A high-finesse Fabry-Perot cavity with a frequency-doubled green laser for precision Compton polarimetry at Jefferson Lab

5.5 MB·

by  A. Rakhman

A high-finesse Fabry-Perot cavity with a frequency-doubled green laser for
precision Compton polarimetry at Jefferson Lab
A. Rakhmana,b,∗, M. Hafezc, S. Nandad, F. Benmokhtare,f, A. Camsonned, G.D. Catesg, M.M. Daltond,g,
G.B. Frankline, M. Friende,h, R.W. Michaelsd, V. Nelyubing, D.S. Parnoe,i, K.D. Paschkeg, B.P. Quinne,
P.A. Soudera, W.A. Tobiasg
aSyracuse University, Department of Physics, Syracuse, NY 13244, USA
bResearch Accelerator Division, Spallation Neutron Source, Oak Ridge National Laboratory, Oak Ridge, TN 37831, USA
cOld Dominion University, Applied Research Center, Norfolk, VA 23529, USA
6
dThomas Jefferson National Accelerator Facility, Newport News, VA 23606, USA
1
eCarnegie Mellon University, Department of Physics, Pittsburgh, PA 15213, USA
0
fDuquesne University, Pittsburgh, PA 15282, USA
2
gUniversity of Virginia, Department of Physics, Charlottesville, VA 22904, USA

r hHigh Energy Accelerator Research Organization (KEK), Tsukuba, Ibaraki, Japan
a iUniversity of Washington, Center for Experimental Nuclear Physics and Astrophysics and Department of Physics, Seattle,
M WA 98195, USA

9
2

]
t Abstract
e
d A high-finesse Fabry-Perot cavity with a frequency-doubled continuous wave green laser (532 nm) has
-
s been built and installed in Hall A of Jefferson Lab for high precision Compton polarimetry. The infrared
n (1064nm)beamfromaytterbium-dopedfiberamplifierseededbyaNd:YAGnonplanarringoscillatorlaser
i
. isfrequencydoubledinasingle-passperiodicallypoledMgO:LiNbO3 crystal. Themaximumachievedgreen
s
powerat5Winfraredpumppoweris1.74Wwithatotalconversionefficiencyof34.8%. Thegreenbeamis
c
i injected into the optical resonant cavity and enhanced up to 3.7 kW with a corresponding enhancement of
s
y 3800. The polarization transfer function has been measured in order to determine the intra-cavity circular
h laser polarization within a measurement uncertainty of 0.7%. The PREx experiment at Jefferson Lab used
p this system forthe first timeand achieved1.0% precision inpolarization measurements of anelectron beam
[
with energy and current of 1.06 GeV and 50 µA.

2
v
Keywords: Compton polarimetry, Fabry-Perot cavity, Laser Polarization, Polarized electron beam
1
5
2
0 1. Introduction [5] are destructive for the beam properties due to
0 solid targets they use and cannot be operated si-
. The Continuous Electron Beam Accelerator Fa-
1 multaneously with the physics experiments. They
cility (CEBAF) at Jefferson Lab (JLab) delivers a
0 can only be operated at low intensity (< 10 µA)
6 highly polarized electron beam in the 1 – 11 GeV
and at low energy (< 5 MeV) respectively; there-
1 energy range for currents up to ∼ 200 µA. A key
fore, the physics experiments have to assume that
v: part of the JLab physics program consists of high- beam polarization remains constant when the in-
i precisionparityviolatingelectronscatteringexper-
X tensity or the energy of the beam is several or-
iments[1–3]thatrequireafastandpreciseelectron
ders of magnitude higher. Alternatively Comp-
r
beam polarization measurement. Conventional po-
a ton polarimetry has been used with great success
larimetry techniques such as Møller [4] and Mott
at JLab [6, 7] and elsewhere [8–12] to measure
the electron beam polarization in a continuous and
∗Corresponding author: Research Accelerator Division, non-destructive manner. However, unlike at high-
SpallationNeutronSource,OakRidgeNationalLaboratory,
current (> 20 mA) storage rings [8, 9, 12] and at
Oak Ridge, TN 37831, USA. Tel.: [phone]; Fax.:
[phone]. high-energy (> 50 GeV) colliders [10, 11], the rela-
Email address: [email protected](A.Rakhman)
Preprint submitted to Nuc. Instr. and Meth. A March 30, 2016
tivelylow-currentandlow-energynatureoftheCE- had serious limitations in providing a good signal-
BAF electron beam has required the use of high- to-noise ratio for beam energies below 2 GeV [23].
finesse optical cavities [13, 14] to enhance the laser Under this circumstance, it is highly desirable to
power beyond commercially available lasers to pro- have a shorter-wavelength and higher-power light
vide sufficient electron-photon collision luminosity. sourcesincethiswillresultinlargerback-scattered
Compton polarimetry determines the electron electronandphotonenergiesandalargermeasured
beam polarization based on a well-known spin- scattering asymmetry, and hence increased ability
dependentscatteringcross-section[15]betweenpo- to control systematic uncertainties.
larized electrons and photons. The longitudinal TheLeadRadiusExperiment(PREx)[1]atJLab
electronbeampolarizationP isextractedfromthe aims to provide the first model-independent evi-
e
experimentalComptonscatteringasymmetryA , dence of the existence of a significant neutron skin
exp
according to, in 208Pb. PREx proposed to obtain a statistical
andsystematicprecisionsof3.0%and2.0%respec-
A =P P A , (1)
exp e γ th tively on the parity-violating electroweak asymme-
where A is the theoretical Compton scattering try of polarized electrons from 208Pb to determine
th
asymmetry that can be derived by combining the the neutron radius of 208Pb at 1.0% accuracy. The
calculated spin-dependent Compton cross section uncertainty goal for beam polarimetry at 1.06 GeV
with the appropriate experimental response func- was 1.0%. In order to meet the high-precision re-
tion [16], and P is the circular photon beam po- quirement, the existing Compton polarimeter un-
γ
larization. The scattering asymmetry A can be derwent an upgrade [24, 25] in 2010. The up-
exp
measured based on the detection of either the scat- grade includes a new photon detector with an up-
tered photons or the scattered electrons and de- graded integrating data acquisition (DAQ) system,
pends on the energies of the colliding electron and a new electron detector, and a new Compton pho-
photon beams. For an electron beam of 1.0 GeV, tonsourcethatincludesagreenlaser(532nm)with
the maximum scattering asymmetry for infrared a new high-finesse Fabry-Perot cavity [26].
(1064 nm) photons is 1.8%, whereas for green Thispaperwilldescribetheupgradedexperimen-
(532nm)photons,itis3.5%. Foranelectronbeam tal setup in Hall A at JLab, with an emphasis on
of 6 GeV, these values are about 10% and 18%, re- the frequency-doubled green laser system and opti-
spectively. Therefore, Compton polarimetry at low cal cavity. A unique technique to extract the laser
beam energies (< 2 GeV) is difficult to perform. beam polarization inside a resonant Fabry-Perot
External optical cavities have frequently been cavity will be given and finally, the first electron
employed to increase the laser power from a light beam polarization measurement results from the
source through a process of coherent addition. Re- PREx experiment [1] at a beam energy and cur-
centlyopticalcavitieshavebeenwidelyemployedin rent of 1.06 GeV and 50 µA will also be presented.
manyareassuchasX-/γ-raygenerationviainverse- Withanenhancementfactorandintra-cavitypower
Compton scattering [17, 18], high harmonic gener- of 3,800 and 3.7 kW, we report the highest power
ation [19] and electron beam polarimetry [14, 20]. Compton polarimetry facility operating at 532 nm
According to the best of our knowledge, there are ascomparedtoexistingfacilitiessuchasMAMI[21]
only three facilities in the world have ever built a and Hall C at JLab [7].
Comptonpolarimeterwithanopticalcavity;Mainz
Microtron(MAMI)[21],HadronElectronRingAc-
celerator (HERA) [20] and JLab [14]. In Hall A of 2. Overview of upgraded Compton po-
JLab, a Compton back-scattering polarimeter had larimeter
been operational since 1999 [22]. Its photon source
hadatypicalcirculatingpowerof∼1.5kWinsidea The upgraded Compton polarimeter in Hall A
monolithic Fabry-Perot cavity where the circularly consistsofamagneticchicane,aphotonsourcethat
polarizedphotonsscatteredfromthelongitudinally iscomposedofalasersystemandFabry-Perotcav-
polarized electrons. It provided 1.0% statistical ity, a photon detector, and an electron detector as
(and 1.2% systematic) precision with ∼ 25 min- shown in Fig. 1. The electron beam enters from
utes’datatakingfor4.6GeV,40µAelectrons[22]. the left. When the four identical dipoles of the chi-
This polarimeter used a continuous-wave (CW) in- cane, referred to as D , D , D and D , are ener-
1 2 3 4
frared(IR,1064nm)laserasitsphotonsourceand gized, the beam is deflected vertically and crosses
2
the electron beam polarization. A scale 3D view of
D Magnetic Chicane Electron Detector D
1 4
the Compton polarimeter in the Hall A tunnel at
JLab is shown in Fig. 2.
Electron Beam Fabry-Perot Cavity Photon Detector The photon detector is located under dipole D4
D2 D3 GSO andconsistofasinglecrystalofCe-dopedGd2SiO5
(GSO) and a single PMT (Photomultiplier Tube).
Fig. 1: A not-to-scale schematic of the upgraded Compton It is mounted on a motorized table with remote-
polarimeterinHallAatJLab. Amagneticchicaneconsists controllablemotionalongbothaxes(horizontaland
offouridenticalmagneticdipoles(D1,D2,D3 andD4)that vertical) transverse to the electron beam direction.
deflect the primary electron beam and send it through the
Fabry-Perot cavity. The scattered Compton electrons and The scattered photons from the CIP go through
photons will be detected by electron and photon detectors, a thin vacuum window and lead disk in the vac-
respectively. uum pipe before they hit the detector window.
The photon detector DAQ uses a customized 12-
bitFlashAnalog-to-DigitalConverter(FADC)that
thephotonbeamenhancedintheFabry-Perotcav-
integrates the scintillation photon signals from the
itythatislocatedatthecenterofthechicane. The
PMTandsamplesthedataatratesof200mW[24].
Fabry-Perot cavity is enclosed in a vacuum cham-
The electron detector consists of four silicon mi-
ber that is connected to electron beam pipes from
crostrip planes spaced horizontally 1 cm from each
both ends and sits on an optical table where the
other with a 200 micron vertical rising offset be-
laser and other optical elements are located. The
tween planes. Each plane consists of 192 strips of
ComptonInteractionPoint(CIP)isatthecenterof
siliconwithawidthof240microns. Theplanesare
thebeamcrossinginthecavity. Thecrossingangle
inclined at an angle of 58 mrad from the vertical
betweentheelectronandphotonbeamsis24mrad.
position. The detector is mounted on a translator
TheelectronsundergoComptonscatteringwithcir-
with a remote-controllable stepper motor and can
cularly polarized photons resonating in the Fabry-
travelverticallyupto120mmfromthemainbeam.
PerotcavityfedbyaCWgreenlaser(532nm). The
backscattered photons are detected in the photon
detector while the scattered electrons can be de- 3. Optical and mechanical system
tected in the electron detector located a few mm
above the primary beam in front of D . Approxi- The optical elements of the polarimeter are
4
mately one electron in every one billion undergoes mounted on a vibration-damped optical table be-
Compton scattering. Unscattered electrons, sepa- tween dipoles D and D (Fig. 2). A schematic of
2 3
rated from the scattered particles by D in the chi- the optical system of the polarimeter is illustrated
3
cane,continueonintotheHallAtargetforthepri- in Fig. 3. Based on their main functionality, we
mary experiment. Independent electron and pho- cancategorizetheopticalelementsintofourgroups
tonanalysesgivecomplementarydeterminationsof necessary for achieving the required power and cir-
Electron Detector
Supports
Scattered
D
Electrons 4
Optical Cavity
Direct Beam to Hall A
D3 Scattered
Photons
D
2
Photon Detector
D
1 Deflected Beam
Optical Table Shielding
Fig.2: Ascale3DviewoftheComptonpolarimeterintheHallAtunnelatJLab.
3
cular polarization in the cavity. The first group periodically poled MgO:LiNbO (PPLN, HC Pho-
3
is the laser source that provides a green beam at tonics) doped with 5% MgO in order to minimize
the wavelength of 532 nm. The second group con- photo-refractive damage. The crystal is 0.5 mm
sists of the elements that transport, align and fo- thick, 3 mm wide and 50 mm long, and the quasi-
cus the incident beam for coupling to the optical phase matching (QPM) period is 6.92 µm with a
cavity. The third group includes elements for con- 50% duty cycle. The input and output surfaces are
trolling and measuring the polarization. The last antireflection(AR)coatedfor1064nmand532nm,
group consists of elements which allow the use of respectively. The crystal is placed in an externally
thereflectedbeamfromthecavityintheelectronic controlled, temperature-stabilized oven developed
feedbacksystemtoachievefrequencylockingofthe in-house. Theovenismountedonastagecomposed
laser to the cavity. ofafour-axistiltaligner(9071-V;Newport)inorder
toestablishaprecisealignmentforphasematching.
3.1. Frequency doubling Atemperaturecontroller(TECSource5305;Arroyo
Instruments) with a nominal resolution of 0.01 ◦C
As shown in Fig. 3, the seed laser is a narrow-
provides the required phase-matching temperature
linewidth (< 5.0 kHz) diode pumped Nd:YAG
for second harmonic generation (SHG). The crys-
nonplanar ring oscillator (NPRO) laser (Light-
talopticalaxisismatchedwiththedirectionofthe
wave 126; JDSU) that delivers a linearly polar-
output polarization (vertical) of the YDFA. A pair
ized CW IR (1064 nm) beam of up to 250 mW.
of 0.5 inch lenses (L and L in Fig. 3) with fo-
Theoutputofthislaserhasbeenfiber-coupledtoa a b
cal lengths of 13.8 mm and 15 mm (with a total
ytterbium-dopedfiberamplifier(YDFA,YAR-10K-
effective focal length of ∼ 120 mm) are then used
1064-LP-SF;IPGPhotonics)throughasingle-mode
to focus the beam waist diameter (∼ 80 µm) into
polarization-maintaining (PM) fiber. The YDFA
the center of the crystal. This makes the Rayleigh
takes ∼ 10 mW of seed laser and provides a lin-
length of the pump laser focus to be around 5 mm
early polarized (extinction ratio: 20 dB) output
and ensures the beam size at the crystal entry and
with a maximum power of 10 W at 1064 nm in
exitfacestobearound400µm. Thiscriticalfocus-
CW mode. The frequency doubling crystal is bulk
CCD
1
PeriscoMpe2 Mr2 M1
L3
We Mce Periscope
Me Ms
CavitylVacuumlChamber Mcs Ws M3
QWP2
NPROlLaser L2 transmissionlsignal PDT HBS CCD2
tollockinglelectronics
QWP1
SinglelMode
PMlFiber Mr3 W Mr4
o
PPLN DC1BD PBS PDR lroecflkeicntgiolenllescigtrnoanlilctos llasto
n
YDFA La Lb Oven
DC2 L0 S1 S2
FOI HWP L1 Mr1
Fig. 3: The optical system consists of a NPRO seed laser, a fiber amplifier, a frequency doubling setup and a Fabry-Perot
cavity enclosed in a vacuum chamber. (YDFA: ytterbium doped fiber amplifier; PPLN: periodically poled lithium niobate;
DC1, DC2: dichroic mirrors; BD: beam dump; FOI: Faraday optical isolator; HWP: half-wave plate; PBS: polarizing beam
splitter;QWP:quarter-waveplate;HBS:holographicbeamsampler;Mce,Mcs: concavecavitymirrors;CCD:cameras;PDR,
PDT:photodiodes;L:lenses;M:planemirrors;W:vacuumwindows;S:photodetectors)
4
a.u.) 1 Theory W) 2 64.4C)
Power ( 0.8 Graph Measured ower ( 11..68 ShTe H(m% G p / WP e or- ac w tm ue )rr e= 1 . 3 7 – 0.02 64.3oature (
HG Output 0.6 D T = 0.6 oC m Output P 11..241 64.2Temper
S 0.4 2 n 0.8 64.1
3
5 0.6
0.2
0.4 64
0.2
0
62 62.5 63 63.5 64 64.5 65 65.5 66 0 63.9
Temperature (oC) 0 1 2 3 4 5
1064 nm Input Power (W)
Fig. 4: Experimental (red circles) and theoretical [27, 28]
Fig. 5: Average green power (solid circles) output and cor-
(blue line) temperature tuning curve for PPLN crystal at
responding phase matching temperature (open squares) at
beamwaistdiameterof80µmandIRpumppowerof5W.
different IR pump power from the fiber amplifier. A theo-
retical fit (green line) defined by Eq. (2) is used to extract
thenormalizedSHGconversionefficiency.
ing also ensures the incoming beam is not clipped
bythetinycrystalthicknesswhilekeepingtheSHG
peak phase-matching temperature to ensure maxi-
efficiency close to the maximum. The generated
mum green power. For a loosely focused Gaussian
greenbeamisseparatedfromtheresidualIRbeam
beam, the normalized conversion efficiency η for
after the crystal via a pair of 1.0 inch dichroic mir- nor
first-order QPM is defined as [29],
rors (R > 99% at 532 nm, T > 95% at 1064 nm)
noted as DC and DC (Altos Photonics). The
1 2 I =η LI2, (2)
lenses, oven and dichroic mirrors are all mounted 2 nor,QPM 1
on a separate linear translation stage seated on a
8ω2d2L
rail, andthewholesystemiscontainedinanenclo- η = 1 , (3)
nor,QPM n2n c3ε π2
sure box. 1 2 0
Withtheexperimentalsetupasdescribedabove, whereI andI arethefundamentalandSHGpow-
1 2
wehavemeasuredseveralproperties,suchascrystal ers, ω is the frequency of the fundamental beam,
1
temperaturetuningcurve, beamqualityandpower d is the nonlinear coefficient of PPLN, ε is the
0
stability, of the second harmonic beam. The mea- permittivity of free space, and n and n are the
1 2
sured temperature tuning curve at 5 W IR pump refractive indexes at these wavelengths. In Fig. 5,
power is shown in Fig. 4. The red circles show the asthetheoreticalfit(greenline)definedbyEq. (2)
experimentaldatawhilethebluelinerepresentsthe shows, the green power varies quadratically with
theoretical fit for the first-order QPM interaction the IR pump power with a normalized conversion
predictedbytheSellmeierequationforPPLNcrys- efficiencyη of1.37%W−1cm−1. Itisnecessaryto
nor
tal [27, 28]. The experimental results give the full point out that all powers are direct measurements
width at half-maximum (FWHM) phase-matching afterDC byanopticalpowermeter(PM140;Thor-
2
temperaturebandwidth∆T =0.6◦Catthephase- labs) without correcting for the residual reflection
matching temperature 64.0 ◦C. at the crystal face or losses in the dichroic mirrors.
Fig. 5 shows the average green power (solid cir- Theexperimentalresultsshowthatthereisnosign
cles) output from PPLN and corresponding phase- of saturation in the SHG power at 5 W IR pump
matching temperature (open squares) at different power. This is mostly due to a larger beam waist
IR pump powers from the fiber amplifier. It is size at the crystal center so that there is no pump
important to note that the value of the phase- depletion. Therefore, the achieved conversion effi-
matching temperature changes slightly when the ciency is somewhat lower than the maximum con-
IR pump power is changed. The phase-matching version efficiency of 2.62%W−1cm−1 provided by
temperature ranges between 64.0 ◦C and 64.3 ◦C thevendor. Alloftheseconsiderationssuggestthat
for IR pump power between 200 mW and 5 W. At thermal lens effects are negligible. The maximum
each power level the crystal is maintained at the achieved green power at 5 W IR pump power is
5
W) 1.8
mm) 1.2 wer (1.75
1) (eter (2e0.81 Output Po 1.7 ∆I2/I2 = 0.8% in 12 hours
m m 1.65
a n
Di 0.6 2
m 53 1.6 Cavity Power
a
Be 0.4 W (x)
1.55
W (y)
0.2 M2 (x) = 1.08– 0.01
M2 (y) = 1.07– 0.01 1.50 2 4 6 8 10 12
0 Time (hours)
-200 -100 0 100 200
Position (mm)
Fig. 7: Second harmonic power stability at the output of
PPLNismonitoredat1.74Wfor12hours.
Fig. 6: The transverse intensity profile (shown in the in-
set)ofthegreenbeammeasuredbyalaserbeamprofilerat
1.74 W green power. Closed and open circles are the beam
single-pass SHG in a PPLN crystal pumped by a
diameters in the horizontal (x) and vertical (y) directions
andcontinuouslinesshowthetheoreticalfitstoextractthe YDFA has been used with the high-finesse Fabry-
M2 [30]valuesinbothdirections. Perot cavity as a photon source for the Compton
polarimeter during the three-month run period of
the PREx experiment [1] in 2010. In the following
1.74 W with a total conversion efficiency of 34.8%.
we describe how we establish a frequency locking
A primary goal of this work is to develop a stable
of this beam to a resonant frequency of the high-
and good-quality green beam at the Watt level, so
finesse Fabry-Perot cavity.
we did not pursue higher conversion efficiencies.
Thequalityofthegreenbeamisofmajorimpor-
3.2. High-finesse Fabry-Perot cavity
tanceinpowerenhancementwithanopticalcavity,
asanydeviationfromaperfectGaussianprofilewill The green beam is propagated through a series
leadtoarejectionofpowerbythehigh-finessecav- of optical elements (Fig. 3) to properly match the
ity and will not contribute to the power buildup fundamental cavity mode (TEM ) at the injection
00
insidetheFabry-Perotcavity. Inordertocheckthe point. After exiting the PPLN crystal and reflect-
beam quality, we focused the green beam after the ing from two dichroic mirrors (DC and DC ), the
1 2
PPLN into a 200 µm waist diameter with an addi- green beam is focused by a lens L (f = 75 mm)
0 0
tional lens and measured the beam size at different at 6.0 cm from the frequency doubling setup with
axialpositions. Fig. 6showsthegreenbeamprofile waist diameters 370 µm and 450 µm in the hori-
at1.74Wmonitoredbyalaserbeamprofiler(LBA- zontal and vertical directions. The focusing of the
FW-SCOR20; Spiricon) along with the theoretical incident beam at the CIP is accomplished by three
fit. Closedandopencirclesarethemeasuredbeam lenses denoted L , L and L , respectively. A di-
1 2 3
diameters in the horizontal (x) and vertical (y) di- verging lens L (f = -1.0 m) at 405 mm from
1 1
rections. The result of the theoretical fits (solid the frequency-doubling setup nearly collimates the
lines) to the experimental data determines an M2 beam after it passes through an optical isolator
[30] value of ∼ 1.1 in both dimensions confirming (isolation: 40 dB, IO-3-532-HP; Thorlabs) and a
theTEM spatialmodeintheSHGprocess. How- half-waveplate(HWP).Inordertoachievethecor-
00
ever, there is a small astigmatism mainly arising rect beam waist size at the cavity center, we have
from the difference in refractive indexes along the two more lenses, L and L (f = -50 mm, f =
2 3 2 3
horizontal and vertical axes of the PPLN crystal. 200 mm), installed at 820 mm and 1.0 m from the
As shown in Fig. 7, the peak-to-peak stability of frequency-doubling setup. Here in our setup, L
1
greenpowerat1.74Wis0.8%fortheentireperiod and L are on fixed mounts and L is mounted on
2 3
of12hours,whilethefundamentalIRpowerstabil- a remote controlled translation stage (M-605.1DD;
ity from the YDFA is measured to be better than Physik Instrumente). This allows fine tuning and
0.6% over 4 hours. precise alignment of the laser beam to the cavity
The frequency-doubled green beam based on by using another pair of remotely controlled mir-
6
(PPLN Profile)
rors (M and M ) that allow four degrees of mo-
tion (2 t1ranslatio2ns, 2 rotations) for the laser beam mm) 3 L3 = 0.2 m vertical
with respect to the optical axis of the cavity. Both 1) (2e horizontal
the proper focusing and precise alignment of the er ( 2
et
iomrdpeirngtoinagchlaiesevregboeoamdmtoodtehemacatcvhitiynga.rTehneeemdierdroirns m Diam 1 L1 = -1.0 m L2 = -0.05 m Mce Mcs
Mr1, Mr2, Me andMs arefixedat45◦ withrespect Bea 0
Cavity
totheincidentbeam. Acamera(CCD )facingthe
1 -1
mirror M monitors the positions of the incident
r2
andreflected beams fromthecavity. Anothercam- -2
era(CCD )atthecavityexitisusedformonitoring
2
theprofileofthetransmittedbeamfromthecavity. -3
As shown in Fig. 3, our cavity consists of 0 500 1000 1500 2000 2500 3000 3500
Distance from Laser Head (mm)
two (M , M ) identical highly reflective (R =
ce cs
99.975%) concave mirrors (Advanced Thin Films) Fig.8: Thecalculatedbeamsize(beamenvelope)versusthe
separated by 850 mm. The mirrors have a typical distance along the beam path from the frequency doubling
transmittance of 240 ppm (parts per million) and setup.
total nominal (scattering and absorption) loss of <
10 ppm. This leads to a maximum power enhance-
√ a quarter-wave plate (QWP ), a holographic beam
2
ment of G ≈ R/(1−R) = 4,000. The coating is
sampler (HBS) and a Wollaston prism (Fig. 3 and
madeofalternatingdielectricquarter-wavelayersof
Fig. 9)thatareusedformeasuringandmonitoring
SiO and Ta O on a super polished (surface RMS
2 2 5 the laser polarization (discussed in Section 4).
roughness=0.5˚A)fusedsilicasubstrate(thickness
4 mm, diameter 7.75 mm) by an ion beam sputter-
3.3. Mechanical design of the cavity
ing technique. The mirrors have a typical radius of
curvature of 0.5 m that requires a Gaussian beam The strongest requirements on the design of a
waist diameter of 348 µm at the center of the cav- Fabry-Perot cavity in a Compton polarimeter are
ity[30]. Thisgeometrywaschosentomaximizethe that it must be rigid and must have a small cross-
electron-photon collision luminosity when the laser ing angle with the electron beam. The small cross-
light is frequency locked to the cavity and crosses ing angle increases the electron-photon interaction
with the electron beam. luminosity. The geometry of the cavity is deter-
Acarefulstudyofbeampropagationfromthefre- minedbythetotaldistancebetweenthetwomirrors
quency doubling setup to the cavity through each (850 mm), the radius of curvature of the mirrors
opticalelementhasbeenperformedusingOptoCad (500 mm) and the crossing angle between the laser
[31]. Fig. 8 shows the calculated beam size ver- and electron beams. With this design the cross-
sus the distance along the beam path from the fre- ing angle is 24 mrad, the gap between the electron
quency doubling setup. Experimental verification beam and mirror edge is 6.12 mm and the laser
of the laser waist size at the center of the cavity is beam size on the cavity mirror is 898 µm.
not possible with the beam profiler. Therefore, we A mechanical drawing of the cavity is shown in
createdanauxiliaryopticalpath(2.375m)thathas Fig. 9. The cavity and all the other optical ele-
thesamepathlengthfromthedoublingsetuptothe ments are mounted on an optical table with a foot-
cavitywithalltheopticalelementsinplace. Instead print of 1.5 m × 1.2 m. This table is placed on
ofM ,weusedanuncoatedmirrorsubstratemade a laminar flow damping system consisting of four
ce
of the same fused silica material with the same ra- pneumatic posts with auto-leveling valves to iso-
dius of curvature. We experimentally checked the late the vibrations from the ground. The opti-
beam waist size and location after this mirror sub- cal table is located inside a small room equipped
strate, andcomparedittoourcalculation. Asmall with a laminar flow fan filter unit. In order to
correction with L (few mm) is necessary in order ensure thermal and mechanical stability, an Invar
3
to get an average beam size of 350 µm at a dis- (64FeNi) frame consisting of three cylindrical rods
tance of 425 mm (corresponding to the cavity cen- attachedtotwoverticalplates,eachwithanoctago-
ter) from this mirror substrate. At the exit of the nalcutout,formsthebackboneofthecavityframe.
cavity,thereisasetofopticalelementscomposedof The cavity vacuum chamber is a cylindrical vessel
7
Fig.9: 3DmechanicaldrawingofthelaseropticsandFabry-PerotcavitysystemintheHallAtunnelatJLab.
madeofstainlesssteelwithadiameterof4.5inches axis. The mirror holder is machined such that the
(Fig. 9)connectedtotwooctagonalgimbalmounts rotationaxisofeachmirrorliesonthesameaxisas
through two soft bellows. The bellows allow the thegimbals. Thebearingsare0.25inchindiameter
gimbal mounts to tilt freely in both the horizontal and0.4inchinlength, andeachcansupportaload
and vertical planes when they are adjusted by four up to 100 lbs. Two remote controlled picomotors
piezoactuatorsunderatmosphere. However,inour (8302; Newfocus) attached to each gimbal mount
experiment, the bellows were too stiff under vac- are used for aligning each cavity mirrors by tilting
uum for the piezo actuators to move the gimbals the gimbal mounts in both the vertical and trans-
remotely. Therefore the procedure involves lock- verse planes with respect to the laser beam prop-
ing the cavity under atmosphere, then locking the agation direction. A pair of counteracting spring
gimbals and drawing vacuum1. The cavity mirror plungers (maximum load 13 lbs) are attached to
is mounted on a mirror holder attached to these the gimbal mounts to keep the alignment in posi-
gimbal mounts which are also made of Invar. Each tion. The picomotors are interfaced to the EPICS
gimbal mount is supported by four stainless steel [32] slow control system (discussed in Section 3.4)
cylindricalbearings(C-FlexBearingCo.) thatform that allows remote alignment of the cavity mirrors.
two axes (horizontal and vertical) for each gimbal A pair of vacuum window substrates (W and
e
mount and allow them to tilt freely around each W in Fig. 3) that are made of fused silica (3 mm
s
thickand0.7inchindiameter)allowthelaserbeam
1 Eventhoughwewereabletokeepthecavityalignment to enter and exit the cavity via a pair of 0.5 inch
in position during the three-month period of the PREx ex- turning mirrors (M and M in Fig. 3) oriented at
e s
periment,webelievethisfeaturecouldberedesignedtomake 45◦ with respect to the incident laser beam. The
theremotealignmentofcavitymirrorsundervacuumpossi-
vacuum windows are AR coated for 532 nm and
ble.
8
The error signal is created by phase detection be-
tween the photodiode (PDR in Fig. 11) output
of the modulated reflection signal and the modu-
lation signal from the function generator (FG220,
Yokogawa). The error signal includes information
on cavity resonance, and is then used to build the
feedbacksignalssuppliedonboththefastandslow
control ports of the seed laser. The electronic feed-
back system has been designed and built by Saclay
[13]. The simplified flow chart is shown in Fig. 11.
The optical elements that are critical to beam
alignment and polarization control and measure-
ment, the lasers (seed laser, fiber amplifier and
Fig. 10: The cavity vacuum chamber attached to two ion
PPLNdoublingsystem)andthelockingelectronics
pumps on the optical table and connected to the electron
beampipeintheHallAtunnelatJLab. Theelectronbeam are all interfaced to a remotely controlled system
pipeabovethecavityisusedforastraightbeamtotheex- through a local workstation. Electronics specific to
perimentalHallwhentheComptonpolarimeterisnotused.
the feedback control loop, the function generator
(rampgenerator, modulationsignalgenerator), the
oscilloscope and the workstation are located in the
welded to stainless steel flanges (Fig. 9) by the
controlroom. Thelasers,stepperandservomotors
glass-metal soldering technique. Two soft bellows
with their control units, photodiodes, CCD cam-
on both sides of each gimbal mount are connected
eras and the preamplifiers are located on the opti-
tobothsidesofthecavityvacuumchamberthrough
caltableintheHallAtunnel. Themonitorsrelated
apairofflangesandallowthegimbalstotiltfreely.
Each of the 45◦ turning mirrors is mounted to an to the cavity resonance mode and beam alignment
areconnectedtocamerasCCD andCCD through
aluminumholderthatisattachedtoastainlesssteel 2 1
coaxial BNC cables and are located in the control
flange mounted on a post. Each aluminum holder
room. Fig. 12 shows a functional view of the laser
has a slit 4 cm long and 1cm wide that allows the
and cavity system with control units.
electron beam to pass through. When the cavity is
Anautomaticswitchingsystemfrom“open-loop”
installed in the electron beam line (Fig. 10), the
modeto“closed-loop”modearoundthecavityreso-
stainless steel flanges are connected to the beam
nanceregionpermitsthesystemtotransitionauto-
pipebyanothersetofsoftbellowsthatisolatesany
matically between the resonant and locked states.
vibrations from the rest of the electron beam pipe.
It consists of an electronic circuit and an EPICS
Two beam position monitors located on both sides
program that manage the laser temperature scan
of the cavity are used to monitor changes in beam
via the slow control port of the seed laser. A 10 V
position of the electron beam during beam tuning
peak-to-peak triangular ramp, together with a si-
anddatataking. Twoionpumpscanprovideavac-
uum of ∼ 10−9 Torr inside the cavity after several nusoidal modulation signal at Ω = 928 kHz with
hours of bake-out at ∼ 100 ◦C. 50 mV amplitude, is supplied on the laser piezo-
electric transducer (PZT) via the fast control port
of the seed laser.
3.4. Feedback and slow control system
Two fast photodiodes (S1223; Hamamatsu) are
Maintaining the resonance and therefore the en- usedformonitoringthereflected(PDR)andtrans-
hancement in the cavity requires feedback control mitted (PDT) signals. They are held at a constant
of the laser frequency. The optical frequency sta- voltage level of 5 V. The currents from the pho-
bilization must be better than the free spectral todiodes are transformed into voltage signals via a
range (FSR=176.5 MHz) divided by the finesse trans-impedance amplifier that allows signal trans-
(F=12,000), whichis∼15kHz(cavitybandwidth). missionacross100metersofcoaxialcables. Aband-
This is equivalent to a stabilization of the 850 mm passfilterisappliedonthereflectedsignalfromthe
cavitylengthtobetterthanthelaserwavelengthdi- cavity after it is mixed with the modulation signal
vided by the finesse, which is ∼ 0.04 nm. The elec- at frequency Ω and amplified. The mixing and fil-
tronic feedback system that controls the laser fre- teringextractsan errorsignalcorrespondingtothe
quency uses the Pound-Drever-Hall technique [33]. change in reflected light at the laser modulation
9
FunctionuGenerator Digital
Modulation Analog
u9B8ukHz
Phase
Lock Linux
Ramp Shifter Control Workstation
FAST LowuPass
Filter
Sum SLOW Amp Mixer EthernetNetwork
FeedbackuSystem
PDR
B 7 6 v
PBS QWP PDT 7Q Q6 Q9 Qg
Piezo TEC FiberuAmplifier SHG Cavity MV MV CV CV
SeeduLaser I I
VMEuCrate
Fig.11: Asimplifiedflowchartofthefeedbackcontrolsystemusedforlaserfrequencylockingtothecavityfundamentalmode.
frequency. The value of Ω was determined by min- theslowdriftinlaserfrequencywhilethefastmode
imizing the laser Residual Amplitude Modulation allows the efficient reduction of the laser frequency
(RAM)[13]. Theerrorsignalisinjectedintoaseries jitter. The output signals of these two modules are
ofthreeseparateintegrators(proportional-integral- applied directly to the two laser control ports that
derivative or PID) common to the slow and fast control the seed laser frequency.
control loops. The two control modes play comple- The VME (VERSA Module Eurocard) crate,
mentary roles: the slow mode is for compensating used for controlling the electronics, is located in
VME Crate
766905 21 Q Optical
VM11VCVC1 VM7 AD Power Meter
II
QuadrantCell Servo Motor Picomotor Stepper Motor
M
Photodiodes Controller Driver s Controller
W
s
M
cs
QWP2 PDT
M HBS
3
M Wollaston
2 M
1 L2 CCD2
CCD1 Mce L3
W QWP
M e Me 4Q2 1 PBS Mr1
4Q1 r2 HWPL1
FOI Reflection Signal to
Locking Electronics
Slow Frequency
Control
Fast Frequency
Control
Fig.12: Afunctionalviewofthelaserandcavitysystemillustratestheelectroniccontrolunitswiththeopticalelements.
10