ebook img

350 micron Polarimetry from the Caltech Submillimeter Observatory PDF

1 MB·
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview 350 micron Polarimetry from the Caltech Submillimeter Observatory

Draftversion January18,2010 PreprinttypesetusingLATEXstyleemulateapjv.10/09/06 350µm POLARIMETRY FROM THE CALTECH SUBMILLIMETER OBSERVATORY Jessie L. Dotson,1 John E. Vaillancourt,2 Larry Kirby,3 C. Darren Dowell,4 Roger H. Hildebrand,3,5 and Jacqueline A. Davidson6 Draft version January 18, 2010 ABSTRACT We present a summary of data obtained with the 350µm polarimeter, Hertz, at the Caltech Sub- 0 millimeter Observatory. We give tabulated results and maps showing polarization vectors and flux 1 contours. The summary includes over 4300 individual measurements in 56 Galactic sources and 2 0 2 galaxies. Ofthesemeasurements,2153haveP 3σp statisticalsignificance. Themedianpolarization ≥ of the entire data set is 1.46%. n Subject headings: catalogs — H II regions — ISM: clouds — ISM: magnetic fields — polarization — a submillimeter J 6 1 1. INTRODUCTION tion vectors. The table contains all the data satisfying TheUniversityofChicago350µmpolarimeter,Hertz7, the criterion F > 3σF, where F is the total unpolarized ] flux density. A was operated at the Caltech Submillimeter Observatory (CSO)from1994to2003. Manyoftheresultshavebeen G 2. INSTRUMENTANDOBSERVATIONS published in some form but usually without giving com- h. plete information. In this paper we present a catalog The instrument, calibrations, and data analysis tech- p of all the measurements giving coordinates, position an- niqueshavebeendescribedindetailby Schleuning et al. - gles, degrees of polarization, and uncertainties as well (1997), Dowell et al. (1998), and Kirby et al. (2005). o as maps showing polarization vectors and flux contours Here we give a brief overview of the instrument and ob- tr derived from the same measurements. An earlier paper serving techniques. s by Dotson et al.(2000) presented a similar summary for An interference filter provided peak transmission at a results at 60µm and 100µm from the far-infrared po- 353µm and a bandwidth of 62µm (see Figure 1 in [ larimeter, Stokes, operated on the Kuiper Airborne Ob- Dowell et al.1998). Afterpassingthroughaquartzhalf- 1 servatory. wave plate turned by a stepper motor, the incoming ra- v For these bright sources we attribute the polarized diationwasseparatedintoorthogonalcomponentsofpo- 0 flux to thermal emission from magnetically aligned non- larizationbyapolarizinggridinclinedat45◦ totheoptic 9 spherical dust grains (e.g., Hildebrand 1988). The de- axis. Thereflectedandtransmittedcomponentswerede- 7 gree and angle of the polarization are useful for inves- tectedsimultaneously foreachpixelonthe sky by corre- 2 tigating the intrinsic properties, alignment mechanism, sponding pairs of NTD germanium bolometers in arrays . 1 and populations of dust grains (e.g., Hildebrand et al. of 32 pixels each, arranged in a 6 6 matrix with the × 0 1999, Hildebrand & Dragovan 1995) and the struc- corners omitted. Winston cones concentrated the radia- 0 ture of magnetic fields (e.g., Hildebrand et al. 1990, tionintocylindricalcavitiescontainingsiliconbolometers 1 Schleuning et al. 2000, Houde et al. 2002). cooled to 0.26K by a dual stage 3He cryostat. v: We discuss the instrument and observations in Sec- Thehalf-waveplatewasmovedin30◦stepsthroughsix i tion 2, and the instrumental and systematic effects in positions. Ateachpositionstandardchoppedandbeam- X Section 3. In Section 4 we present the object list (Ta- switchedphotometric observationswere performed. The ′ ′ r ble 1) and table of results (Table 2). In Table 1 we also chop throw varied between 2 and 8 depending on the a give references where earlier papers have been based on sizeoftheobject. Thechopfrequencywas 3Hz. Each ∼ the samedata. In allothercases,this is the firstpresen- data file took approximately 4 – 5 minutes including tation of complete data sets. timespentintegrating,beamswitching,andsteppingthe All results in Table 2 and the figures are for polar- halfwaveplate. Inexcellentweatherconditionsthenoise ization vectors. The inferred directions of the magnetic equivalentfluxdensityformeasurementofpolarizedflux field vectors (not shown) are orthogonalto the polariza- was 3 – 4 JyHz−1/2. Relative photometric measurements were obtained si- 1NASA Ames Research Center, Astrophysics Branch, MS245- multaneously with the polarimetric measurements. Ar- 6,MoffettField,CA94035 ray flat fielding was obtained by pixel dithering as de- 2DivisionofPhysics,Mathematics,&Astronomy,CaliforniaIn- scribedbyDowell et al.(1998). Theindividualdatafiles stitute of Technology Current address: Universities Space Re- searchAssociation,NASAAmesResearchCenter,MS211-3,Mof- werecorrectedfortheirgainbynormalizingtothebright fetField,CA94035 flux peak. In most cases,we did not Nyquist sample the 3Enrico Fermi Institute and Department of Astronomy & As- imageplane; the center-to-centerspacingof the detector trophysics,UniversityofChicago ′′ 4JetPropulsionLaboratory array was measured to be 17.8. As a result, the resolu- 5alsoDeptartment ofPhysics tion in our flux maps is √2 our nominal beamsize 6UniversityofWesternAustralia of 20′′. ∼ × 7 Now replaced by a new polarimeter, SHARP (Novaketal. Peak fluxes for all objects presented here are given in 2004;Lietal.2008). Table 1 andin the captions ofFigures 2 – 56. Fluxes for 2 eleven objects were found in the literature (see Table 1, between the source and reference positions. However, “Flux Reference” column). These eleven objects, along this is not the case for polarimetry because the polar- with observationsofW3OH, Mars,Uranus,andJupiter, izations do not subtract simply. Methods to estimate weretakenasfluxstandardsforcalibratingallremaining these systematic uncertainties have been developed by objects. Thecalibrationswereperformedwithrespectto Schleuning et al. (1997) and Novak et al. (1997). the standardswithin eachobservingrun. Averagefluxes Estimates for the limits to these uncertainties include: are reported for objects observed in multiple runs. (1) ∆P+ , the maximum amount by which the actual sys value of P could be larger than the measured value, 2.1. Instrumental Effects − (2) ∆P , the maximum amount by which it could be sys Allofthemeasurementsmadeduringanobservingrun smaller, and (3) ∆φ , the maximum amount by which sys werecombinedtocalculatethe telescopeandinstrumen- themeasuredpositionanglecouldbe inerror. Themag- tal polarization as described by Platt et al. (1991). The nitudes ofthese limits aredependent onthe ratioof flux mean instrumental polarization across the array varied in the source beam to that in the reference beam and from 0.23% – 0.38% on different observing runs. The the ratio of the polarization in the source and reference telescope polarizationvariedfrom0.04%– 0.48%on dif- beams (for more details see Schleuning et al. 1997 and ferentruns,withameanvalueof0.22%. Allresultswere Novak et al. 1997). For any given polarization measure- corrected for telescope and instrument polarizations as ment, these maximum uncertainties can be estimated if well as a polarization efficiency of 95%. the polarized flux is known in both the source and ref- erence positions. However, few large scale 350µm pho- 2.2. Systematic Effects tometric maps, and no large scale submillimeter polari- Amajorobstacletothemeasurementofsubmillimeter metric maps, exist from which to make these estimates. polarizationisthe inevitable fluctuationsinatmospheric Therefore,we havemade no effort to estimate these sys- transmission and emission. The removal of background tematics here. sky emission is mostly achieved by observing reference regionsawayfrom the sourceof interest and subtracting 2.3. Positive Bias the reference position measurement from the on-source Since polarization is an inherently positive quantity, a position measurement. Rapid (3 – 4 Hz) modulation positive bias is introduced into any measurement with (“chopping”) of the telescope secondary mirror between noise. For high signal-to-noise ratios (P/σ & 4) mea- off-sourceandon-sourcepositionsallowsremovalofmost p surements can be best corrected for this bias using the fluctuations in sky emission. Slow ( 0.05Hz) beam ∼ relation switching (“nodding”) facilitates removal of linear gra- dients in the sky emission. The effects of fluctuations in P0 ≈qPm2 −σp2 (1) the transmission of the atmosphere are largely removed by observing two orthogonalcomponents of polarization whereP0 is the correctedpolarization,Pm the measured simultaneously. While the instrumental design of Hertz polarization,andσp themeasureduncertaintyinthepo- (two-component polarimetry) and our observing strat- larization (e.g., Simmons & Stewart 1985). Extending egy (chop-nod) allow us to remove many of the effects thisrelationtoslightlylowersignal-to-noiseyieldsabias of the sky fluctuations, these strategies do not allow us less than 0.06 Pm for Pm > 3σp. Given this small × to correct for fluctuations on all the relevant physical level of bias, and the fact that equation (1) is not an and time scales, so there is still a residual noise in our exact solution for P/σp . 4, we have made no attempt measurements due to sky fluctuations. A relatively re- to correct the polarizations presented in Table 2. How- cent data analysis approach, developed by Kirby et al. ever, we caution that readers wishing to make precise (2005), capitalizes on making repeated observations of comparisonsbetween this archiveandother polarization the source and measurements of the atmospheric optical measurementsshould consider that some bias-correcting depth to further reduce the residual effects of sky fluc- scheme may be appropriate for some studies. tuation. While our traditional two-array measurement Thebiascorrectionhas beenappliedtoallplottedvec- of the polarization (e.g., Platt et al. 1991) is quite effec- tors in Figures 2 – 56, including those with P/σp < 4. tive at measuring polarized flux, the Kirby et al. (2005) With this correction, any measurement with Pm < σp approachsignificantly improvesthe estimationofthe re- yieldsP0 =0. However,upper limits onthe polarization ducedStokesparameters(polarizedfluxdividedbytotal can still be placed on these measurements (Vaillancourt flux). This new approachhas allowedus to extend mea- 2006). These points are shown by circles on the maps surements to fainter sources. when P +2σp < 1%. This 2σ criterion represents con- If the distance between the source and the reference fidence levels between 95 – 99% depending on the exact positionsissmallcomparedtothesizeofthesource,then level of the polarization signal-to-noise. it is possible to erroneously subtract source flux along withtheskybackground. Largechopthrows( 6′)were 3. RESULTS ∼ chosen for most Hertz observations in order to mitigate We have made two cuts to the polarization data set this problem. The size and range of directions of the in order to eliminate excessively noisy points. First we chop used for each object are given in Table 1. requirethateachpositionbemeasuredwithatleast3po- Despite the large chop throws available at the Caltech larizationfiles. Second, every measuredpoint must have Submillimeter Observatory, it is not always possible to a flux signal-to-noise ratio 3. When the flux S/N < ≥ avoid the problem of source flux in the reference beam. 3, we have not successfully measured total flux, making This is not a significant problem for submillimeter pho- itimpossible to havesuccessfully measuredthe polariza- tometry as one can at least report the flux difference tion. Evenwithflux S/Njustabove3,itis unlikely that 3 Fig. 1.— Distributionof all measured polarizations (solid line) and of measurements with a 3σp or better statisitical significance (dash-dotted line). Nocorrection has been madefor positive bias (seeSection2.3). we have successfully measured the source polarization, but in many cases the flux information may be of use. We have made no cuts based on the value of the polar- izationorofthestatisticaluncertaintyofthepolarization measurements; such cuts could bias the statistics of the polarization distribution. The cuts we did apply should only reinforce the already existing bias of this data set towards brighter regions of submillimeter flux. The distribution in degrees of polarizationis shown in Figure 1. Among the 4372 measured points, 2153 have 3σ or greater statistical significance, 990 6σ, and 350 ≥ 12σ. The median P and P/σ for all of the measure- p ≥ ments are 1.46% and 2.93, respectively. The distribu- tion of measured polarizations and how the distribution changes with wavelengthare discussed by Hildebrand et al. (1999). TheresultsaretabulatedinTable2andshowninFig- ures 2 – 56. The plotted vectors have all been debiased (see Section 2.3). The results listed in Table 2 have not been debiased. All maps include a reference vector and ′′ a shaded circle indicating the Hertz beam size of 20 . Coordinates in right ascensionand declination are offset from the coordinates given in the caption and Table 1. Solid vectors denote measurements where P/σ 3; P ≥ dotted vectors denote 2 P/σ < 3; and open circles P ≤ denote P + 2σ 1% . Note that any vector which P ≤ ◦ meets the criterion P/σ > 3 has ∆φ < 10 and any P ◦ vector with P/σ > 2 has ∆φ < 15 . Flux contours P ′′ were generated by smoothing the data with a 20 gaus- sian, choosen to mimic the Hertz beam. No smoothing was applied to the polarizations and fluxes tabulated in Table 2. 4 Fig. 2.—NGC253. Offsets from0h47m33s.1,−25◦17′15′′ (J2000). Contoursat10,20,...,90%ofthepeakfluxof110Jy. 5 Fig. 3.—W3. Offsetsfrom2h25m40s.7,62◦5′52′′ (J2000). Contoursat10,20,...,90%ofthepeakfluxof480Jy. 6 Fig. 4.—NGC1333. Offsetsfrom3h29m3s.8,31◦16′3′′ (J2000). Contours at10,20,...,90%ofthepeakfluxof75Jy. 7 Fig. 5.—L1551. Offsetsfrom4h31m34s.2,18◦8′5′′ (J2000). Contoursat10,20,...,90%ofthepeakfluxof95Jy. Fig. 6.— IRAS 05327-0457. Offsets from 5h35m14s.4, −4◦55′45′′ (J2000). Contours at 60, 70, 80, 90, 95, and 98% of the peak flux of 25Jy. Thelengthofa10%vector isshownforscale 8 Fig. 7.—OrionMolecularCloud(OMC-4). Offsetsfrom5h35m8s.2,−5◦35′56′′ (J2000). Contours at30,40,...,90%ofthepeakfluxof 37Jy. 9 Fig. 8.—OrionMolecularCloud(OMC-1). Offsetsfrom5h35m14s.5,−5◦22′32′′ (J2000). Contoursat4,6,8,10,20,...,90%ofthepeak fluxof2100Jy. 10 Fig. 9.—OrionMolecularCloud(OMC-3). Offsetsfrom5h35m23s.5,−5◦1′32′′ (J2000). Contours20,...,90%ofthepeakfluxof110Jy.

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.