Measuring the temperature of high-luminous exitance surfaces … · 2016. 9. 9. · Measuring the...
Transcript of Measuring the temperature of high-luminous exitance surfaces … · 2016. 9. 9. · Measuring the...
Measuring the temperature of high-luminous exitance surfaces with infrared thermography in LED applications
Indika U. Perera* and Nadarajah Narendran Lighting Research Center, Rensselaer Polytechnic Institute, 21 Union St., Troy, NY 12180
ABSTRACT Recently, light-emitting diode (LED) lighting systems have become popular due to their increased system performance. LED lighting system performance is affected by heat; therefore, it is important to know the temperature of a target surface or bulk medium in the LED system. In-situ temperature measurements of a surface or bulk medium using intrusive methods cause measurement errors. Typically, thermocouples are used in these applications to measure the temperatures of the various components in an LED system. This practice leads to significant errors, specifically when measuring surfaces with high-luminous exitance.
In the experimental study presented in this paper, an infrared camera was used as an alternative to temperature probes in measuring LED surfaces with high-luminous exitance. Infrared thermography is a promising method because it does not respond to the visible radiation spectrum in the range of 0.38 to 0.78 micrometers. Usually, infrared thermography equipment is designed to operate either in the 3 to 5 micrometer or the 7 to 14 micrometer wavelength bands. To characterize the LED primary lens, the surface emissivity of the LED phosphor surface, the temperature dependence of the surface emissivity, the temperature of the target surface compared to the surrounding temperature, the field of view of the target, and the aim angle to the target surface need to be investigated, because these factors could contribute towards experimental errors. In this study, the effects of the above-stated parameters on the accuracy of the measured surface temperature were analyzed and reported.
Keywords: light-emitting diodes, LED lighting system, testing, lens temperature, surface temperature, IR thermography, measurement error, surface emissivity
1. INTRODUCTION Light-emitting diode (LED) technology has improved significantly over the past decade. The technology has now reached a level where LED chips emit high radiant power, in the order of watts per chip. To create “white” light for illumination applications, the emission of these LED chips are combined with phosphor-encased optical encapsulant layers.[1] Past studies report increases in LED package lumen degradation due to both chip and encapsulant degradation caused by high temperature.[2],[3] In order to quantify LED lens degradation and improve LED performance, as well as for safety-related aspects, it is necessary to measure the lens temperature accurately during operation. With increased demand for larger lumen packages with smaller LED package footprint area, the lumen output and therefore luminous flux densities of these lenses increase. The luminous flux exiting or leaving a surface is termed as luminous exitance.[4]
In measuring the bulk medium or surface temperature of these LED lenses, either contact (intrusive) or non-contact type measurement methods can be employed. Intrusive-type measurements are commonly conducted with thermocouples. These thermocouple measurements are reported to have significant errors caused by the optical radiation emitted through the lens surface being measured.[5],[6] In addition, measurement errors can be caused by the geometrical configuration, attachment method, and heat transfer interactions of intrusive-type temperature measurements.[7] Infrared temperature measurement systems are non-contact type measurement methods that lack some of the stated sources of error such as attachment method and heat transfer interaction present in contact-type methods. Infrared temperature measurement systems are generally designed to be sensitive between 3–5 µm and 7–14 µm wavelength bands.[8] The optical emission of an LED system used for illumination is in the range of ~0.4–0.8 μm where the infrared temperature measurement systems are insensitive.
*Corresponding author: [email protected]; +1 (518) 687-7100; www.lrc.rpi.edu/programs/solidstate
Fifteenth International Conference on Solid State Lighting and LED-based Illumination Systems, edited by Matthew H. Kane, Nikolaus Dietz, Ian T. Ferguson, Proc. of SPIE Vol. 9954,
99540K · © 2016 SPIE · CCC code: 0277-786X/16/$18 · doi: 10.1117/12.2240650
Proc. of SPIE Vol. 9954 99540K-1
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Infrared temtemperaturestemperature m
In order to accurate estisurface emisdistance, bacstudy investi
This study iserror with uswas not inve
A number odifficulties oreference.
In the first stwas coated otemperature the LED mocharacteristic
A single-banspectral rangused for the Oleum V217rotate the blathe pivot axis
Figure 1. Sche
3.1 Characte
The infrared and the entirview (FOV) Figure 2. To was mounted
mperature meas as an alternmeasurement s
estimate the timation. Additssivity and geckground tempgated the effec
s mainly concesing infrared testigated nor di
of methods areof these metho
tage, the selecton one-half ofcomparison w
odule was estics.
nd wavelength ge from 7.5–14
study. Figure 78838 high-temack-coated surfs remaining ho
ematic of the hig
erization of th
imaging camere setup was pof the thermal facilitate the h
d on a heater bl
asurements havnative to consystem is very
arget surface tionally, the raeometric confiperature and suct of these para
erned with invmperature meascussed, and it
e available in ods, an altern
ted reference pf the LED len
with the charactmated using t
infrared imagi μm (Jenoptik 1 illustrates t
mperature sprayface in the two orizontal.
gh-temperature b
he applied pain
era was aimed laced ~50 mmimager was co
heating of the block. Two calib
ve been usedntact-type temimportant to ob
temperature, tange of tempergurations, suc
urface emissiviameters on the
2. Mvestigating andasurement systt is assumed th
literature to cative two-stag
paint coated onns surface. Thterized paint-cthe infrared im
3.ing camera moIR-TCM 384 w
the experimenty paint) coatedperpendicular
black paint chara
nt coating
normal to the cm from the optovered by the bblack paint-coabrated, J-type
d in the electmperature meas
btaining an acc
the emissivity ratures during
ch as infrared ity of the targLED lens surfa
METHODOLd quantifying mtems. Calibratioat the measure
characterize ange method wa
n a target surfahe different macoated surface. maging camera
EXPERIMEodule with a fowith a 384×24tal setup used
d surface. The er axes (indicate
acterization expe
crossing-point ics of the therblack paint-coaated surface to30 AWG therm
tronics and Lsurements.[9],[10
curate, repeatab
of the target operation thatemperature
et, can also aface temperatur
LOGY measurement eon of the infrarment system is
n unknown taas developed u
ace was characaterial surfaceThen the surf
a based on the
ENT focal plane arra44 pixel resolut
for the characexperimental sed by the black
erimental setup.
of the pivot axrmal imager. Tated surface ofo different tempmocouples wer
LED industries0] The calibrable measureme
surface becomat might causemeasurement
ffect the estimare.
errors associatered temperaturs in calibration
rget surface.[8]
using a flat-bl
cterized. Then ts were then cface temperatue experimental
ay detector (FPtion of bolomecterization of
setup had the ak arrows) on the
xes on the blacThis was to ensf the calibrationperature valuesre embedded in
s to measure ation of the ent.[8]
mes a prerequ changes to thsystem aiminated temperatu
ed with the syre measuremenn.
] Due to the plack paint coa
the characterizcharacterized bure during operlly determined
PA) having a setric FPA detecthe black pain
ability to indepe horizontal pl
ck paint-coatedsure the entiren block, as inds, the calibration machined bli
surface infrared
uisite for he target g angle, ure. This
ystematic nt system
practical at as the
zed paint based on ration of
d surface
sensitive ctor) was nt (Rust-endently ane with
d surface e field of icated in on block ind holes
Proc. of SPIE Vol. 9954 99540K-2
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
in the aluminblock and the
An Omega Tthermocouplthermocoupl
Figure 2. Sche
The heater brespect to thtemperature. temperature. that the infradefault surfaunderestimattemperature imager meastemperature.
Figure 3. Char
num calibratione blind holes fo
TC-08 8-channes. The therme calibrator.
ematic diagram o
block was heathe embedded t
Figure 3 illuThe line runni
ared thermal imace emissivityted the surfacwas greater th
surement accur
racterization of t
n block. The cor the thermoco
nel USB thermocouples and
of the infrared th
ted to a numbthermocouple
ustrates the infing diagonally
mager should hy of ε=1.00 (ice temperaturehan ±1.5 K betracy).[11] Figure
the surface emiss
ross section onouples (thermo
ocouple data athe data acqui
hermal imager an
er of predefinmeasurementsfrared imaging
y across the chahave estimatedideal blackbode at temperatutween 0 and 1e 3 shows that
sivity of the blac
n the right of Focouple wells).
acquisition moisition system
nd experimental
ned temperature, the infrared
g camera-estimart is provided d at a given thedy) overestimaures ≥80°C.
100°C and ±2%t the paint coat
ck paint-coated s
Figure 1 indica.
dule was usedwere checked
setup.
e values and othermal image
mated temperatas a visual gu
ermocouple-mated the tempThe deviation% of the measting’s surface
surface.
ates the dimens
d to acquire thed for calibratio
once steady-staer was used toture vs. the thide to illustrateeasured tempeerature at tem
n from the thsured temperatemissivity dec
sions of the ca
e temperature fn using a Fluk
ate was achievo estimate thehermocouple-me the ideal tem
erature. This shmperatures ≤40hermocouple-mture (based on creases with in
alibration
from the ke 714B
ved with e surface measured mperature hows the 0°C and
measured thermal
ncreasing
Proc. of SPIE Vol. 9954 99540K-3
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Table 1 shomeasured teemissivity va
Table 1. Effec
The calibratirule) at a timwas varied tFigure 4 indidependent emtemperature aiming anglefrom the thetemperaturesrotation angl
Figure 4. Angu
To assess thethe black paithermal imagaddition, thehigh-temperamaintained a
ows the effectiemperature andalues were used
ctive surface emi
Temperatur
≤8
>80–
>160
ion block was me, as illustrateto three distincicates the variamissivity chanrepresents the
es less than 30°ermocouple-mes. A greater dee exceeded 30°
ular dependence
e effect of FOVint-coated surfger optics to t
e surface emissature black paat 25±2°C.
ive surface emd the infraredd in the subseq
issivity of black
e range (°C)
80
≤160
–200
rotated both ied in Figure 1ct temperature ation of the infrnge with the e mean temper° from the surfeasured tempe
eviation more t° for surface te
e of infrared ther
V and target bface on axis withe backgrounsivity of the b
aint (ε>0.9) an
missivity calcud thermal imaquent estimatio
paint-coated sur
n the positive , while maintavalues while
frared thermal ithermocouple-
rature of four face normal oferature was lethan the measuemperatures ab
rmal imager temp
background emith the infrared
nd plate was vbackground pland with no sur
ulated to haveager-estimated
on of surface te
rface.
Effe
and negative aining the surfameasuring theimager-estimat-measured temmeasurements
f the target surss than the murement accura
bove ~50°C.
perature estimat
missivity, a simd thermal imagvaried to changate was also chrface coating (
e the minimumtemperature.
mperature.
ective surface e
0.96
0.92
0.90
direction on oface temperature temperature ted temperatur
mperature. Thes except for thrface and therm
measurement acacy of the ther
te.
milar experimenger, as indicatege the FOV ohanged by inte(ε<0.2). The b
m error betweThese estima
emissivity
one axis (accorre constant. Thusing the emb
re after adjustine infrared therhe 0° temperatmal imager optccuracy of thermal imager w
nt was conducted in Figure 5. of the black paerchanging the
background sur
een the thermoated effective
rding to the righe surface tembedded thermong for the temprmal imager-eture measuremical axis, the de device at th
was observed w
ted with the noThe distance f
aint-coated sure plates with trface temperat
ocouple-surface
ght-hand mperature ocouples. perature-stimated
ment. For deviation he tested when the
ormal of from the rface. In the same ture was
Proc. of SPIE Vol. 9954 99540K-4
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Figure 5. Conf
Figure 6. FOVε>0.9 (left) an
Figure 6 showith the therdistance corr
In order to aexperimentalFigure 7 shomeasurement
figuration of the
V and backgrounnd background su
ows there was rmocouple-mearesponded to a
ssess the effecl setup was us
ows the observt accuracy of th
experimental se
nd emissivity efurface emissivity
no significantasured temperaFOV 75 mrad.
ct of backgrouned with the ad
ved results fromhe device with
etup for FOV an
ffect on infraredy ε<0.2 (right).
t effect on theature. The 50 m.
nd temperaturedded feature om the experimeh the thermocou
d background ef
d thermal image
infrared thermmm distance c
e on the infraref heating the pent. The infraruple-measured
ffects.
er-estimated tem
mal imager-estcorresponded to
ed thermal imaplate with the red imager-estitemperature v
mperature: backg
timated tempero a FOV 300 m
ager-estimated through hole iimated temper
values.
ground surface e
rature when comrad, and the
temperature, tillustrated in F
ratures were w
emissivity
ompared 200 mm
the same Figure 5.
within the
Proc. of SPIE Vol. 9954 99540K-5
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Figure 7. Back
3.2 Characte
These initialfollowed by surface tempFigure 8. Themissivity vamodule compside of the coto the predeadjusted unti
Figure 8. Sche
Figure 9 sholens. It also close to the gmodule was rsurface. The with increase
kground tempera
erizing LED m
l experiment rangular-depen
perature, the LEen the LED malues for the ponents are apoated and uncotermined valuil the two temp
ematic of the LE
ws the tempershows the anggeometric centrotated along tflat areas of th
e in temperatur
ature effect on in
module surfac
results showedndent errors atED module wa
module was moudifferent mate
pproximately aoated halves is ue based on thperature values
ED module and o
ature matchinggular variation ter of the LEDthe horizontal ahe LED modulre.
nfrared thermal i
ce
d that the majt aiming angleas coated with unted on a hea
erial surfaces iat the same tem
assumed identhe paint charac
matched.
orientation for th
g of the black pof the estimat
D module indicaxes with increle used to asse
imager temperatu
or systematic es larger than the characteriz
ater block and in the LED m
mperature. Alontical. The surfacterization. Th
he experiment wi
paint-coated suted temperaturecated a larger seasing temperaess the emissiv
ure estimate.
error stems f30°. In order
zed black painthe infrared th
module. At anyng the line of
ace emissivity ohen the unkno
ith the infrared th
urface and the e. The estimatspreading of thature. This was ity of the solde
from surface er to estimate tt on half the su
hermal imager y steady-state symmetry, theof the black-cown material s
hermal imager.
silicone surfaced temperature
he estimated tedue to the geo
er mask did no
emissivity adjuthe LED lens urface, as illuswas used to astemperature, t
e temperature ooated side was surface emissiv
ce of the LED e on the same emperature as tometry of the Lot show a simi
ustments exitance
strated in ssess the the LED on either adjusted vity was
primary location the LED
LED lens lar trend
Proc. of SPIE Vol. 9954 99540K-6
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Figure 9. Eval
3.3 Estimati
Two identicamodule was eliminating lmounted on Perfluoroalkolens temperatemperature thermocoupl2500 Siliconthermal imathermocoupl
Figure 10. Infr
Figure 10 shthe case tem
luation of LED p
ing high-lumin
al samples of tcoated with
uminous flux (heat sinks an
oxy (PFA) insuature at the geof the LED mes. To attach t
ne, Two Part Aager. A two-pe at the case te
rared thermal im
hows the infrarmperature locati
primary lens emi
nous exitance
the same LEDthe same blac(optical radiatind operated inulated thermoc
eometric centerodule was alsothe thermocoupAdhesive and Spart permanenemperature loca
mager and thermo
red imager- andion of the blac
issivity by match
surface tempe
D module that ck paint whileon) from being
n constant curcouples (Omegr of the lens eo measured usiple to the LEDSealant) was unt adhesive (Aation.
ocouple tempera
d thermocouplck paint-coated
hing temperature
erature
was used in the the other wag emitted fromrrent mode. Thga SRTC-TT-Jexitance surfacing both the in
D lens exitanceused and was rArctic Alumin
ature measureme
le-measured ted (no radiant f
e.
he characterizaas uncoated. T
m the primary lehe infrared th
J-36-36) were uce at the samenfrared thermale surface, an opremoved after na Thermal A
ents of the energi
emperatures at flux-emitting L
ation process wThe black coaens surface. Bohermal imager used alternatele drive currentl imager and thptical silicone each measurem
Adhesive) was
ized LED modu
the geometric LED module) a
were selected,ating was suffioth LED modu
and J-type 3ly to measure tt conditions. The same type oelastomer (Nu
ment with the s used to att
le.
center of the and the uncoat
and one ficient in ules were 6 AWG the LED The case of J-type uSil CV-infrared
tach the
lens and ted LED
Proc. of SPIE Vol. 9954 99540K-7
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
modules. Thepaint coatingcase tempera
The uncoatedafter adjustinthe charactergeometric cewas assessedcenter overes
A prototype packages (XPsilicone layeE3634A dc pto create a simm³. The alu36 AWG Persink near theselected to mtemperature b
Figure 11. Sch
Figure 12 scharacterizatthermocoupltemperature higher of 1.5
e thermal imagg matched the ature location.
d LED moduleng for the solderization procesenter was measd during the chstimated the su
LED light engPEBRY-L1-00
er heat sink, aspower supply×ilicone layer (uminum heat sirfluoroalkoxy e silicone layermaintain a unibased on past s
hematic of the LE
shows the siltion of the surfe measuremenrange of three K or 2% of th
ger temperaturethermocouple
e case temperater mask surfacss used for thesured with the
haracterization urface temperat
gine was desig000-00P01) arrs illustrated in2). A silicone diameter of 4 ink was attach(PFA) insulater on the side fiform temperatstudies.[12]
ED light engine
licone layer tface emissivitynt of the hea
e measurementhe measured tem
e estimates aftemeasurement
ture measuremce emissivity, we thermal imag
infrared imagstage. Figure 1ture due to the
gned with an Lranged in an arn Figure 11. Telastomer (Numm and 1 mmed to a heating
ed thermocoupfacing away froture distributio
with heated silic
temperature ey as described at sink tempes. The estimatmperature at al
er adjusting forboth at the ge
ments from the twhich was founger temperatur
ger and adjuste10 also indicatabsorption of t
LED array conrray to provide
The LEDs wereuSil CV-2500 Sm thick) and wg element to cole was used toom the LED iron on top of t
cone layer exper
estimated usinin the previou
erature. The eted temperaturell conditions.
r the temperatueometric center
thermal imagend during the cre estimation. ed for the surfates the thermocthe optical radi
nsisting of 21 Ce a uniform irre operated at aSilicone, Two Pwas embeddedontrol the tempeo measure the trradiance. Thethe silicone lay
riment setup.
ng the infrareus sections. Therror bars repe from the infr
ure-dependent r of the lens an
r and the thermcharacterizatioThen the lumiace emissivity couple measureiation.[5],[6]
Cree XLamp Xadiance on thea 700 mA conPart Adhesive in an aluminuerature of the stemperature of diameter of thyer, which is
ed thermal imhis temperaturepresent the mifrared thermal
emissivity of tnd at the LED
mocouple also n stage. This vinous exitanceof the silicone
ement at the ge
XP-E2 royal ble silicone layernstant current and Sealant) w
um heat sink 6silicone layer. Af the silicone lahe phosphor laat the same h
mager after a e is compared inimum to mimager was w
the black D module
matched validated e surface e, which eometric
lue LED r and the (Agilent
was used 60×60×1 A J-type
ayer heat ayer was heat sink
similar with the
maximum ithin the
Proc. of SPIE Vol. 9954 99540K-8
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
Figure 12. Inf
The infrared measurementfactors were
The authors H11014), thInstitute and
The authors wassistance in
[1] Muell(2000
[2] Barto3279,
[3] TamuV ope
[4] Murd2nd e
[5] Perera(2013
[6] Fu, XInt. J.
[7] FiglioSons
frared imager tem
thermal imaget accuracy aftthe main sourc
would like to ahe Federal Av
the Lighting R
would like to apreparation of
ler-Mach, R. a0); doi: 10.1117
n, D.L. et al., 17–27 (1998)
ura, T. et al., “Ieration,” J. Lum
doch, J.R., “Intd., ch. 1, sec. 7
a, I.U. et al., 3); doi: 10.1117
X. and Luo, X., Heat and Mas
ola, R.S. and BInc., Hoboken
mperature estima
er was able to er the surface ces of systemat
acknowledge tviation AdminiResearch Cente
acknowledge thf this manuscrip
and Mueller, G7/12.382840.
“Life tests and; doi: 10.1117/
Illumination chminescence 87-
troducing Ligh7, pp. 20-30, V
“Accurate me7/12.2023091.
“Can thermocss Transfer 65,
Beasley, D.E., , NJ, 309–367
ate of the silicon
4. Cpredict the suremissivity ad
tic error in the
ACKNOWthe Alliance foistration (FAAer (IRA) for fin
he efforts of Jept and the staff
REFG.O., “White-l
d failure mech/12.304426.
haracteristics o-89, 1180–118
ht and Seeing,”Vision Commun
easurement of
couple measure, 199-202 (201
[Theory and D(2011).
ne layer with lum
CONCLUSIOrface temperatudjustments andestimated temp
WLEDGMEor Solid-State IA Research Gnancial assistan
ennifer Taylor ff at the Lightin
FERENCESlight-emitting d
anisms of GaN
f lighting array2 (2000); doi:
” in Illuminatinications, New
LED lens sur
e surface tempe3); doi: 10.101
Design for Me
minous exitance
ONS ure of the lum
d maintaining peratures from
ENTS Illumination Sy
Grant Number nce during this
of the Lightinng Research Ce
diodes for illu
N/AlGaN/InGa
y using 10 cand10.1016/S0022
on Engineerinw York (2003).
rface temperatu
erature of light16/j.ijheatmass
echanical Mea
surface.
minous exitancethe aiming an
m the infrared th
ystems and Te13-G-009), Rresearch proje
ng Research Ceenter.
umination,” Pr
aN light emittin
dela-class whit2-2313(99)005
ng: from Ediso
ure,” Proc. SP
t emitting diodstransfer.2013.0
asurements, 5th
e surfaces to wngle below 30°hermal imager.
chnologies (ARensselaer Polect.
enter for her in
oc. SPIE 3938
ng diodes,” Pro
te LEDs under 588-8.
on’s Lamp to th
PIE 8835, 883
de module accu05.072.
h ed.], John W
within the °. These .
SSIST # ytechnic
nvaluable
8, 30–41
oc. SPIE
r AC 100
he LED,
3506-1-6
urately?”
Wiley and
Proc. of SPIE Vol. 9954 99540K-9
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx
[8] Kaplan H., [Practical Applications of Infrared Thermal Sensing and Imaging Equipment, 3rd ed.], SPIE-International Society for Optics and Photonics, Bellingham, WA (2007).
[9] Garg, J. et al., “Micro fluidic jets for thermal management of electronics,” Proc. ASME 2004 Heat Transfer/Fluid Engineering Summer Conference 4, 647–654 (2004); doi: 10.1115/HT-FED2004-56782.
[10] Arik, M. and Stanton, W., “Effect of chip and bonding defects on the junction temperatures of high-brightness light-emitting diodes,” Optical Engineering 44(11), 111305-1-8 (2005); doi: 10.1117/1.2130127.
[11] JENOPTIK Laser, Optik, Systeme GmbH, “JENOPTIK IR-TCM 384 & 640 Manual,” Revision 003 (2009).
[12] Perera, I.U., “Thermal management of remote phosphor layer in a light-emitting diode system,” M.S. thesis, Dept. Lght., Rensselaer Polytechnic Institute, Troy, NY (2013).
Proc. of SPIE Vol. 9954 99540K-10
Downloaded From: http://proceedings.spiedigitallibrary.org/ on 09/09/2016 Terms of Use: http://spiedigitallibrary.org/ss/termsofuse.aspx