Performance of a novel PMMA polymer imaging bundle for field acquisition and wavefront sensing

13.4 MB·

by  Samuel Nathan Richards

Publications of the Astronomical Society of Australia(PASA)
(cid:13)c AstronomicalSocietyofAustralia2017;publishedbyCambridgeUniversityPress.
doi:10.1017/pas.2017.xxx.
Performance of a novel PMMA polymer imaging bundle for
field acquisition and wavefront sensing
S. N. Richards1,2,3,4∗, S. Leon-Saval1,4, M. Goodwin2 J. Zheng2, J. S. Lawrence2, J. J. Bryant1,2,3,
7 J. Bland-Hawthorn1,4, B. Norris1,4, N. Cvetojevic1,2,4,5,6 and A. Argyros6
1
0 1SydneyInstituteforAstronomy,SchoolofPhysics,UniversityofSydney,NSW2006,Australia
2AustralianAstronomicalObservatory,POBox915,NorthRyde,NSW1670,Australia
2
3CAASTRO:ARCCentreofExcellenceforAll-skyAstrophysics

n 4SydneyAstrophotonicInstrumentationLaboratory(SAIL),SchoolofPhysics,UniversityofSydney,Sydney,NSW2006,Australia
a 5CUDOS:ARCCentreofExcellence,CentreforUltrahighbandwidthDevicesforOpticalSystems
J 6InstituteofPhotonicsandOpticalScience(IPOS),SchoolofPhysics,UniversityofSydney,Sydney,NSW2006,Australia

3
1
Abstract

] Imagingbundlesprovideaconvenientwaytotranslateaspatiallycoherentimage,yetconventionalimaging
M
bundles made from silica fibre optics typically remain expensive with large losses due to poor filling fac-
I tors (∼40%). We present the characterisation of a novel polymer imaging bundle made from poly(methyl
. methacrylate)(PMMA)thatisconsiderablycheaperandabetteralternativetosilicaimagingbundlesover
h
p shortdistances(∼1m;fromthemiddletotheedgeofatelescope’sfocalplane).Thelargeincreaseinfilling
- factor(92%forthepolymerimagingbundle)outweighsthelargeincreaseinopticalattenuationfromusing
o PMMA (1dB/m) instead of silica (10−3dB/m). We present and discuss current and possible future multi-
r
t objectapplicationsofthepolymerimagingbundleinthecontextofastronomicalinstrumentationincluding:
s fieldacquisition,guiding,wavefrontsensing,narrow-bandimaging,aperturemasking,andspeckleimaging.
a
[ TheuseofPMMAlimitsitsuseinlowlightapplications(e.g.imagingofgalaxies),howeveritispossibleto
fabricate polymer imaging bundles from a range of polymers that are better suited to the desired science.

1
v Keywords: instrumentation: miscellaneous – telescopes – instrumentation: adaptive optics – instrumenta-
0
tion: high angular resolution
6
7
3
0 1 INTRODUCTION the fibre, leaving only the ends bundled together. The
. delicate process of making a flexible silica fibre bundle
1 Over the course of the past few decades, the act of
0 increases the cost to ∼US$1000 per metre. High filling
bundling optical fibres together to translate an image
7 factorsilicafibrebundlesdoexist(e.g.Bland-Hawthorn
in a spatially coherent manner has been optimised to
1 et al. 2011; Bryant et al. 2014), though their large in-
: increase the spatial resolution, yet retaining efficiency
v dividual core diameters (∼100µm) results in a coarse
andflexibility.Thesedevicesareknownasimagingfibre
i spatialresolution,andwiththebundleonlybeingfixed
X bundles(orcoherentfibrebundles),andarewidelyused
at one end, they lack the ability to preserve a spatially
r for remote sensing with applications such as biomedi-
coherentimageandthereforearenotclassedasimaging
a cal endoscopy. Most fibre bundles are made using sil-
bundles.
ica optical fibres, however the process used to make
The use of polymer as an alternative to silica in fi-
a ∼1mm diameter bundle that remains flexible sac-
bre imaging bundles presents an attractive solution to
rifices efficiency due to the fibre filling factor. This is
many of the issues faced by using silica (e.g. the abil-
because for most of the length, each fibre needs to be
ity to obtain a high filling factor whilst retaining flexi-
loose, otherwise a 1mm bundle would be a solid rod
bility). One of the most common polymer fibre materi-
withnearzeroflexibility.TheSCHOTTLeachedImage
alsispoly(methylmethacrylate)(PMMA),howeverthe
Bundles1 are a good example of this, where the inter-
use of PMMA (and most polymers) over silica comes
corecladdingmaterialisetcheddownoverthelengthof
at the cost of attenuation (on the order of 1dB/m
for PMMA versus 10−3dB/m for silica). This increase
∗E-mail:[email protected] in attenuation limits the applications of polymer fi-
1http://www.schott.com/lightingimaging/english/medical/ bre to cases where fibre lengths on order of a metre
medical-products/transmitting-images_leached-image-
are required (e.g. endoscopy, localised remote sensing).
bundle.html
1
2 Richards et al.
Albeit,longerlengthsofthesepolymerimagingbundles
can be used in applications where the source is of high
enough power to overcome the attenuation loss.
An area where fibre optics continue to increase in
their application is astronomy, where they have be-
come the dominant method of translating the light of
thousands of galaxies and stars from the telescope fo-
cal plane to nearby spectrographs. Most focal planes of
currentlarge(4-10m)telescopesare(cid:46)1mindiameter,
but there are soon to be focal planes of 1-2m diameter
uponthearrivalofExtremelyLargeTelescopes(ELTs).
These large focal planes require precise field acquisi-
tion and guiding, for which coherent imaging bundles
are well suited. In this application, at one end, multi-
ple imaging bundles are deployed over the focal plane
targeting guide stars, and at the other end, all of the
imaging bundles are brought together with one camera
situated away from the focal plane imaging their end
faces.ThismethodhasbeenadoptedbytheSloanDig-
italSkySurveymulti-objectspectrograph(SDSS;Smee
et al. 2013) and the Sydney-AAO Multi-object Integral
field spectrograph (SAMI; Bryant et al. 2015). SDSS
use silica imaging bundles from Sumitomo2, and SAMI
use polymer imaging bundles from ESKA3.
Basic characterisation of the Sumitomo silica image
bundles has been performed by Haynes et al. (2006),
whoanalysedthespatialfidelity,crosstalkandscatter-
ing,andconcludedthatmoreinvestigationintoimaging
bundlesisneededtoassesstheirsuitabilityinastronom-
ical applications. The Sumitomo bundle has 2µm cores
spacedby3µminasemi-regularhexagonalpattern,re-
sultinginafillingfactorof∼40%(Udovichetal.2008).
This immediate substantial loss from the filling factor
isdetrimentalinastronomicalapplicationsofthesespa-
tially coherent imaging bundles, where every photon
counts. In addition to the filling factor, core sizes of
2µmareveryclosetobeingpurelysingle-modedatop-
ticalwavelengthsresultinginanevengreaterlossdueto
very poor modal coupling (Horton & Bland-Hawthorn
2007). In the case of the Sumitomo bundle, the outer
diameter is only 800µm, meaning that even without
separating the individual cores along the length of the
fibre, the bundle retains some flexibility.
More recently, SCHOTT have developed a Wound
Fibre Bundle4 for the Defense sector, that creates a
spatially coherent array of smaller 6×6 fibre arrays
with each fibre being 10µm in diameter. Its efficiency Figure1.:(top)Back-illuminatedmicroscopeimageofa
over the wavelength range 400−1500 nm is typically coherent polymer bundle showing the entire face of the
40-50%, including attenuation, fill-factor and Fresnel bundle.(middle)amagnifiedimageof(top)showingthe
surfacereflectionlosses.Detailedcharacterisationonits hexagonal structure of the cores. There are 7095 cores
optical performance has yet to be carried out. hexagonally packed, with each core being 16µm (full-
diagonal) resulting in an overall diameter of 1.5mm.
2SumitomoElectricLightwaveCorp., Dust in the microscope imaging system gives the ap-
http://www.sumitomoelectric.com pearance of damaged cores. (bottom) a projected Hα
3ESKATM;MitsubishiCorp.,http://fiberopticpof.com
mapofaspiralgalaxybybutt-couplingtheinputendof
4SCHOTTWoundFibreBundle,http://www.schott.com/
lightingimaging/english/defenseproducts/wound.html thebundletoa1.5mmgrayscaleprintoutofthegalaxy.
PASA(2017)
doi:10.1017/pas.2017.xxx
Performance of a novel polymer imaging bundle 3
Our interest then lies with the ESKA polymer imag- This is a factor of 2-3 improvement over the best silica
ing bundles (PMMA), which have a larger filling frac- imaging bundles (25−40%). This gain in filling factor
tionthansilicaimagingbundles,yetretainspatialreso- iscriticalforastronomy,wheremissingspatialinforma-
lutionandflexibility.InSection2wepresenttheresults tion can be detrimental to the science. When the filling
of laboratory characterisations of two different sizes factor of this polymer imaging bundle is compounded
of polymer imaging bundles. In Section 3 we explore with the typical attenuation of PMMA (1dB/m), it is
their various applications in the context of astronom- easytosee,fromFigure2,thatinshortlengths(<4m)
ical instrumentation, presenting recent on-sky demon- thepolymerimagingbundlestheoreticallyout-performs
strations, and concluding our work in Section 4. even the best silica imaging bundles (0.01dB/m and
40% filling factor).
By evenly back-illuminating the bundle, using an
2 Optical Performance
OLED flat-screen source through a Bessel-Johnson R-
Tobetterunderstandtherangeofpossibleapplications band filter, the flat-field response profile of the 1.5mm
polymer imaging bundles have in astronomical instru- polymer imaging bundle was found by taking the mean
mentation,weperformedlaboratorycharacterisationon of image pixel intensities within radially increasing an-
their filling factor, flat-field response, spectral through- nuli (see Figure 3). Each annulus has a width of 16µm
put,cross-talkandfocalratiodegradation.Weexplored (onecore),withtheimageinitiallydarksubtractedand
twosizesofESKApolymerimagingbundles,withexter- thensubtractedbythemaximuminter-coreintensityto
naldiametersof0.5and1.5mm.Theyarebothmadeby reduce the contamination of cladding light. The radial
the samemethod, with the 0.5mm bundle being drawn profile is observed to have a slight decrease of ∼1%
downfurtherintheprocess,resultinginthesamenum- out to a radius of ∼600µm before decreasing rapidly
ber of cores, but smaller core diameters. Both bundles to∼90%attheedgeofthebundle.Weconjecturethat
have 7095 hexagonally-shaped cores, with each core di- thistypeofradiallydecreasingprofileislikelylinkedto
ameter being 16µm (full-diagonal) in the 1.5mm bun- the intrinsic scattering and crosstalk properties of the
dle and 6µm (full-diagonal) in the 0.5mm bundle. In bundlestudiedinSection2.3.Linearpropagationlosses
some of the tests we present the results of both bun- resulting in scattered light out of the central cores will
dlesizes,althoughweprimarilytestedthe1.5mmbun- result in an increased crosstalk between neighbouring
dle.Aback-illuminatedmicroscopeimageofthe1.5mm cores, however the same scattering mechanism in the
bundle can be found in Figure 1. In all tests, the end outer diameter cores will result in power loss out of the
faces were machine polished5 stepping down through composite waveguide, i.e. bundle. Variations in inten-
30,12,9,3,1,0.3µm grit paper. sity within the annuli, in addition to the radial profile,
are easily corrected for by flat-fielding in a data reduc-
tion process. The 1.5mm polymer imaging bundle is
2.1 Filling factor and flat-field
therefore deemed acceptable for astronomical applica-
Thedominantsourceoflossesinfibrebundles,interms tionsthatrequireaccuratefluxinformationofasource.
of total efficiency, is a poor filling factor, where the to- This is true even when the source is extended spatially.
tal area at the bundle end-face that can be sampled by
the cores is less than the overall area. As the cores are
2.2 Spectral transmission
what guides the incident light and maintain the spa-
tial coherence (with the rest of the bundle consisting of Althoughtheuseofpolymercanincreasethefillingfac-
the cladding), any light not coupled is at best lost, or tor of imaging bundles dramatically (see Section 2.1),
at worst couples into the cladding that contributes to it has much larger optical attenuation compared to sil-
cross-talk. Due to the small core sizes found in imag- ica (1dB/m and 10−3dB/m respectively). To see how
ing bundles, sufficient cladding to confine propagating the attenuation varies as a function of wavelength, the
modesresultsinlargerseparationsbetweencores.Ifthe spectral throughput of both polymer imaging bundles
space between cores becomes too small then the modes was measured via the cut-back method (see Figure
can easily leak into adjacent cores, spreading out the 4). A reference spectrum was obtained by coupling a
spatial information and increasing light losses. Creat- white-light source into a 200µm core-diameter silica
ing a large refractive index difference between the core patchcordandmeasuredwithanopticalspectrumanal-
and the cladding allows the core-to-core separation to yser (OSA). End-face polished (0.3µm grit) and con-
besmaller,whichispossibletoachievewhenusingpoly- nectorised lengths of the imaging bundles were then
mer rather than silica (see Section 2.3). By analysing a butt-coupled to the patch-cord and fed into the OSA.
front-illuminated image of the 1.5mm polymer imag- The lengths of imaging bundles were 10,5,2,1m, re-
ing bundle, the filling factor was measured to be 92%. polishing and re-connectorising after each cut. The me-
dian spectral transmission per metre of each bundle
5KrelltechRev2MicroPolisher was determined using the measurements of the mul-
PASA(2017)
doi:10.1017/pas.2017.xxx
4 Richards et al.
1.0 1.02
) 1.00
d
0.8 n
a
nd) -Rb 0.98
a (
-Rb 0.6 ncy 0.96
Efficiency( 0.4 malisedefficie 00..9924
0.2 or
N
0.90
0.0 0.88
10-2 10-1 100 101 102 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Fibrelength(m) Distancefromcentre(mm)
Figure 2.: Theoretical R-band efficiency as a function Figure 3.: A normalised radial profile of the 1.5mm
of fibre length for a 1.5mm polymer imaging bundle polymer imaging bundle. Each measurement is the
(blue line) and a best silica imaging bundle (dashed mean intensity of pixels within annuli of widths 16µm
line),wheretheefficiencyincludesthefibreattenuation placed on an R-band back-illuminated flat-field image
(1dB/mand0.01dB/mrespectively)andthefillfactor of the bundle. Error bars represent the standard devi-
(92% and 40% respectively). ation of each measurement.
1.00
U B V R I
0.80
r
e
t
e
m
r
e 0.60
p
n
o
si
s
mi 0.40
s
n
a 1.5mm
r
T
0.5mm
0.20
CYTOP
Silica
u g r i z
0.00
400 600 800 1000 1200 1400
Wavelength(nm)
Figure 4.: Spectral throughput of the 0.5 and 1.5mm polymer imaging bundles (red and blue respectively). The
green line is the transmission profile of a new polymer material, CYTOP, taken from Argyros (2013), and the
magenta line is the manufacture’s transmission of AFS200220T silica fibre. Bessel-Johnson (UBVRI; dotted black
line) and SDSS (ugriz; solid black line) spectral filters are overlaid for reference (arbitrarily scaled). There was no
transmission of wavelengths less than ∼350nm and greater than ∼1100nm for either bundle.
tiple lengths normalised by the reference spectrum. an additional loss of ∼40%. Both transmission curves
The absorption bands seen in the spectral transmis- have been corrected for Fresnel end face losses. Figure
sion curve (Figure 4) are due to the molecular vibra- 4 also shows the spectral transmissions for a new poly-
tions of the PMMA, becoming essentially opaque to mer, CYTOP6 (Cyclic Transparent Optical Polymer),
light of wavelengths shorter than ∼350nm and longer and a single silica fibre (AFS200220T7).
than ∼1100nm. The aforementioned losses that arise
from the single-mode nature of the cores in the 0.5mm 6CYTOP,AsahiGlassCo:
http://www.agc.com/kagaku/shinsei/cytop/en/
bundle are clearly evident in the scaling difference be-
7AnhydroguideFiberSilica(AFS),FiberguideIndustries:
tween the transmission of the two bundles, resulting in http://www.fiberguide.com/product/optical-fibers/
PASA(2017)
doi:10.1017/pas.2017.xxx
Performance of a novel polymer imaging bundle 5
0 1.0
) 1 0.8
y
t
ensi sity
t n
in 2 0.6 te
d n
e i
s d
ali se
orm 3 0.4 mali
n r
( o
10 n
g
o 4 0.2
l
5 0.0
Figure5.:(left)Alogscaledinvertedimageafterbutt-couplingandaligningasilicasingle-modefibretothecentre
of one core. (right) The intensity profile of the cut shown overlaid on (left) going from left to right. The red line is
the log scaled intensity profile and the blue line is intensity on a linear scale. Both lines have been normalised to
the maximum of the injected core. 3% of the injected light has been scattered to the surrounding cores.
2.3 Focal ratio degradation, cross-talk and ratiodegradation(FRD),wheretheconeangle(numer-
scattering ical aperture; NA) of the input light becomes larger at
the output of the fibre (or that the encircled energy of
Due to the chemical composition and imperfections of
a given profile is less at the output than at the input
optical fibres, conservation of the input light is diffi-
light).
cult. This is even more so in standard polymers (e.g.
Knowledge of a fibre’s FRD is important when de-
PMMA); as demonstrated by the high attenuation lev-
signing an optical system that makes use of optical fi-
elswhencomparedtosilica,duetothehigherscattering
bres. It dictates the power of the optical lenses used
ofthepolymermedia.Unlikewhenmakingopticalfibres
to image the near-field (face) of the fibre, whether as
from silica, the ability to mix polymers together makes
part of an imaging system or spectrograph. As the in-
it possible to increase the numerical aperture of the fi-
dividual core diameters of the polymer imaging bun-
bre by creating a larger refractive index difference be-
dles are small (6µm and 16µm), it was not possible to
tweenthecoreandthecladding.Thisallowsforahigher
measure the FRD via the standard method of inject-
number of modes to propagate down the fibre and bet-
ing a focussed spot of light (of varying NAs) down a
ter constrains the mode field diameters. However, the
single core and measuring the NA of the output light
chemicalstructuresofPMMAandsilicaareverydiffer-
(Bryant et al. 2014). Another technique to measured
ent,leadingtoamuchlargerscatteringtermforPMMA
FRD is to inject collimated light at varying angles, but
thansilica.Thisincreaseinscatteringalsoincreasesthe
this too requires injection into a single core. Therefore,
chanceofcouplingbetweendifferentmodes.Lightscat-
no FRD measurement was taken, but the large accep-
tering can enhance the coupling from low order modes
tance NA of the polymer imaging bundle would indi-
to higher order modes closer to the core mode cut-off
cate that its FRD would be not too dissimilar to that
thatareweaklyguidingandlightlytocoupletocladding
of a standard silica fibre. As an absolute reference, the
modes.Whenfibresarebroughtclosetogether,asinthe
maximum NA of both polymer imaging bundles was
case of imaging bundles, the scattered light that leaves
measured by overfilling the bundle’s input NA (in the
the core has a chance of coupling into its neighbour-
R-band), and radially measuring the far-field intensity
ing cores. If the propagating modes are poorly confined
using a power meter with a 500µm pinhole mask at a
duetoasmalldifferenceinrefractiveindexbetweenthe
radial distance of 20cm. This size of pinhole mask was
core and the cladding, the propagating light has a high
chosen to be large enough for the power meter to suf-
chance of evanescently coupling to neighbouring cores.
ficiently detect the incident light, yet small enough to
Thesecouplingtermsareknownascross-talk,andmin-
providediscretemeasurementsatthevariousanglesand
imising this property is important for imaging bundles
shieldthepowermeterfromscatteredlightinthelabo-
as to not spread out the spatial information of the in-
ratory. The maximum NA (at 5% maximum intensity)
putlightsource.Thecouplingofmodestohigherorder
of the 0.5mm and 1.5mm polymer imaging bundle is
modes that were not coupled to at the input as well as
0.46±0.03and0.49±0.03respectively,withtheerrors
the lading of modes to the cladding is known as focal
PASA(2017)
doi:10.1017/pas.2017.xxx
6 Richards et al.
Figure 6.: Three 1.5mm polymer imaging bundles used as guiding probes for SAMI. (left) sky-flats of the three
bundles imaged through a 1.5:1 re-imaging lens to a an Apogee Alta U6 CCD. Dust in the imaging system gives
theappearanceofdamagedcores,howeverthelowerbundleisdamagedonitsrightside.(middle)threeguidestars
being probed by the polymer imaging bundles. Typical V-band magnitudes and exposures times for SAMI guide
stars are 13-15 and 1s respectively. (right) image of Saturn and its rings through one of the bundles (∼2 arcsec
seeing). Each imaging bundle has a 23 arcsec field-of-view. The outline of each bundle is traced by a black circle.
All three images are inverted.
representing the range of NAs found at intensity values 3 Applications of polymer imaging bundles
eithersideofthe5%maximumintensity.Therefore,due in astronomy
to the small overall diameter of these imaging bundles,
In this section we present current and possible (but
designing an optical system to image the face of the
by no means exhaustive) applications of polymer imag-
bundle, or even 100 bundles packed together, with f/1
ing bundles in astronomical instrumentation that use
optics (NA = 0.5) is straightforward.
lengths<2m.Forthisweconsiderthetrade-offsofthe
To measure the cross-talk in the 1.5mm polymer
increase in filling factor at the cost of an increase in
imagingbundle,asilicasingle-modefibre(NA=0.1,in-
attenuation. Fundamentally, polymer imaging bundles
jectedwithR-bandlight)wasbutt-coupledandaligned
translate images, which in the context of astronomy is
to the centre of a single core, and the intensity of all of
best suited to getting an image from the focal plane of
thesurroundingcoreswasimaged(seeFigure5).Itwas
a telescope to a nearby location where bulkier systems
foundthat2.9±0.2%oftheinjectedlightwasscattered
reside (cameras etc). The need of imaging bundles in
into the surrounding cores. This is the mean measure-
astronomy increases as the diameter of the telescope’s
ment and standard deviation of five different locations
focal plane increases, more so in the era of the ELTs.
overthefaceoftheimagingbundle(withtheexception
of the outermost cores).
Due to the high-NA, i.e. large mode confinement, of
the polymer imaging bundles, it is unlikely that the
3.1 Field acquisition and guiding
modes’ evanescent fields are significant to allow effi-
cient coupling between cores, meaning that the pri- The most obvious application of imaging bundles is for
mary driver for the cross-talk is more likely to be the telescope guiding as part of a multi-object instrument,
large scattering term of the polymer’s chemical struc- orwheretheweightofaguidecameraonthefocalplane
ture. This is due to the larger molecular chains and is an issue (e.g. small telescopes). It is possible to use
clusters of the polymer in addition to the molecular pick-offmirrorsonarmstodomulti-objectguiding,but
absorptions of their vibrational states. Unlike a stan- ifthereareopticalsciencefibrescoveringthefieldplate
dard silica fibre when unspooled, the polymer imaging then it is difficult to avoid collisions. Imaging bundles
bundle retains a relaxed bend radius of ∼15cm. It is permit a flexible solution to translating a spatially co-
possible to straighten the fibre via thermal annealing, herent image of a guide star from the focal plane to
although doing so risks damaging the fibre (e.g. chang- a nearby camera. An advantage of using polymer over
ing the polymer’s chemical / physical structure). silica for multi-object imaging bundles (other than the
increaseinfillingfactor)isthattheendpreparationofa
polymer imaging bundle is much more customisable, as
theproductresemblesasinglefibre.Silicaimagingbun-
PASA(2017)
doi:10.1017/pas.2017.xxx
Performance of a novel polymer imaging bundle 7
dleshavetobecarefullyconnectorised(normallybythe thatevenwithoutfore-optics,multiple1.5mmpolymer
manufacturer) to minimise stress or fractures, whereas imagingbundlesareabletoacquireandguideonafield.
the polymer imaging bundles can either be mounted in If an increased field-of-view per imaging bundle is de-
custom connectors, or just left bare whilst maintaining sired, then the use of fore-optics (miniature lenses or
the ability to be polished. fibre tapers) is straightforward.
SAMI is a multi-object integral field spectrograph
that deploys 13 silica fibre-bundle integral field units
3.2 Multi-object Wavefront Sensing
(hexabundles) over a 1 degree field at the 3.9m Anglo-
Australian Telescope (Bryant et al. 2015). For field ac- The use of adaptive optics (AO; Babcock 1953; Hardy
quisition and guiding it uses three of the 1.5mm poly- 1998) to correct for atmospheric turbulence is an es-
merimagingbundles(1mlengthsandeachwith23arc- sential part of the science cases proposed for ELTs, but
sec field-of-view) mounted in the same plug-plate con- is also a technology that is currently available on tele-
nector design as the hexabundles. All of the imaging scopes all over the world. Performing AO on a single
bundlesfeedasingleguidecameralocatednearbytothe on-axis target is now standard practice for ≥8m class
sideofthefocalplane(seeFigure6).Althoughonlyone telescopes, though recent studies into multi-object AO
oftheguidestarsisusedfortelescopeguidinginSAMI, (MOAO;Hammeretal.2002;Roussetetal.2010;Vidal
having multiple guide stars over a 1 degree field allows etal.2014)showthepowerofmaximisingthescientific
forbetterunderstandofthewide-fieldatmosphericcon- output of large aperture wide-field telescopes, such as
ditions,andthetip/tiltdeterminationofthefieldplate. the 25.4m Giant Magellan Telescope (GMT; Bernstein
The light-weight and flexible nature of the polymer et al. 2014). For MOAO to be realised on telescopes
imaging bundles makes them suited to many types of such as the GMT, many wavefront sensors would need
fibrepositioningtechnologies.Inadditiontouseinplug- to be deployed across the focal plane, localised to each
plate instruments, the 1.5mm polymer imaging bundle science target.
has also been proven to be compatible with the Star- The use of polymer imaging bundles becomes an at-
Bugs positioning system (Gilbert et al. 2012; Kuehn tractivesolutiontodistributedwavefrontsensors,where
et al. 2014), and multiple 0.5mm polymer imaging eachbundlehasminiatureShack-Hartmannfore-optics,
bundles will perform the guiding for the multi-object relaying the wavefront information to a high frame
fibre spectrograph instrument, TAIPAN on the UK rate camera situated off the focal plane. These minia-
Schmidt Telescope (Kuehn et al. 2014, see Figure 7). tureShack-Hartmannwavefrontsensors(mini-SHWFS;
Even though there is a loss in efficiency when using the Goodwin et al. 2014) were demonstrated on the 3.9m
0.5mm polymer imaging bundle instead of the 1.5mm AAT in December 2015 (see Figure 8; a full descrip-
polymer imaging bundle, the smaller core sizes of the tion of these devices is provided by Goodwin et al. in
0.5mmpolymerimagingbundlearebettersuitedtothe preparation).
plate scale of the UKST (0.44 to 1.08 arcsec per core The 1.5mm polymer imaging bundle, with its 16µm
respectively, with the latter being too coarse to accu- cores and high filling factor, is able to critically sample
rately sample the guide star’s point-spread-function). the Shack-Hartmann spots of the wavefront sensor. A
The 0.5mm polymer imaging bundle is also being con- laboratory closed-loop adaptive optics experiment was
sidered to replace the current guide bundles on the 2df carried out to quantify the suitability of the polymer
instrument on the 3.9m Anglo-Australian Telescope, imagingbundleinwavefrontsensing(seeFigure8).The
which uses a pick-and-place positioning system with fi- set-up for the experiment was as follows: (1) Collimate
bre retractors (Lewis et al. 2002; Sharp et al. 2006). thebeamofa635nmlasertoabeamwidthof12.5mm.
Investigation into the stresses induced on the 0.5mm (2) send the collimated laser beam through a rotatable
polymerimagingbundleanditslongevitywhenusedin phase-screenwitheffectiveseeingof0.78arcsec.(3)im-
sprung retractors is underway. age the pupil of the collimated beam onto an ALPAO
As focal planes become increasingly larger, particu- 97-15 Deformable Mirror (DM). (4) re-image the pupil
larlyintheeraofELTswherefocalplaneswillbe1-2m planethrougha100µm-pitchmicrolensarray,resulting
diameter, the need of fibre bundles to perform guid- in the 1.5mm polymer imaging bundle (1m length) re-
ing also increases. This is where the adaptability of the ceiving a 14×14 Shack-Hartmann spot pattern, which
polymer imaging bundles to different positioning tech- contains the wavefront information at the pupil plane.
nologies has an advantage, in addition to the large in- (5) re-image the end-face of the 1.5mm polymer imag-
creaseinfillingfactor.Typicalplate-scalesfortheELTs ing bundle with 2× magnification onto an Andor Zy-
are a few arcsec/mm, meaning that the field-of-view lar 5.5 MP sCMOS detector. (6) Control the rotation
of a 1.5mm polymer imaging bundle would also be a speed of the phase-screen, the SCMOS detector, and
few arcsec. Blind pointing of modern large telescopes is the DM via "S/W ALPAO Core Engine (ACE)" soft-
commonly ∼1 arcsec (Bernstein et al. 2014), meaning ware using MATLAB/C++. The software was cus-
tomised to fit this experiment, which was able to run
PASA(2017)
doi:10.1017/pas.2017.xxx
8 Richards et al.
at ∼250 frames per second in closed-loop mode. The
latency in the experiment was found to typically be 2-
3 frames. Optically, the set-up was designed to work
in f/10, and to mimic the University of Hawaii’s 2.2m
telescope.
From Figure 8 (bottom) it is possible to see that
the 1.5mm polymer imaging bundle can indeed criti-
cally sample the Shack-Hartmann spot pattern as part
of a closed-loop adaptive optics instrument, enabling
the the Shack-Hartmann spot pattern to be relayed to
a nearby, yet arbitrarily located, sensor. Only at wind-
speeds greater than 10 ms−1 does the wavefront im-
provement become negligible, with the RMS approach-
ing that of an uncorrected wavefront (open-loop). Cus-
tom designs of polymer imaging bundles can be man-
ufactured depending on system requirements, however
this is somewhat limited without the ability to self-
manufactureasaccesstoproprietypolymersandequip-
ment is restricted.
In addition to being used for high-order wavefront
sensing, the polymer imaging bundles are also suited Figure7.:Microscopeimageofa0.5mmdiameterpoly-
to distributed low-order wavefront sensing for use in mer imaging bundle mounted in a TAIPAN StarBug
ground-layerAO(GLAO;Hartetal.2010)orthemon- with a 3D-printed ferrule.
itoring of natural guide stars in laser-tomography AO
(LTAO; Viard et al. 2002), which would also benefit
any attempt of laser-tomography MOAO with either
Rayleigh or Sodium laser guide stars.
An alternative method in wavefront sensing is the
curvature sensor (Roddier 1988; Roddier & Roddier
1993),wheretheintensityimageoftheintra-andextra-
focus are simultaneously obtained, allowing the second
derivativeofthewavefronttobemeasured.TheVisible
and Infrared Survey Telescope for Astronomy (VISTA)
builtintoitsdesigntheabilitytoobtaintheintra-focus
of one star and the extra-focus of another (but nearby)
star with two respectively independent CCDs (Patter-
son & Sutherland 2003). The assumption is that differ-
ential aberrations between the two stars are negligible.
However,mostAOsystemsthatarebasedoncurvature Figure8.:(top)RaytraceofminiatureShack-Hartmann
wavefront sensors (Rigaut et al. 1998; Arsenault et al. sensor, where A is the starlight from the telescope (f/8
2003; Watanabe et al. 2008) use a resonant vibrating at AAT cassegrain focus), B is a 3mm diameter colli-
mirror with sinusoidal piston variations to obtain the matinglens,Cisa110µmpitchmircolensarray,andD
intra- and extra-focuses. This is a very delicate proce- is the polished front face of a 1.5mm polymer imaging
dure, and limits the adaptability to MOAO. Based off bundle. (left) Image of a prototype miniature Shack-
the design proposed by Guyon (2010), another method Hartman sensor illuminated with a 630nm laser at f/8
istousetwoimagingbundlesforeachcurvaturesensor, (top). Dust inside the wavefront sensor blocks ∼2 of
with a beam splitter projecting the intra- and extra- the lenslets. (right) On-sky demonstration of the same
focus images of a star onto each bundle. The imaging wavefront sensor as shown in (left) translating the spot
bundles then feed into a sensitive high frame-rate cam- pattern from the focal plane to a high frame rate cam-
era off the focal plane, and the intensity images are era (sCMOS). This stacked image is equivalent to a 1s
analysed to obtain the wavefront information. A full exposure.Thede-centringofthepupilisduetoasmall
description of these devices and recent on-sky tests are tip/tiltintheopticalalignmentofthewavefrontsensor
given by Zheng et al. (2014, in preparation). to the focal plane. The wavefront sensor’s field-of-view
(sub-aperture spacing) is ∼2 arcsec. Both images have
beeninvertedincolour,withacircledrawntoshowthe
boundary of the imaging bundle.
PASA(2017)
doi:10.1017/pas.2017.xxx
Performance of a novel polymer imaging bundle 9
1.0
0 m/s 3 m/s 7 m/s no phase screen
1 m/s 5 m/s 10 m/s
0.8
)
s
n
0.6
o
r
c
i
m
(

S 0.4
M
R
0.2
0.0
0 5 10 15 20 25 30 35 40
seconds
Figure 9.: Laboratory closed-loop adaptive optics data using a 1.5mm polymer imaging bundle as part of a mini-
Shack-Hartmann wavefront sensor (mini-SHWFS, see text for setup description). (top left) The 14×14 Shack-
Hartmannspotpatternrelayedbythe1.5mmpolymerimagingbundle.Thecentre-of-massofeachspotisequated
in relation to the centre of each zonal box, with the offsets represented by red arrows. A mask is applied to skip
zones/spots that have too little flux. Dust inside the mini-Shack-Hartmann sensor results in some zones to the left
of the bundle being skipped. (top right) The flux map showing the masked out zones as black zones. (bottom)
Closed-loop results for six different simulated wind speeds of the 0.78 arcsec phase screen, in addition to no phase
screen.TheRMSisameasurementoftheerrorcomparedtothatofaflatwavefront(bestimagequality).Thesmall
RMSdifferencebetweenthe0ms−1 andnophasescreenisaverificationthatthemini-SHWFSissuccessful.There
is little to no wavefront improvement for wind speeds above 10 ms−1, with the RMS increasing to the uncorrected
wavefront due to system latency (2-3 frames).
PASA(2017)
doi:10.1017/pas.2017.xxx
10 Richards et al.
3.3 Other applications 4 Conclusions
In addition to being used for field acquisition, guid- Imagingbundlesareausefulwayoftranslatingspatially
ingandwavefrontsensing,polymerimagingbundlesare coherentimages,andovershortdistances,imagingbun-
also suited to a variety of other astronomical applica- dlesmadefrompolymercanbemoreefficientandprac-
tionsthatrequiretheefficientandflexibletranslationof tical than than those made from silica. We performed
images over short distances. One such application is in laboratory characterisation of ESKA PMMA imaging
multi-object narrowband imaging, where each imaging bundles, and explored their various applications in the
bundle is positioned over a target and has a miniature context of astronomical instrumentation. These cheap
tuneable narrowband filter that is tuned for its target polymer imaging bundles (∼$10 per metre) remain
(e.g. a galaxy’s redshift). The ability to obtain high- flexibleasasingle1.5mmdiameterunitwith7095cores
spatial resolution narrowband images (e.g. Hα) of 105 providing a spatial filling factor of 92%. Their R-band
independent galaxies over a survey lifetime can provide transmissionwasmeasuredtobe∼85%permetre,and
adatasetthatisunparalleledinunderstandingthestar theirflat-fieldresponseyieldsaradialdecreaseof∼3%
formation morphology of a galaxy as a function of its towardstheoutercores.WithamaximalNAof0.5,the
environment. An issue if using the polymer bundle pre- polymer imaging bundles are well suited to a range of
sented here is the spectral absorption features of the telescope focal ratios. As most applications will image
PMMA.Thereisalargeabsorptionfeatureat∼740nm theend-faceoftheimagingbundleswithacamera,any
(seeFigure4),whichwouldbedetrimentaltoHαobser- FRD can be accounted for in the camera design.
vationsofgalaxiesatredshifts,0.06(cid:46)z (cid:46)0.2.Depend- The polymer imaging bundles characterised here are
ingonthescienceinquestion,differentpolymerscanbe already in use for field acquisition and guiding as part
usedtominimisespectralfeatures,whilststillretaining oftheastronomicalinstrument,SAMI,andinthenear-
other benefits such as high fill-factors. Argyros (2013) future, TAIPAN. This application is the most obvious
provides a review of the common polymers used in the useofspatiallycoherentimagingbundles,thoughother
manufacture of fibre optics, with CYTOP having the demonstrated applications include multi-object minia-
best spectral transmission (high and flat; Figure 4) in ture wavefront sensing. Polymer imaging bundles are
addition to operating at near-infrared wavelengths. highly customisable in design (including which poly-
It is also possible to envisage an instrument that mers to use depending on desired wavelength ranges)
hasdistributedminiatureaperturemasks(Tuthilletal. making them well suited to a range of astronomical ap-
2000)overthefocalplane,withtheintensityofeachin- plications, such as multi-object narrow-band imaging,
terference pattern projected onto the polymer imaging multi-objectaperturemaskingandmulti-objectspeckle
bundles, all feeding a sensitive high frame-rate camera imaging. There are of course many other applications
off the focal plane. This application would work much not yet explored, and the use of polymer imaging bun-
better if the polymer used in the bundles is one that is dlesinfutureastronomicalinstrumentshaspotentialto
efficient at wavelengths in the near/mid-infrared, such become routine.
that extrasolar planet detection by imaging is feasi-
ble (Burrows 2005; Ireland 2012). Multi-object speckle
imaging (Saha 2002) is also feasible with distributed
polymer imaging bundles without fore-optics, again 5 Acknowledgments
withallbundlesfeedingasensitivehighframe-ratecam-
We would like to thank Chris Betters, Simon Gross,
era off the focal plane. This would be useful in a fast
Alexander Arriola, David Brown, and the staff at the
survey to understand the fraction of multiple star sys-
Anglo-Australian Telescope for their help in the char-
tems (Duchêne & Kraus 2013).
acterisation and commissioning of the different instru-
Over time, an increasing number of telescope instru-
ments that make use of the polymer imaging bundles.
ments are achieving multi-object functionality as they
Most of the characterisation work was performed at
aim to maximise the available science over the focal
theSydneyAstrophotonicInstrumentationLaboratory,
plane, so it is only natural to keep thinking of feasible
fundedbyanARCLaureateFellowshipawardedtoJoss
ways to make use of the power and large focal plane
Bland-Hawthorn.
of the ELTs to come. There are likely many more ap-
Thisresearchwasalsoconductedinpartnershipwith
plications of polymer imaging bundles for astronomical
the Australian Research Council Centre of Excellence
applications, and we leave the reader to imagine.
for All-sky Astrophysics (CAASTRO), funded under
the Australian Research Council (ARC) Centre of Ex-
cellence program (project number CE110001020), with
additional funding from the seven participating univer-
sities and from the NSW State Government’s Science
Leveraging Fund.
PASA(2017)
doi:10.1017/pas.2017.xxx