HAL Id: hal-02471181
https://hal.archives-ouvertes.fr/hal-02471181
Submitted on 13 May 2020
HAL
is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire
HAL, estdestinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
International round-robin experiment for angle-resolved light scattering measurement
A. von Finck, T. Herffurth, A. Duparré, S. Schröder, M. Lequime, Myriam Zerrad, S. Liukaityte, C. Amra, S. Achour, M. Chalony, et al.
To cite this version:
A. von Finck, T. Herffurth, A. Duparré, S. Schröder, M. Lequime, et al.. International round-robin ex-
periment for angle-resolved light scattering measurement. Applied optics, Optical Society of America,
2019, 58 (24), pp.6638. �10.1364/AO.58.006638�. �hal-02471181�
International Round-Robin Experiment for Angle Resolved Light Scattering Measurement
A. VON F INCK ,
1,*T. H ERFFURTH ,
1A. D UPARRÉ ,
1S. S CHRÖDER ,
1M. L EQUIME ,
2M.
Z ERRAD ,
2S. L IUKAITYTE ,
2C. A MRA ,
2S. A CHOUR ,
3M. C HALONY ,
3Q.
K UPERMAN ,
3Y. C ORNIL ,
3A. B IALEK ,
4T. G OODMAN ,
4C. G REENWELL ,
4B. G UR ,
5S. B RINKERS ,
5G. O TTER ,
5A. V OSTEEN ,
5J. S TOVER ,
6R. V INK ,
7A. D EEP ,
7AND D.
D OYLE
71Fraunhofer Institute for Applied Optics and Precision Engineering IOF, A.-Einstein-Str. 7, 07745 Jena, Germany
2Institute Fresnel, Domaine Universitaire de Saint Jérôme, 13397 Marseille cedex 20, France
3Light Tec SARL, Pôle d’Activités Hyérois, 1128 Route de Toulon, 83400 Hyères, France
4National Physical Laboratory, Hampton Rd, Teddington, Middlesex, TW11 0LW, United Kingdom
5TNO Industrie en techniek, Stieltjesweg 1, 2628 CK Delft, The Netherlands
6The Scatter Work, Inc., 2100 N. Wilmot, Suite 202, Tucson AZ 85712, USA
7European Space Agency ESTEC, Optics Section (TEC-MMO), Keplerlaan 1, PO Box 299, 2200 AG Noordwijk, The Netherlands
*Corresponding author: [email protected]
Received XX Month XXXX; revised XX Month, XXXX; accepted XX Month XXXX; posted XX Month XXXX (Doc. ID XXXXX); published XX Month XXXX An international round-robin experiment has been conducted to test procedures and methods for the measurement of angle resolved light scattering. ASTM E2387-05 has been used as the main guide, while the experience gained should also contribute to the new ISO standard of angle resolved scattering currently under development (ISO/WD 19986:2016). Seven laboratories from Europe and the USA measured the angle resolved scattering from Al/SiO2 coated substrates, transparent substrates, volume diffusors, quasi volume diffusors, white calibration standards, and grating samples at laser wavelengths in the UV, VIS and NIR spectrum. Results were sent to Fraunhofer IOF that coordinated the experiments and analyzed the data, while ESA-ESTEC, as the project donor, defined conditions and parameters. Depending mainly on the sample type, overall good to reasonable agreements were observed, with largest deviations at scattering angles very close to the specular beam. Volume diffusor characterization unexpectedly turned out to be challenging. Not all participants provided measurement uncertainty ranges according to GUM, often, a single general scatterometer-related measurement uncertainty value was stated. Although relative instrument measurement uncertainties close to 1% are sometimes claimed, the comparison results did not support these claims for specular scattering samples as mirrors, substrates, or gratings. © 2017 Optical Society of America
OCIS codes: (290.0290) Scattering; (290.5820) Scattering measurements; (120.4800) Optical standards and testing http://dx.doi.org/10.1364/AO.99.099999
1. Introduction
The increasingly demanding requirements of optical systems in the recent few years, especially those of space optical systems, have driven light scattering specifications of optical components to the extreme.
Light scattering reduces the optical throughput of optical systems, the lateral resolution of imaging optics, the spectral resolution of spectrometer systems, and it increases stray light problems. As a result, light scattering specifications have become one of the most critical performance parameters of high-end optical instruments and components. Testing these important specifications consequently drives demands for reliable measurements.
Light scattering distributions are described by the angle resolved quantities BSDF (Bidirectional Scattering Distribution Function) or ARS (Angle Resolved Scattering), which are defined in the ASTM
standard E2387-05 [1]. This ASTM standard is currently the only active standard available, however, there is a corresponding new ISO standard under development (ISO/WD 19986:2016) [2]. Practically, a quantification of BSDF / ARS is typically performed by in-house developed scatterometers that consequently differ in terms of measurement parameters, optical set-up, and also procedures. This can result in large deviations that are sometimes observed between measurements performed by different laboratories on the same sample. Consequently, a test of the reliability of BSDF measurements and procedures is important for ESA and space optics manufacturers.
Previously performed round-robin experiments of angle resolved light scattering date back to the late 1990s, and were conducted to test the former ASTM E1392 standard [3, 4], which has been by now replaced by ASTM E2387. In addition, round robin experiments have been conducted to verify the standards of the integrated scattering quantities TS (Total scattering, ISO 13696 [4]) and TIS (Total
int th pr pr an ex th Se anse
2.
Th irr th po
Fig coth an ad inc
Fig be θs
ca los co tra [5by re int
sca catec th m co
tegrated scatterin e active ASTM E2 reviously been re
This new Roun roject, where ESA nd Fraunhofer I xperiments. This p
e basic light sc ection 3 contains nd conducted, foll ection 4.
. Definitions
he BSDF is defi radiance dEe; whi e solid angle ΔΩs
ower Pi, and to the BS where φs and θ g. 1), respectively osθs factor, where e incident radian ngle in situations dequate accuracy cidence and isotr
g. 1. Light scatter eam, 3 - reflected - polar scattering Numerical inte alculation of the h
sses. For norma orresponds to
ansmission (TSf) ]. Integration ov y directional–he flectance (THR) tegrated scatterin
which is denoted attering, respect areful instrumen chniques are sug at signal voltage easuring the ab omprised in equat
ng, ASTM F1048 2387 standard fo ported.
nd-Robin experim A, as the project d
IOF, as the pro paper is organize cattering definit
details about ho lowed by a comp
fined as the sur ich equals to the p
s, normalized to t e projected surfa
(
s, s)
SDFθ ϕ = dE dL θs are the azimuth y. BSDF can be c e ARS correspond
nt flux. It is ofte where out of pla y, which is typica ropic scattering sa
ring geometry and beam, 4 - transm g angle, s - azimu egration of an hemispherical sca al incidence, inte
Total Scatterin hemispheres, re er the complete emispherical ref . For different ng (S) can be used
∆Ps
Pi ∬
d by Sf for forwar tively. For the p nt calibration is ggested in ASTM-E
values from the bsolute power v
tion (1) and (2). T
[5]) [7, 8]. A com or angle resolved ment was part donor, defined the
oject contractor ed as follows. Sec tions and calibr ow the experimen prehensive analy
rface radiance d power ΔPs of the that solid angle, t ace area [1,9]
(
i s
lim0
ΔΩ
=ΔΩ→ Δ P
P E
L
e s e
hal and polar sca onverted into AR ds to the radiant en convenient to ane scattering can ally the case for m
amples [10].
d definitions: 1 - s mitted beam, θi -
uth scattering ang ngle resolved attering quantitie
egration within ng in the refle
espectively, as de hemisphere is s flectance or to integration limi d
rd scattering and practical measur mandatory. Di E2387, but they a
detector are eva values Ps and P The BSDF can the
mprehensive test d scattering has n of an ESA-EXPR e frame condition r, coordinated t ction 2 summariz ration procedur nts were organiz ysis of the results
dLe scaled by t light scattered in to the incident lig
( )
s s
s s
cos ,
θ ϕ θ
Ω , (
attering angles (s RS by an addition intensity scaled omit the azimu n be predicted wi moderate angles
sample, 2 - incide angle of inciden gle.
quantities allow s or light scatteri
θs=2° and θs=8 ection (TSb) an efined in ISO136
ometimes denot otal hemispheric
its, the paramet
, (
by Sb for backwa rement of BSDF, ifferent calibrati all have in comm aluated, rather th Pi directly, as e en be calculated b
t of not RO ns, the zes res.
zed in
the nto ght
1) see nal uth by ith of
ent ce,
ws ing 85°
nd 96 ted cal ter
2) ard , a on on han e.g.
by
wher Vs is r corresp order to compon the det dynami achieve factor k differen other un
• Ab
• froRe a r sca VIS(Po
• Dif ref Vs
an sam
3. Rou
A. Layo The lar layout o sample contam campaig sequent potentia In o experim parame experim illumina incident in the fo
••
••
•
• A do the exp docume technica main gu test the the part good co
(
BSDFθ re kc is a calibrati recorded from ponds to the det
o measure both nents as well as tection system n ic range of typica ed by adaption te kA. This factor nee nt calibration rou nknown variable bsolute calibratio om set-up geome elative calibration
reference sample attering angles al S-NIR meas olytetrafluoroeth ffuse calibration:
ference sample w is recorded over nd numerically in mples are also us
und-Robin Ex
out and organiza rge number of p
of the experime contamination d mination, as observ gns [7], had to b tially by the pa ally insensitive to order to find c ments, the part eters that are fea ments. The mo ation wavelength t angle without o ollowing basic me
Illumination wav Illumination pol Illumination det Incident angle:
measured in Litt Range of scatter diffusor sample samples -90° to Scattering angle samples 0.5°
ocument containi periments was p ent also contain al and project ma uide to be follow eir own methods rticipants were a ondition. Also, it w
) (
s i
s s
s,ϕ θ
θ V k
V
≈ ΔΩ ion factor that co
the detector d tector signal in
extremely low s extremely high p needs to be wo ally over 10 order echniques that a eds to be determ utines can be im es:
on: measuremen etries.
n: kc is determine e with known B llows reducing ca surements, t hylene) samples a : kc is determined with known integ r the correspond ntegrated (see al sed for this calibr
xperiments
ation of the exp participating lab ents to shorten
during the exper rved during earlie be avoided. One u articipants, a qua o contamination.
common measu ticipants were asible with their ost critical mea hs, illumination p obscuration of the easurement para velengths (nm): 3 arization: s-pol tection: unpolariz
11°, except for t trow configuratio ring angles: -90°
e (#W and #D s +90°
e step sizes: 0.1 ing the detailed prepared and s ned sample hand
anagement infor wed, although the s and procedures asked to check if was advised to us
) (
c s s s s
cos
, θ
θ ϕ θ
A
A kk
V
k =
orresponds to Vi∆ during measurem
the incident bea scattering signals powers from spe orking linearly in rs of magnitude. T are described by mined before calib mplemented to d nt of Vi while ∆Ωs
ed by measurem BSDF. Recording alibration uncert typically wh are used for calib d by measuremen grated scattering, ding range of sca lso [1]). Typically ration.
eriments boratories requir
the timescale a riments. In parti er sequential RR m
unique sample w asi-volume-diffus urement parame asked to prov set-ups and ava asurement param
polarization, and e reflected beam.
ameters:
325, 633, 1064, 1 zed
the grating samp on
to +270° for tra amples in sectio 1° for grating sa measurement pa sent to the part
dling informatio rmation. ASTM E participants wer s. After receipt of f the samples had
se mechanical blo
)
s s s
cos ,
θ ϕ θ
A
, (3)
∆Ωs. The signal ment, and Vi
am. Hence, in s from optical ecular beams, nside a large This is usually an additional bration. Then, determine the Ωs is calculated ment of Vs from Vs at multiple tainty. For UV- hite PTFE bration.
nt of Vs from a e.g. TS, where attering angles y, white PTFE
red a parallel and to reduce icular, sample measurement was measured ser, that was eters for the vide possible ailable for the meters were d the smallest . This resulted 1550
ples that were ansparent and on C), opaque amples, other arameters for ticipants. The on as well as 2387 was the re also free to f the samples, d arrived in a owers to blow
off we a r lat(u m sca lar Th prra
pa co nabe On re Ap or duco co Jul No ES
Fig
B.
T De
ff single particles ere to be discusse representative m terally scan the usually called ras
easurement. Thi attering propert rger than the actu he participants w repared tabular anges (for k=2) as
The experiment articipant receive orrelated with the ame. The sample efore that, all sam
ne laboratory joi sults and sample pril 2016. After rder to character uring the experim orrelated with the ontaining results
ly 2016. The int ovember 2016, a SA-ESTEC.
g. 2. Schematic of
Participating la Table 1: Laborat
esignation
ESA E
IF I
IOF LT NPL
TNO TN
TSW
s before measur ed with IOF. The measurement pos surface with fix ster scan or ma s allows the iden ties across the sa ual measurement were asked to
sheets with BSD s a function of the
ts were conduct ed a random nu e experimental re s were distribute mples were scree
ined the experim es were returned that, the screeni rize sample com ments (see Fig.
e BSDF data from and analysis of t ternal report for nd the results we
f the layout of the
aboratories and tories that partic
(alphabet O European Space A
Institute Fresnel, Fraunhofer Inst Precision En Light Tec National Physic
Middles NO Industrie en t
The Scatter Wor
rement. Other cl participants wer sition, therefore i xed incident an apping) prior to ntification of loca ample surface th
t uncertainty of th fill in their res DF and measure e scattering angle.
ted anonymousl umber between I esults instead of w
ed to the particip ened at the coord ments in Decemb
d to the coordina ing experiments parability and sa 2). These result m the participants the experiments r the participant ere finally presen
e experiments.
instrumentatio cipated in the RR
tic order) Organization Agency ESTEC, No
Netherlands Espace Photoniq
France titute for Applied ngineering, Jena, G c SARL, Hyères, Fr cal Laboratory, T sex, United Kingd techniek, Delft, Th rks, Inc., Tucson, A
leaning techniqu re requested to fin it was suggested
d scattering ang o the actual BRD al variations of t hat are often mu he instrument [1 sults in previous
ement uncertain .
ly. Therefore, ea I and VII, which with the participa pants in June 201 dinating laborato
ber 2015. The la ating laboratory
were repeated ample degradati ts were eventua . An internal repo was sent to ESA ts was released nted in June 2017
n
R measurement
oordwijk, The que, Marseille, d Optics and Germany
rance Teddington, dom he Netherlands
Arizona, USA ues
nd to gle DF the uch 1].
sly nty ach is ant 15, ry.
ast in in on ally ort in in 7 at
ts
Seven measur informa Europea one from
Figur by the compon function disadva filter (SF the scat apertur (FS). Us measur (AS). Ad that m conveni detecto
Fig. 3.
resolved optical The corresp design corresp a modif illumina One pa includin the det (symme detectio Tab(ra
P
n internationa rements (see Tab ation are listed
an research instit m the USA provid re 3 displays the
participants. T nents to measu nal principles antages. The beam
F) and illuminate ttered light from res are used to d sually, either a ph rements in the VI dditional detecto may affect the
iently used to m r field of view.
(Color online) d light scattering
set-ups used by designs describe ponds to ASTM
(2) with exchang ponds to ASTM se fied illumination ation spot diamet articipant extend ng the source opt tection side, this etric to the illum on optics being re ble 2: Set-up and andom order). P Participant
number I II III IV V VIIVI
al laboratories ble 1). Complete
in Appendix A tutions, but also o ded results.
schematics of the They all incorpo ure scattered lig
and therefore m from a light so es the sample (S).
m the sample wi efine the solid an hotomultiplier tub IS and NIR. The p or optics contrib near angular li modify the solid an
Schematics of t g instrumentation y the participan ed in the ASTM2 design (1), set-u ged (AS) and (F et-up (3). One pa n, such that (PH) ter is adjusted by ded set-up (C) tics were replaced
participant also mination) with th eplaced by a fiber d calibration ove Participant VII u
Set-up A A C C A C / C B
participated e names, address A. Most of the f one company fro e different optica orate the same
ght, but still h different adv urce (L) is cleane . The detection sy th a detector (D ngle (AS) and the be or a photo dio pinhole is usually bute additional s imit, neverthele ngle or to precis
the optical desig n used by the part nts mostly corre
2387 standard [1 up (B) correspo
S) apertures, wh articipant used se ) is imaged onto y adapting the (PH with fiber opti d by a fiber coupl o used a collima he aperture (AS) r coupled sensor.
erview of the pa used 2 different Calib
me abs abs abs rel abs diffuse rel
in the RR ses, and other facilities were om France and al set-ups used fundamental have different vantages and ed by a spatial ystem records D), where two e field of view ode is used for y imaged onto scattered light ess, they are sely define the
gns for angle ticipants.
espond to the 1]: Set-up (A) nds to ASTM hile set-up (C) et-up (A) with o (S) and the H) size in (SF).
ics: the (PH) led source. On ated detection including the
articipants t set-ups.
bration ethod solute solute solute lative solute lative
/ relative
de On th
Sa
#
Q sa pe to us 3 mch pr pe (sp an to illu als an min
C.
In forsa typ su on wa susa wa m Ho blo tecaz
co sp an pieac un pr poco re Ta mTh
One participant etector is remove ne participant use e detection field
ample Subst
#W #M
#G
#S
#D Fused S QVD 2-plate q For the chara amples, it is of erformance at sca reduce stray lig se under-filled illu
mm of diamete haracterization o recision is mor erformance. Ther peckle) is usuall nd/or illuminatio laser sources.
Two participan umination spot d so give more fre nd were used by
ethods (see also Table 2.
Sample set order to cover a r the measurem ample set was to
pe should be as uppliers were ask ne production run as that the sam uppliers. This po amples, as the sta
as not alway easurements. He owever, all sam owers and, if ne chniques. All sam zimuthal sample o The final samp omprehensive set pecular or diffuse nd (iii) transpare ece for each pa cquired; except in nique piece was u For typical op roportional to the ower spectral de onnection is still g levant rms surf able 3 and were icroscopy (AFM herefore, master
used set-up C w ed for scattering ed set-up (A) wit of view is smalle
trate / sample typ N-BK7 Fused Silica
Borofloat White PTFE Silica volume diffu quasi volume diff acterization of f most importan attering angles cl ght problems. Th umination, small er), and are b of diffusely-scat re important th refore, a reductio ly achieved by u on with an exten
nts used high diameters from 5
edom in the cho y three participan Section 2) used b
a broad scenario o ment of scattered be created. More s identical as po ked to provide q n. One main cond mples had to be osed a challenge ndard quality wi s sufficient fo ence, some sam mples were clean
ecessary, using d mples were labele
orientation were ple set composit t comprises (i) lo e scattering samp ent and opaque s rticipants plus o n the case of the used for the exper ptical compone e interface rough ensity function (P given but more co face roughness
derived by topo M) and white lig r PSD functions
where the lens be angles close to t th the over-illumi er than the illumi
pe
A reflectiv fusor
fusor
low-scattering nce to maximiz ose to the specul herefore, the corr l illumination spo ased on laser ttering samples han sensitivity on of sample relat
using large illum nded spectral ban precision set-u 5 mm to 8 mm. B oice of the illumin nts. The set-ups a by the participan
of different exper d light, a possib eover, all sample ossible. For this quotes of sample dition posed by t e available from e for some of t th respect to sam or comparable mples provided h ned before ship drag wiping or u ed and the front marked with an tion is displayed ow and high scat
ples as well as d samples. Of each one additional sp e QVD sample, w
riments.
ents, light scatt ness properties r PSD), for interfe omplex [13-16]. T
values at λ=633 graphy analysis ght interferomet s were calculate
etween sample an the specular bea ination concept ( ination spot on t Table 3: Samp Front surface Al / SiO2[85 nm]
- ve holographic gr
- - specular-reflecti- ze sensitivity, t lar beam, as well responding set-u ots (typically abo sources. For t s, however, hi or near specul ted statistical noi mination spot siz ndwidth compar ups with effecti Broad-band sourc nation waveleng and the calibrati nts are summariz
rimental challeng bly comprehensi es from one samp
purpose, differe es manufactured the project initiat m standard optic
the low scatteri mple contaminati light scatteri had to be rejecte pment using du ultra-sonic cleani side as well as t arrow.
d in Table 3. T ttering samples, ( diffraction gratin h sample type, o pare sample we where only a sing ttering is usua
rms roughness an rence coatings t The light scatteri 3 nm are listed using atomic for try (WLI) [17-1 ed and integrat
nd am.
i.e.
the
sample) highly d uniform
ples for the expe Dimens ø50ø50
rating 25
ø ng ø6
the as ups out the igh lar ise zes red ive ces gth, on zed
ges ple ive ent in tor cal ing on ng ed.
ust- ing the The (ii) gs, ne ere gle
ally nd the ng in rce ted 9].
between range o covered than the This sid orientat in the ou
Fig. 4. M
Fig. 5. T experim screenin allowed degrada
). Over-illuminat diffuse scattering m illumination.
eriments.
sions (mm) r 0.8x9.5
0.8x6.35 0
5x25x6 ø40
ø50x5 5
60x20
n 0.055 µm-1 and of surface rms rou d. One side of the e other side and de was defined ted such that the ut-of-plane direc
Master PSD functi
The samples fro ments in one run
ng experiments w d characterizing
ation during the c
tion is primarily g samples [12] a
rms (nm) 0.61 / 0.52 4.3
40
- S
500 / 310 d 1.6 µm1200 -1 (see Fig ughness values b
e diffusor sampl to exhibit a dire as the front sid e direction of high ction.
ions of the individ
om each type w n using the light s were repeated a
the sample com course of the exp
y used to charac and typically exce
Order Num PAV-PM-2037- PW1-2025-UV GH25-12V (Tho SG 3043-U (Spher HOD-300 (Her g. 4). It can be see- etween 0.5 nm a es was found to ctional roughnes de. The sample hest roughness w
dual sample types
were screened b scattering sensor after the RR expe mparability as w periments.
cterize rather els in a highly
mber -C (CVI) V (CVI)
orlabs) re Optics) raeus) en that a large and 1.2 µm are be a rougher ss component.
azimuth was was orientated
s.
before the RR r “horos”. The eriments. This well as sample
D.
Th on re sp 21 pr easp he th aft be sh sp pr pa
Fig (ri desa
qu Tosa wa th av
co as na sp ro
#
#
#
#
#
. Sample screen he samples from ne run using the cords the near an pecular reflection
1]. It was moun rovide spatially r ach sample type pecular beam).
emisphere except e transmission d ter the RR experi As an example, efore and after th howed the highes pare sample “s” th rescreening and article contamina
g. 6: Compariso ight) results of th eviation from th ample pool and m Although even t uantitative post-m o derive a single ample Sb, the data as applied. The s e relative devia verage of the integ
Based on the omparability and possible. The s atural sample we pare samples we oom during the RR
Table 4: Sampl screening exp calcu
I II
#M 18% 8%
#W 3% 2%
#G 6% 8%
#S 1% 1%
#D -6% 5%
ing
m each type, resp light scattering s ngular scattered n, at an incident a nted fixed over a resolved scans of e (integrated bet All samples w t for the diffusor direction. The scr ments to analyze Fig. 6 shows the he experiments. T st deviation from hat was not used postscreening s ations between th
on of the pre-sc he eight M# sam he average were marked as spare “s this qualitative ev measurement an
representative i a reduction algorit
sample comparab ation of its inte
grated scattering
,
̅
scans, spare s homogeneity of spare samples w ear without huma ere boxed and st R experiments.
le comparability periments. The s ulated without th
III IV 0% -5% -9 5% -21% 8 -15% 14% -1 1% -1% 0 -1% -1% 0
pectively, were sc sensor “horos” (F light within a con angle of θi=8°, and an XY-translation f the integrated s tween about 1°
were measured r samples, which reening experime e sample wear.
screening scan f The second samp
the average and for the RR exper scans qualitative he scans for partic
creening (left) a mples. The sample
e removed from s”.
valuation would nalysis of these s ntegrated scatter thm suggested in bility of sample egrated scatterin g values of all sam
, ,
, 1
samples were f the residual sam were also used t
an interaction. Fo tored in a cond
y N as determin standard deviat he spare sample
V VI VII 9% 0% -13%
% 1% 2%
14% 8% -7%
% 0% 0%
% 5% -2%
canned together Fig. 5). This sens ne of 8° around t d at λ=650 nm [2 n sample stage scattered light Sb
to 8° around t in the reflecti were measured ents were repeat for the #M sampl ple in the first ro was selected as t riments. Compari ely already revea
cipant I, II, and VI
and post-screeni es with the highe m the participan
be helpful, a mo scans is preferab ring value for ea n DIN-ISO13696 [ e N is calculated ng value from t mples ̅, :
( picked such th mples were as go
to characterize t or this purpose, t itioned ISO7 cle
ned by the pre- tion (std.) was es (s).
s std.
27% 10%
-19% 9%
28% 12%
-1% 1%
15% 4%
in sor the 20, to of the on in ted ow les the ng als II.
ng est nts’
ore ble.
ach [5]
by the
3) hat od the the ean
Samp change experim
For determi constan compen In T respecti measur Table post-sc
#M
#W#G
#S
#D In su (for scat Moreov notewo effects experim sample scatterin not show
4. Res
A total o (see An the resu systems instead A. Com To ef parame angle r where n influenc all par measur invarian samples log sca Integrat to +175 sample (integra respecti +5°). Th 7 to 12 f From deviatio relative
ple degradation of its integrated ments Sb,post,N:
the screening ine the calibratio nt offset between nsated by adaptin Table 4 and 5,
tively. Negative v rement uncertain e 5: Sample degr creening experi calculated
I II
4% 264%
0% 215%
3% 9%
-1% -1%
1% 0%
ummary, light sc attering angles of ver, it can be obs orthy degradation
of natural samp ments can probab handling during ng samples #M a w a significant de
sults
of 73 BSDF meas nnex B). Addition ults, as some pa s, with a differe of BSDF, or with mparison by integ
ffectively compar eter per angle res resolved scans w no measurement ces from the inst rticipants delive rements were shi nce postulated b s. Then, the data ale and integrat tion ranges from 5° in transmissi (integration lim ation limits bet tively), and the # he calculated inte for all provided m m these values an on of the integrat e deviations from
of sample N d scattering valu
, ,
, ,
measurements, on factor kc in E n the pre- and p ng kc of the post-s the results for values of sample nty.
radation N as d iments. The aver d without the sp
III IV V
1% 3% 0%
-2% 1% -2%
2% 1% -2%
-1% 1% 1%
1% 1% 1%
cattering related 1° to 8°) of 1% <
served that all sp n during the exp ple degradation bly be excluded. H
g the experimen and #W, while th egradation.
surements were p nal data processi articipants provi ent sampling of unremoved mea grated scatterin re this huge num solved scan is be were numerically t artifacts (especi trument signatur ered data. Bef ifted to normal in by Harvey [22], a were interpolat ted to yield Sb
m +5° to +50° in r on direction we mits between +18 tween +140° to
#G samples (inte egrated scatterin measurements of nd per illuminat ted scattering va m the arithmetic
N is quantified by ue before Sb,pre,N
1
calibration unc Equation (3) wou post-screening sc
screening experim and are e degradation ar
determined by t rage degradatio pare samples (s)
V VI VII
% 1% 14%
% -1% 98%
% 6% 2%
% 1% 1%
% 1% 0%
differences betw
< < 12% have b pare samples did periments. This
during the cour Highest degradat nts was observed
he high scattering
provided by the ing was necessar
ded data in oth the scattering a asurement artifac ng values mber of measurem
eneficial. For this y integrated in ially detector obs
re) were expecte fore integration ncidence accordi except for the ted on a 0.1° grid
by means of E reflection directi ere used, except 85° to +235°), the
o +175° and + egration limits b
g values are plot f the individual sa tion wavelength, alues was calcula c mean were ca
y the relative and after the
(4) certainties to uld result in a ans. This was ments.
summarized, re a result of
the pre- and on (avg.) was
.
s avg.
1% 41%
2% 44%
3% 3%
1% 0%
0% 1%
ween samples been achieved.
d not exhibit a indicates that rse of the RR tion caused by d for the low g samples did
7 participants ry to evaluate er coordinate angle, in ARS cts.
ments, a single s purpose, the regions of θs
scurations and ed and where n, the BSDF ing to the shift QVD and #S d on a vertical Equation (2).
on and +125°
t for the QVD e #D samples +5° to +35°, between 0° to tted in Figures ample types.
the standard ated, and their alculated. The
sta m tab cowa be ea inc 1.
Fig sca wo topwa pr at pr hig effca m de
Fig by
32 63 10 15
2.
Fig sca wesca pa de m da prpa Be de co
andard deviation easurements wa bles 6 to 11, th omparability an
avelength scalin etween the repor ach specific type
cluded in the calc Mirror samples gure 7 and Table attering decreas ould be expected pography. Two avelength-scaling robably caused b
1550 nm sam rescreening resu gher scattering s fect could possib annot be observe
easurements sho eviation of 9% for
g. 7: Comparison y red circles.
Table 6: Analy s I 18%
4%
25 nm 11%
33 nm 2%
064 nm 550 nm
Window sample gure 8 and Table attering decrease ere observed by aling characteris articipants VII a egradation effec easurement of pa ata also containe rescreening resul articipant IV indic esides the outlier eviation of 40%
onstraints and ex
n was also calcula s in part too low hese deviations
d sample degr ng or experienc
rted data and th of measuremen culations.
#M
e 6 show the res ses with increa d for smooth sam
outliers were g analysis (see by calibration issu mple degradation ults revealed tha sample than the bly be present i ed for 633 nm illu
ow a very good r all illumination w
n of the #M samp
ysis of the measu samples with ou II III IV
-5%
3%
2%
-
es #W
e 7 show the res es with increasin comparison and stic. The shape at 633 nm and cts (see Fig. 2 articipant V is pro d negative BSDF lt compared to t cates a possible s rs, the measurem
%, which is co xpectations of th
ated even if the n for statistically si are directly com radation. Outlie ce based on ob he expected shap nt artefact) were
sults of the #M s asing illuminatio mples that scatte
observed by co Appendix C). Th ues; however, for n effects are a at participant I sample average in the 325 nm m umination. Besid
overall agreemen wavelengths.
ple measurement
urement deviat utliers removed
V VI V
-9% 0% -1
0% 1% 1
-3% -9%
0% -4% O
-10% 10%
0% O
sults of the #W s ng illumination w d by observation of the BSDF m 1550 nm could 22 in Annex B obably affected b F values). A simil the 633 nm BSD sample comparab ments exhibit an onsidered reaso his comparison,
number of provid ignificant results.
mpared to samp ers (identified bserved deviatio pe of the BRDF f removed and n
samples. Integrat on wavelength,
er purely from t omparison and hese outliers we
r the measureme also possible. T received an 18 (see Table 6). Th measurements, b des the outliers, t nt with a standa
ts. Outliers mark
tions of the #M d.
VII std.
13% 12%
4% - 10%
OL 3%
14%
OL -
samples. Integrat wavelength. Outlie of the wavelengt easurements fro d indicate samp B). The 1550 n by noise effects (t
lar deviation of t DF measurement bility related effe n average standa onable within t as light scatteri
ded In ple ons by for not
ted as the ere by ent The 8%
his but the ard
ked
ted ers om th- ple nm the the ect. of ard the ing
measur affected
Fig. 8: C by red c by circle Tab
Sb
325 633 106 155 Sf
325 633 106 155
3. Grati Figure 9 integrat wavelen scatterin particip obscura this me the spe effects c IV, and t standar should b significa emphas measur
rement on trans d by straylight or
Comparison of th circles; forward s es / crosses, resp ble 7: Analysis of samp
I II 3%
0%
5 nm 42%
3 nm 25%
64 nm 50 nm
5 nm 18%
3 nm 27%
64 nm 50 nm
ing samples #G 9 and Table 8 sh ted scattering v ngth. No outlier ng angles betwee pant I (see Fig
ation, also, a sligh easurement, whic ecular reflected
could have affec the 1064 nm me rd deviation of 22
be noted that at antly higher than sis the extreme rements on gratin
sparent low scatt contamination is
he #W sample me scattering Sf / bac pectively.
f the measurem ples with outlier
III IV V -21% 8%
1% -2%
-7%
-27% 34%
38%
OL 26%
-37% 29%
36%
OL
how the results values decrease rs were observe
en -90 and -70° in g. 23) could be htly misaligned in ch specifically af and diffracted cted the 633 nm easurements of pa
2% was found for large scattering a n this (see section e difficulty asso ngs.
ttering optics are ssues.
easurements. Ou ckward scatterin
ment deviations o rs removed.
VI VII
% 1% 2%
% -1% 98%
% -36%
% -12% -20%
% -38%
L OL
% -43%
% -19% OL
% -36%
L OL
of the #G samp with increasing ed. The low BS n the 325 nm me e caused by sa ncident angle was ffected the angul peaks. Sample m measurements
articipant V and V r this sample typ angles the deviat n 4.2, Fig. 15), wh ociated with m
e more easily
utliers marked ng Sb indicated
of the #W
std.
11%
- 39%
28%
54%
- 38%
33%
50%
-
les, again, the g illumination SDF values at easurement of ample holder s observed for lar position of comparability of participant VI. An average pe. However, it tions are even hich serves to making BRDF
Fig ob
326 10 15 4.
Fig sa pr PT ins catyp ba int th pa com
Fig by
sp mlig he an th ba wia f giv co th Sb
va“co
ag
g. 9: Comparison bserved.
Table 8: Analy samples. No out
I 6%
3%
25 nm -6%
33 nm -1%
064 nm 550 nm
PTFE samples # gure 10 and Tabl ample comparabi rescreening set-u TFE is often used
struments for alibration method
pically exhibit andwidth, which
tegrated scatterin e illumination articipant V, ind ontributed BSDFs isalignment issue
g. 10: Compariso y red circles.
The measurem pecification of a
easurements are ght) with 8° inc emispherical coll ngles of +5° and
e RR results, a re ased on the mean
ith outliers remo factor to yield th ven by the ven orrected curves w
e RR measureme value of each p alues derived fro orrect” results bu Besides the ou greement with an
n of the #G samp
ysis of the meas tliers were obse 1550 II III IV
14%
1%
12%
#S
le 9 show the res ility was below th up. These samples d for the calibra the measureme ds “relative” or a rather consta is between abou ng values show n
wavelength. Tw dicating a calib s are slightly asy es.
on of the #S samp
ments were addi sample from the e for unpolarized cident angle (in
lection (instead +50°) . Hence, in presentative BSD n of the BSDF cu
ved, This average he hemispherical ndor for the ind were integrated ents, which could participant. We w om the vendor s ut can be seen as a utliers, the mea n average standar
le measurement
urement deviat erved. No data w 0 nm.
V VI
-14% 8%
-2% 6%
5% 1%
-4% -6%
-36% 36%
sults of the #S sam he measurement s are especially in ation of optical s ent of scattered
“diffuse” are use ant TS in the ut 250 nm and 25
no significant tre wo outliers we bration issue. A ymmetric indicati
ple measurement
itionally compar e same sample t illumination (ins stead of 11° in of integration b n order to make a DF for the sample
rves for the part e BSDF curve wa lly scattered light dividual wavelen over the same s d then be directly want to note tha specification do an additional sou asurements show rd deviation of 3%
ts. No outliers we
tions of the #G was provided for
VII std.
12%
- 5%8%
52%
-
mples. The deriv t uncertainty of t nteresting as wh ystems - includi d light when t ed. These sampl specified spectr 500 nm. Hence, t end as a function ere observed f Also, some of t ing sample/syste
ts. Outliers mark
red to the vend type. The vendo stead of s polariz ncident angle) an
between scatteri a comparison wi e was first obtaine ticipants at 633 n s then corrected t value, which w ngths. Then, t cattering angles compared with t at these calculat not represent t urce of compariso w a good over
%.
ere
r
ved the ite ng the les ral the of for the em
ked
dor or’s zed nd ng ith ed, nm was by the as the ted the on.
rall
Tab sample
3 6 10 15
VS.
3 6 10 15
5. Diffu The res and Tab observe outliers distincti Fig. 26) system/
effects, s the diffu which is FOV is artificia deviatio
Fig. 11:
by red c by circle Tab
Sb
32 63 106 155 Sf
32 63 106 155
6. QVD The res This un the ord
ble 9: Analysis o es with outliers the
I 1%
-1% -
325 nm 2%
633 nm -6%
064 nm 550 nm 325 nm 4%
633 nm -8% 1
064 nm 1
550 nm
usor samples #D sults of the volum able 10. Only ma ed. The measure s, identified by tive asymmetric c
). These are pro /sample misalign small illuminatio fusor. This is in c
s usually minimi smaller than the ally too low. If th
on of the measure
Comparison of th circles; forward s es / crosses, resp ble 10: Analysis o
samp I -6%
1%
25 nm OL 33 nm OL 0
64 nm -1
50 nm 25 nm OL 33 nm OL
64 nm 1
50 nm
sample sults for the QVD s nique sample was der I, VI, II, V, IV,
of the measurem s removed. The d e vendor specific
II III IV
1% 1% -1%
-1% -1% 1%
3% -3% -4%
5% -1%
0.2%
1.9% -4% -6%
1.6% -4%
-4%
me diffusor samp arginal waveleng ements showed a a (typically too characteristic of t obably caused b nment issues: As on spots spread to
conflict with the ized for under-ill e light scattering he outliers are ex
ements is 4%.
the #D sample m scattering Sf / bac pectively.
of the measurem ples with outlier
II III 5% -1% -
0% 1% 1
0.8% -0.2%
1.0% -0.2%
0%
3% -5%
1.2% -7%
0%
sample are displa s measured by th , III, VII. Similarly
ment deviations data were also c cation VS.
V V VI
% 0% 0%
% 1% 1%
OL -2%
% 11% 1%
-7% 3%
OL OL 0.1%
% 10% -1%
-9% 0.4%
OL
ples are summari gth scaling prop a surprisingly lar o low) scattering
the BRDF curves by detector field s a result of volu o a large area on e detector field o umination instru g area, the meas xcluded, the aver
measurements. Ou ckward scatterin
ment deviations rs removed.
IV V VI
1% 0% 5%
1% 1% 1%
0%
OL -2% 2%
OL 1%
OL OL OL OL 3%
OL 6%
ayed in Fig. 12 an he participants s y to the diffusor
of the #S compared to
VII std.
0% 1%
1% 0%
2%
-2% 6%
5%
-0.2% 0.3%
3%
-4% 6%
5%
-4% 0.3%
ized in Fig. 11 perties can be rge amount of g value or a s (see Annex B d of view and ume scattering the surface of of view (FOV) uments - if the sured BSDF is rage standard
utliers marked ng Sb indicated
s of the #D VII std.
-2% 4%
0% 1%
- OL 2%
1%
OL - - OL 5%
6%
OL -
nd in Table 11.
sequentially in r samples, the
BS dis va iss
Fig ma
6 B.
es Ho for co un [2 ex m ins m paac ca lev sa spsca
bu ge be Pare fun wi seun
of th co ho int#W un
SDF measuremen stribution, with alues, probably c
sues.
g. 12: Comparis arked by red circ
Table 11: Analy s I 33 nm OL Measurement
The BSDF me timated as near owever, it is som r the characte omparatively high
ncertainties of abo The GUM (Guide 6] has been deve xpression of unc easurement unce struments is easurement un arameters, e.g.
ccuracies, signal n alibration uncerta vel or optical at ample (mis-)alig peckle effects ca atterometer relat Some participan udgets for their eneral accuracy etween 0.3% a articipants II and
sults with detai nction of the sca ith a focus on a ensitivity. They
ncertainty ranges Unfortunately, p f s-polarized illum
e comparability o oncluding evalua owever, an angle teresting and the W, and #G sam ncertainties. Henc
nts show relative outliers again b caused by FOV
son of the QVD cles.
ysis of the measu samples with ou
II III -4% 2%
uncertainties an easurement un
10% for high d metimes reported rization of diff h and constant sc out 0.2% have be e to the expressi eloped to provide
certainty in me ertainty evaluatio
particularly dif certainty is inf signal drifts, d noise, reference s
ainties, signal ad ttenuator chang gnment. Moreov an lead to unce ted uncertainties nts did not con r BSDF measure of <1%; anothe and 2.1% (mai d III were the o iled BSDF measu attering angle. Th a low measurem provided their s between ∆BSDF participant III use mination because of the results with ation of the m e resolved comp erefore discussed mples, no part ce, for these samp
ely large deviatio being identified
and system/sam
D sample measu
urement deviat utliers removed
IV V V
OL OL 2
nd deviations certainty is so dynamic range m d to be substantia fuse scattering catter level, e.g. Sp een reported [23-
on of uncertainty e an internationa easurements. Ho on of angle resolv fficult. This is fluenced by a detector non-lin
ample inhomoge daption uncertain es), etc., but als ver, sample in ertainties signific s.
ntribute measure ements. One pa er participant re inly depending only laboratories urement uncerta hese participants ment uncertainty r measurements F/BSDF=0.2% an
ed unpolarized il e of budget restr h participant II, an measurement un
parison of the u d (#S and #D sam ticipant provide
ple types, only th
ons in shape of t by low scatteri mple misalignme
urements. Outlie
tions of the QVD d.
VI VII st
2% OL 3%
ometimes rough measurements [1
ally lower than th samples with pectralon®, whe -25].
y in measuremen al consensus for t owever, a detail ved light scatteri s because BSD
large number nearity, position eneity / anisotrop nty (e.g. from ga so instrument an
homogeneity an cantly larger th ement uncertain articipant stated eported a stabil on wavelength s that contribut ainty models as s also used set-u y rather than hi s with very lo
d 1%.
llumination inste rictions. This lim nd does not allow certainty budge ncertainties is st mples). For the # ed angle resolv he relative standa
the ng ent
ers
td.
%
hly 1].
his a ere nt) the led ing DF of ner py, ain nd nd han nty a ity th).
ted s a ups igh ow mits ead w a ets, till
#M, ved ard
deviatio (BSDF particip deviatio outliers BSDFIII/ and III uncerta most pa into the 1. Mirro The measur deviatio between can be angles.
samples are sign 1% or 2 probabl
Fig. 13:
of the # measur
on of the BSDF r F) is to the ave pants, and wher on of the BSDF(
s). For the other s /BSDFII - 1 betwe I (BSDFIII) is ainties. Only the m
articipants provi e analysis.
or samples #M angle resolved r rements is displa on is clearly a fu n about 3% and observed near This clearly ex s of I, IV, V, and V nificantly higher t 2%. The higher ly related to spec
Relative standar
#M sample measu red this sample (o
results (BSDF) erage BSDF(θs) re (BSDF) is (θs) functions fr sample types add een the measurem
directly compa measurements at ided results. Aga
relative standard ayed in Figure 13 unction of the sca 30% for |θs-θi|>2 the specular be xceeds sample VI) and degradatio
than the sometim fluctuations in th ckle and/or signa
rd deviations (to urements at λ=6 outliers removed
/ (BSDF) is ana that was meas the correspond rom all participa ditionally the rela ments of participa ared to their m
t 633 nm are com ain, outliers were
d deviation of th 3 (with outliers r attering angle, w 2°, while the larg eam and for lar comparability ( on. Moreover, the mes stated uncerta the BSDF of part al noise.
op) and BSDF res 633 nm for all par d).
alyzed, where sured by the ding standard ants (without ative deviation ant II (BSDFII) measurement mpared, where e not included
he #M sample emoved). The with deviations gest deviations rge scattering (12% for the ese deviations ainties of near ticipant VI are
sults (bottom) rticipants that
2.
Figof m
mth sca ag an co de 3.
m sa ab thne sa de
Window samples
g. 14: Relative st f the #W sample easured this sam The angle resol easurements is d e scattering angl attering angles l gain, the largest d nd for large scatte omparability (13
egradation, and re Grating samples # The angle resol easurements is d amples, it is a fu bout 10% and 10
e specular diffra ear the specular ample comparabi egradation, and re
s #W
tandard deviation measurements a mple.
lved relative stan displayed in Figur le, with deviation arger than 2° aw deviations can be ering angles. This 3% for the sam eported uncertain
#G
lved relative stan displayed in Figur
nction of the sca 00% for scatterin acted beams. The beams and at la ility (12% for the eported uncertain
ns (top) and BSD at λ=633 nm for a
ndard deviation re 14. Again, it is c ns between abou way from the sp e observed near s is significantly h mples of I, IV, V
nties.
ndard deviation re 15. As for the m
attering angle an ng angles larger t e largest deviatio arge scattering a e samples of I, IV,
nties.
DF results (bottom all participants th
of the #W samp clearly a function ut 6% and 80% f ecular beams. An the specular bea higher than samp V, and VI), samp
of the #G samp mirror and windo nd varies betwe than 2° away fro ons are once mo angles, and exce , V, and VI), samp
hat m)
ple n of for nd am ple ple
ple ow een om eed ore ple
Fig. 15:
of the # measur 4. PTFE
A com Figure constan valley d to abou most m deviatio sample The B (see Fig are unl screenin Similarl in the p Previou showed Howeve polariza particip 1%-2%
from a experim polariza the 6%
III. Hen not clea It is in isotropy used an calculat respecti (vendor which is
Relative standar
#G sample measu red this sample.
samples #S mparison of the 16. The relative nt over the scatt deviation of the m ut 16%. Compare measurements e ons. Also, asymm
/ instrument alig BSDF measurem g. 17), with no ove
likely to be due ng checks showe ly, anisotropy eff prescreening exp us measurement d only small az er, it is importan ation conditions pant III it was un
% have previously different supplie ments have not b ation effects cann difference obser nce, the role of m ar. nteresting to not y and normal in nd the sample co ted THR values a tively. Scaling the r specification fo s somewhere bet
rd deviations (to urements at λ=63
e #S measureme e standard devia tering angle, at a measurements of p
ed to the measure exhibit a higher metries can be o gnment effects.
ments of participan erlap in the unce e to sample com ed these deviation fects larger than periments and ca
ts on PTFE sam zimuthal deviati nt to note that the
(participant II us npolarized). Polar y been observed er [29]. As the P been characterize not be excluded a rved for the meas measurement unc te that if the BSDF ncidence (in fact
ould show mino are 101% and 9 e curve of parti for unpolarized l
tween the results
op) and BSDF res 33 nm for all par
ents at λ=633 nm ation was found
about 8%, while participant V and ements of particip r level of noise observed, possib nt II and III differ ertainty ranges. Th
mparability effec ns should be belo 1% would have an probably also mples of a diffe ions of about 1 ese participants u sed s-polarized li rization effects in d for PTFE samp PTFE samples use ed for polarizatio and could possib surements of par certainties on the F data are integra an incident ang or anisotropic be 95% for particip icipant II to a T light) yields the s of both participa
sults (bottom) rticipants that
m is shown in to be almost e the peak-to- d I correspond pant II and III, e and larger bly related to r by about 6%
he differences cts, since the ow about 1%.
e been noticed o be excluded.
erent supplier 1% [27, 28].
used different ight, while for n the range of ples, however, ed during the on properties, bly account for rticipant II and e deviations is ated assuming gle of 11° was ehaviour), the pant II and III, THR of 98.9%
e dashed line, ants.
Fig th
Figun pa Th TH 4.
m
g. 16: Relative sta e #S sample mea
g. 17: Compariso ncertainties of t articipant II and II he dotted line cor HR according to t Diffusor samples The angle resol easurements is
andard deviation asurements at λ=6
on of the deviatio the #S sample
II. The error bars rresponds to the the vendor specifi s #D
lved relative stan displayed in Fig
s (top) and BSDF 633 nm of all par
ons (top) and the measurements s correspond to u
scaled BSDF of a fication.
ndard deviation g. 18. Standard d
F results (bottom) rticipants.
e BSDF results an at λ=633 nm f uncertainties at k=
n #S sample with
of the #D samp deviations betwe
)of
for nd
=2.
h a
ple een
about 2 angles, a
For t deviated scatterin their re the pre directio the dev compar
Fig. 18:
of the #
2% and 4% can although deviatio the diffusor samp d by about +8%
ng; the uncertai esults, however, c escreening tests, on to deviate by viation of these m
rability effects rat
Relative standar
#D sample measu
be observed for ons at very large a ples, the measure for forward scat inty ranges do n correlate with th which was dete +6% between t measurements is p
ther than measur
rd deviations (bo urements at λ=63
the majority of angles exceed 10 ements of partici ttering, and +1%
not overlap. The he sample compa ermined in forwa those two partici presumably caus rement uncertain
ottom) and BSDF 3 nm.
the scattering 0%.
ipant II and III for backward deviations of arability from ard scattering ipants. Hence, sed by sample nty.
F results (top)
