogen lamp after each ignition to elimi-nate any variability in the strike. This proceduremakes possible a reduction in the uncertainties ofabout 3%.

Additional work has shown that under certainconditions the deuterium lamp with an MgF 2 win-dow may be used as a radiometric standard downto 115 nm [18]. This effort was directed at deter-mining the circumstances under which portions ofthe many-line molecular spectrum below 167 nmcan be used as a continuum.

Acknowledgments

This paper is a condensation of NBS MeasurementServices: Radiometric Standards in the Vacuum Ul-traviolet, NBS Special Publication 250-3. The NBSpublication is an extensive treatment by the presentauthors of the radiometric calibration program car-ried out by the vacuum ultraviolet radiometrygroup in the Atomic and Plasma Radiation Divi-sion of the National Bureau of Standards. NBSSpecial Publication 250-3 is available from the Su-perintendent of Documents, U.S. GovernmentPrinting Office, Washington, DC 20402-9325.

The authors are grateful for the suggestionsmade by W. L. Wiese and J. R. Roberts. Also, theauthors wish to acknowledge the continuing

support of the VUV radiometry program at NBSby the NASA Solar and Heliospheric Branch.Most of the work described here was carried outunder partial funding by this agency.

References

[I] Ott, W. R., Behringer, K., and Gieres, G., Vacuum ultravi-olet radiometry with hydrogen arcs. 2: The high powerarc as an absolute standard of spectral radiance from 124nm to 360 nm, Appl. Opt, 14, 2121-2128 (1975).

[2] Roberts, J. R., and Voigt, P. A., The calculated continuousemission of a LTE hydrogen plasma, J. Res. Natil. Bur.Stand. (U.S.) 75A, 291-333 (1971).

[3] Ott, w. R., Fieffe-Prevost, P., and Wiese, W. L., VUVradiometry with hydrogen arcs. l: Principle of the methodand comparisons with blackbody calibrations from 1650 Ato 3600 A, Appl. Opt. 12, 1618-1629 (1973).

[4] Ott, w. R., and Wiese, W. L., Far ultraviolet spectral radi-ance calibrations at NBS, Opt. Eng. 12, 86-94 (1973).

(5] Wiese, W. L., Electric arcs, in Methods of ExperimentalPhysics, Bederson, B., and Fite, W. L., Eds. (Academic,New York, 1968), Vol. 7B, 307-353.

[6] Larenz, R. W., fiber ein Verfahren zur Messung sehrhoher Temperaturen in nahezu durchliissigen Bogensiulen,Z. Phys. 129, 327-342 (1951).

[7] Behringer, K., Prdzisionmessungen am Spektrum desWasserstoffplasmas, Z. Phys. 246, 333-347 (1971).

[8] Kepple, P. C., Improved stark profile calculations for thefirst four members of the hydrogen Lyman and Balmerseries, Dept. of Physics and Astronomy, Univ. of Mary-land, Report No. 831 (1968).

[9] Maecker, H., and Steinberger, S., Weiterentwicklung derKaskaden-bogenkammer fur hohe Leistungen, Z. Angew.Phys. 23, 456-458 (1967).

[10] Walker, J. H., Saunders, R. D., and Hattenburg, A. T.,NBS Measurement Services: Spectral Radiance Calibrations,Natil. Bur. Stand. (U.S.) Spec. Publ. 250-1 (January 1987).

[11] Boldt, G., Das thermische Plasma als Intensititsnormal-strahler im Wellenldngenbereich von 1100 his 3100 A,Space Sci. Rev. 11, 728-772 (1970).

(12] Morris, J. C., and Garrison, R. L., A radiation standard forthe vacuum ultraviolet, J. Quant. Spectrose. Radiat. Trans-fer 9, 1407-1418 (1969).

[13] Stuck, D., and Wende, B., Photometric comparison be-tween two calculable vacuum ultraviolet standard radia-tion sources: synchrotron radiation and plasma-blackbodyradiation, J. Opt. Soc. Amer. 62, 96-100 (1972).

[14] Bridges, J. M., and Ott, W. R., Vacuum ultraviolet ra-diometry. 3: The argon mini-arc as a new secondary stan-dard of spectral radiance, Appl. Opt. 16, 367-376 (1977).

[15] Saunders, R. D., Ott, W. R., and Bridges, J. M., Spectralirradiance standard for the ultraviolet: the deuteriumlamp, Appl. Opt. 17, 593-600 (1978).

(16] Ott, W. R., Bridges, J. M., and Klose, J. Z., Vacuum-ultra-violet spectral-irradiance calibrations: method and applica-tions, Opt. Lett. 5, 225-227 (1980).

[17] Walker, J. H., Saunders, R. D., Jackson, J. K., andMcSparron, D. A., NBS Measurement Services: SpectralIrradiance Calibrations, Natl. Bur. Stand. (U.S.) Spec.Publ. 250-20 (1987).

38



118] Klose, J. Z., Bridges, J. M, and Ott. W. R., The use of 4. Special Tests of Radiometric Devices in the Neardeuteriuth lamps as radiometric standards between 115 am and Vacuum Ultraviolet, SP 250 No. 40040S (7.6A)and 350 am. Extended Abstracts of the VI InternationalConference on Vacuum Ultraviolet Radiation Physics. Tests of customer supplied radiometric devicesCharlotttsville, VA, 111-52 (June 1980). not included above are performed at the direction

Appendix. NBS Calibration Services in the of the customer.Near and Vacuum Ultraviolet1. Spectral Irradiance Standard, Argon Mini-Arc(140-330 nur), SP 250 No. 40010S (7.6B)

An argon mini-arc supplied by the customer iscalibrated for spectral irradiance at 10 Dm intervalsin the wavelength region 140-330 nm. Absolutevalues are obtained by comparing the radiative out-put with laboratory standards of both spectral irra-diance and spectral radiance. The spectralirradiance measurement is made at a distance of50 cm fronm the field stop. Uncertainties (2o-) areestimated to be less than ±10% in the wavelengthregion 140-200 nm and within ±6% in the wave-length region 200-330 nm. A measurement of thespectral transmission of the arc window is includedin order that the calibration be independent of pos-sible window deterioration or damage.

2. Spectral Radiance Standard, Argon Mini-Arc(115-330 Dn), SP 250 No. 40020S (7.60

The spectral radiance of argon mini-arc radiationsources is determined to within a 2o- uncertainty ofless than 6% over the wavelength range 140-330nm and 11% over the wavelength range 115-140 mD. The calibrated area of the 4 mm diameterradiation source is the central 0.3 mm diameter re-gion. Typical values of the spectral radianceare: at 250 nm, L=30mW cm 2 nm-' sr-'; and at150 anm, 4=3 mW cm<2 nm-' sr-'. The transmis-sion of the demountable MgF2 lamp window andthat of an additional MgF2 window are determinedindividually so that the user may check periodi-cally for possible long term variations.

3. Spectral Irradiance Standard, Deuterium ArcLamp (165-200 mrn), SP 250 No. 40030S (7.61))

The lamp is calibrated at IO wavelengths from165 to 200 nm at a distance of 50 cm. Its spectralirradiances are about 0.05 ,W cm-2 nm-' at165 m, 0.03 W cm-2 nm' at 170 unn, and0.05,aW cm-2 nm-' at 200nm. The approximate2oa uncertainty relative to SI units is estimated tobe less than 10%. The lamp is normally supplied byNBS and requires 300 mA at about 100 V.

39