The Ionized Circumstellar Envelopes of Orion Source I and the Becklin Neugebauer Object

0.29 MB·English

by  R. L. Plambeck

accepted forpublicationin ApJ
PreprinttypesetusingLATEXstyleemulateapjv.5/2/11
THE IONIZED CIRCUMSTELLAR ENVELOPES OF ORION SOURCE I
AND THE BECKLIN-NEUGEBAUER OBJECT
R. L. Plambeck1, A. D. Bolatto2, J. M. Carpenter3, J. A. Eisner4, J. W. Lamb5, E. M. Leitch5,6,
D. P. Marrone4, S. J. Muchovej5, L. M. P´erez3, M. W. Pound2, P. J. Teuben2, N. H. Volgenau5, D. P. Woody5,
M. C. H. Wright1, B. A. Zauderer7
accepted for publication in ApJ
ABSTRACT
3
1 The 229 GHz (λ1.3mm) radio emission from Orion-KL was mapped with up to 0.14′′ angular
0 resolution with CARMA, allowing measurements of the flux densities of Source I (‘SrcI’) and the
2 Becklin-Neugebauer Object (BN), the 2 most massive stars in this region. We find integrated flux
densities of 310±45 mJy for SrcI and 240±35 mJy for BN. SrcI is optically thick even at 229 GHz.
n
No trace of the H30α recombination line is seen in its spectrum, although the v =1, 5(5,0)-6(4,3)
a 2
J transition of H2O, 3450 K above the ground state, is prominent. SrcI is elongated at position angle
140◦, as in 43 GHz images. These results are most easily reconciled with models in which the radio
8
emissionfromSrcIarisesviatheH− free-freeopacityinaT<4500Kdisk,asconsideredbyReid et al.
2
(2007). Bycontrast,theradiospectrumofBNisconsistentwithp+/e− free-freeemissionfromadense
] (ne ∼5×107 cm−3), but otherwise conventional, hypercompact HII region. The source is becoming
R opticallythinat229GHz,andtheH30αrecombinationline,atV =23.2±0.5kms−1,isprominent
LSR
S initsspectrum. ALymancontinuumflux of5×1045 photonss−1,consistentwiththatexpectedfrom
. a B star, is required to maintain the ionization. Supplementary 90 GHz observations were made to
h
measure the H41α and H42α recombinationlines towardBN. Published 43 and 86 GHz data suggest
p
that SrcI brightened with respect to BN over the 15 year period from 1994 to 2009.
-
o Subject headings: ISM: individual(Orion-KL)— radio continuum: stars — radio lines: stars — stars:
r formation — stars: individual (Becklin-Neugebauer-Object)
t
s
a
[ 1. INTRODUCTION Goddi et al. 2011). Tracing the motions backward,
Goddi et al.(2011)findthat560yearsagotheprojected
2 The Kleinmann-LowNebula inOrionis well-knownas
v the nearest region of high mass star formation, 415 pc separation of these two stars in the plane of the sky
5 away (Menten et al. 2007; Kim et al. 2008). It contains was just 50±100 AU. An extensive system of ‘bullets,’
8 atleast2massiveyoungstars. Oneofthese,theBecklin- bow shocks, and ‘fingers,’ visible in lines of H2, FeII,
0 NeugebauerObject(BN),hasbeenstudiedextensivelyat and CO (Allen & Burton 1993; Zapata et al. 2011), also
0 is centered on Orion-KL. Proper motion measurements
infrared wavelengths (Scoville et al. 1983); it is thought
. suggest that the fingers were created by an explosive
1 to be a B star. The other object, Source I (hereafter,
event 500-1000 years ago (Doi et al. 2002; Bally et al.
1 ‘SrcI’), is so heavily obscured by foregrounddust that it
2011). Thus, the currently favored paradigm for Orion-
2 is not directly visible in the infrared, although light re-
KL (G´omez et al. 2008; Bally et al. 2011; Goddi et al.
1 flected by the surroundingnebulosity providesa glimpse
2011) postulates that SrcI and BN were ejected from a
: of its spectrum (Morino et al. 1998; Testi et al. 2010).
v multiple system approximately 500 years ago, and that
SrcI is noteworthy because it is surrounded by a cluster
i the ejection of the stars unbound the surrounding gas
X of SiO masers, one of the few cases in which SiO masers
and circumstellar disks, creating the finger system.
are associated with a young star. It also is known to
ar drive a bipolar outflow into the surrounding molecular BN’sluminosityof(0.8−2.1)×104L⊙(De Buizer et al.
cloud (Plambeck et al. 2009). 2012) suggests that it is a 10 to 15 M⊙ star
(Schaller et al. 1992). Estimates of the mass of SrcI are
Remarkably, proper motion measurements show that
conflicting. SrcI is recoiling at about half the speed
SrcI and BN are recoiling from one another at 35-
40 km s−1 (Rodr´ıguez et al. 2005; G´omez et al. 2008; of BN in the rest frame of the Orion Nebula Clus-
ter (Goddi et al.2011); conservationof momentum then
1Radio Astronomy Lab, Hearst Field Annex, University of suggests that its mass is roughly 20 M⊙. On the other
California,Berkeley,CA94720USA hand, if the SiO masers near SrcI are in Keplerian ro-
2Department of Astronomy, Universityof Maryland, College tation about the star, the inferred central mass is only
Pa3rkD,eMpaDrt2m0e7n4t2oUfSAAstronomy, California Institute of Technol- 7 M⊙ (Matthews et al. 2010); this can be interpreted as
alowerlimit ifthe SiO-emitting gasis supportedinpart
ogy,Pasadena, CA91125USA
4DepartmentofAstronomy,UniversityofArizona,933North by radiation or magnetic pressure. SrcI could well be
CherryAve.,Tucson,AZ85721USA a compact binary with a semimajor axis of < 10 AU
5Owens Valley Radio Observatory, California Institute of (G´omez et al. 2008).
Technology, P.O.Box968,BigPine,CA93513USA
6Department of Astronomy and Astrophysics, University of At cm and mm wavelengths the continuum spectra of
Chicago, 5640SouthEllisAve.,Chicago, IL60637USA BN and SrcI are believed to be dominated by free-free
7Department of Astronomy, Harvard University, Cambridge, emission from ionized circumstellar gas that is at least
MA02138USA
2 Plambeck et al.
partially optically thick up to 100 GHz (Plambeck et al. Wegeneratedaprimary-beamcorrectedcontinuummap
1995). Observations at higher frequencies, where this from the remaining channels, using multifrequency syn-
emissionshouldbecomeopticallythin,canbeusedtoes- thesis to avoid radial smearing.
timate the Lyman continuum fluxes from the stars, bet- TohelpcorrecttheOriondataforatmosphericdecorre-
ter constraining their masses. Such observations must lation,asinglephase-onlyself-calibrationwasperformed
be made with subarcsecond angular resolution in order on the source itself. This means that after cleaning the
to distinguish the compact circumstellar emission from continuum map in the normal way, the brightest pixels
bright but extended dust and molecular line emission in the clean component list were used as a source model
from the surrounding molecular cloud. to refine the antenna-based phases. A time interval of 2
Here we describe high angular resolution 229 GHz ob- minuteswasused,shortenoughto trackthe mostsignif-
servationsofOrion-KLwiththeCombinedArrayforRe- icant atmospheric phase variations, but long enough to
search in Millimeter-Wave Astronomy (CARMA) that obtainasignaltonoiseratioof≥1oneachofthe91avail-
cleanly resolve the emission from both stars. Surpris- able baselines, allowing a robust least squares fit for the
ingly, we find that the emission from SrcI is optically 13 antenna-based phases. Applying the self-calibrated
thick even at 229 GHz, and we fail to find the H30α re- phase corrections increased the source flux densities by
combination line in its spectrum. We argue that these about 50%. The measured rms noise in the final 0.25′′
results favor models in which the radio emission from resolution map is 4 mJy beam−1.
SrcI originates from the H− opacity in a T < 4500 K Even higher resolution images of Orion were ob-
disk. tained in 2009 February using the CARMA A-array,
with antenna separations of up to 1.7 km (1350 kλ).
2. OBSERVATIONS These observations utilized older, single-polarization re-
The spectra of SrcI and BN presented here were ob- ceivers and a 1.5 GHz bandwidth correlator. For these
tained from a single night’s observation of Orion made data the CARMA Paired Antenna Calibration System
with the CARMA B-array on 2011 January 06. Four- (“C-PACS,” P´erez et al. 2010) was used as an adap-
teen antennas were used; projected antenna separations tiveopticsschemetocorrectforblurringbyatmospheric
ranged from 50 to 700 kλ. The phase center was phase fluctuations. C-PACS placed the 8 CARMA 3.5-
05h35m14s.505, −05◦22′30′.′45 (J2000). The weather was meter telescopes next to a subset of the 10-m and 6-
excellent–theatmosphericopacityat225GHzwas0.10, m telescopes, including those at the extremes of the ar-
and the rms atmospheric phase fluctuations were 65 µm ray. As the 6-m and 10-m telescopes observed Orion at
on a 100-m baseline. Double sideband system tempera- 229 GHz, the 3.5-m antennas observed the nearby (1.6◦
turesforthe dual-polarizationreceiverswere100–200K, away)quasar0541-056at31GHzinordertomonitorthe
scaled to outside the atmosphere. For each polarization atmospheric phase variations above each station. The
the correlator was configured to observe three overlap- atmosphericphasedelayto each3.5-mtelescope wasde-
ping 500 MHz windows (covering the frequency range rivedby self-calibrationon0541-056every12 sec. These
225.8–227.2 GHz in the receivers’ lower sideband and antenna-based phases were scaled up by the ratio of the
231.4–232.8GHz in the upper sideband, with 12.5 MHz observing frequencies (229/31 = 7.4) and applied to the
wide channels), plus a single 250 MHz window (covering mm data. The C-PACS calibration accuracy is limited
226.65–226.90 GHz in the lower sideband and 231.75– by the angular separation of the cm calibrator and the
232.00 GHz in the upper sideband, with 3.125 MHz res- mmsource,whichcausesthecmandmmbeamstoprobe
olution). Upperandlowersidebandsignalsareseparated different paths through the atmosphere; by the signalto
in the correlator by phase-switching. noiseratioonthe31GHzcalibrator;andbytheairmass.
The antenna gains were derived from observations of AcontinuummapwasgeneratedfromtheA-arraydata
the calibrators0423-013(2.4 Jy) and 0607-085(1.05 Jy) usingonlybaselineslongerthan250kλ. Thesynthesized
thatwereinterleavedwiththeOrionscansevery10min- beamwas0.15′′×0.13′′atPA14◦. ApplyingtheC-PACS
utes. To minimize decorrelationfromatmosphericphase correctionsmorethandoubledthesignaltonoiseratioof
fluctuations, a self-calibrationintervalof 30 seconds was thismap. AswiththeB-arraymap,wethenperformeda
used. Thecalibratorfluxdensitieswereestablishedfrom single phase-only self-calibration, with a 2 minute inter-
observations of the primary flux standard, Uranus. Be- val, to help correct for residual atmospheric phase fluc-
cause Uranus is heavily resolvedon these baselines, only tuations; this doubled the flux densities of the compact
antenna pairs with separations < 200 kλ were used for sources. The measuredrms noise in the final 0.14′′ reso-
flux calibration. Antenna gain solutions were applied lution map is 2.6 mJy beam−1.
to the Orion data after smoothing to a 10 min interval The integrated flux densities measured from the 0.14′′
(scalar averaging the amplitude corrections). Only data and0.25′′ resolutionmapswere240and255mJyforBN,
with antenna separations greater than 250 kλ were used 310 and 370 mJy for SrcI. Although the discrepancy in
for the Orionmaps, yielding a 0.28′′×0.21′′ synthesized the two SrcI flux densities could conceivably be caused
beam at PA −27◦. by time variability of the source (cf. section 4.2.3), it
With 0.25′′ angular resolution, almost all spectral line is more likely attributable to differences in the sampling
emission from the Orion Hot Core and surrounding of visibilities by the A and B arrays, which lead to dif-
molecular cloud is resolved out. Nevertheless, to ensure ferent sidelobe structure from the adjacent Orion Hot
that continuum flux densities were not contaminated by Core. We will use the A-array flux densities for the re-
residual line emission or absorption, we made maps of mainderofthis paperbecausethehigherresolutiondata
all the spectral channels, and flagged those (28 out of should more effectively filter out emission from the Hot
234) with anomalously high (> 2× average) rms noise; Core. We estimate that the absolute flux scale is accu-
the noisy channels coincided with strong spectral lines. rate within ±15%.
Ionized Envelopes of Orion SrcI and BN 3
Figure 1. (left) 229 GHz continuum maps of Orion-KLfromCARMA C- and B-arraydata. Black contours show the 0.83′′ resolution
C-array map; the contour levels are −100,−50, 50, 100, 200, 300, and 400 mJy/beam. Negative contours are attributable to poorly
sampled extended emission. Red contours show the 0.25′′ resolution B-array image; contour levels are 20, 40, 60, 80, 100, 150, 200, and
250mJy/beam. Extended emissionisalmostentirelyfilteredoutinthis map;themostnegative valueis−30mJy/beam. (right)Spectra
of BN and SrcI near 232 GHz obtained from the 0.25′′ resolution image. Dashed lines indicate the continuum levels. The spectra have
beenHanningsmoothedto30kms−1 velocityresolution. TheH30αlineisdetected towardBN,theH2Ov2=1,5(5,0)-6(4,3) linetoward
SrcI.ArrowsintheSrcIspectrumindicateweakabsorptionfeatures fromtransitionsofethyl cyanide,dimethylether,andmethanol.
Finally, supplementary 3mm observations were made intherighthandpanelsofFigure1. TheH30αrecombi-
with the A-array in 2009 February to obtain spectra nationlineat231.901GHzis prominentinthe spectrum
of the H41α (92.034 GHz) and H42α (85.688 GHz) re- ofBN,butisnotdetectedtowardSrcI.Thebrightlineat
combinationlinestowardBN.Bothoftherecombination 232.687GHzinSrcI’sspectrumisthev =1,5(5,0)-6(4,3)
2
linesandthe strongSiOmaserat86.243GHzweremea- transitionofH O.WeakabsorptionfeaturestowardSrcI
2
sured simultaneously. The data were self-calibrated on are due to transitions of ethyl cyanide, dimethyl ether,
themaserusinga10secinterval. Thespectraofthetwo and methanol. No emission lines were detected toward
recombination lines were averaged together to improve either source in the lower sideband.
the signal to noise ratio. Figure 2 displays the 0.14′′ resolution images of BN
and SrcI from the A-array. BN is nearly circularly sym-
3. RESULTS metric, while SrcI is elongated. Table 1 summarizes the
The left hand panel in Figure 1 shows the 0.25′′ res- results of Gaussian fits to the positions, flux densities,
olution continuum map from the B-array observations, and deconvolved sizes. The 229 GHz source positions
overlaid on a 0.83′′ resolution image obtained with the matchthe43GHzpositionsmeasuredin2009Januaryby
CARMA C-array (Eisner et al. 2008). SrcI and BN are Goddi et al. (2011) within 0.015′′. The deconvolvedsize
clearlydistinguishableonlyinthehigherresolutionmap; of SrcI at 229 GHz, 0.20′′×0.03′′ at PA 140◦, is similar
extended emission from dust and molecular lines domi- tothe sizesmeasuredat8.4GHz (0.19′′×<0.15′′ atPA
nate the lower resolution image. 136◦; G´omez et al. 2008) and at 43 GHz (0.23′′×0.12′′
The upper sidebandspectraofBNandSrcI areshown at PA 142◦; Goddi et al. 2011), although the source ap-
pears to be thinner along its minor axis at 229 GHz.
The flux density we measure for BN in the 0.14′′ reso-
lution map is roughly 2× higher than the value recently
reported by Galv´an-Madrid et al. (2012) from an anal-
ysis of ALMA Band 6 Science Verification data. These
early ALMA data had a 1.3′′ ×0.6′′ synthesized beam.
WesuspectthatthefluxdensityofBNisdepressedinthe
ALMA map because of negative sidelobes from poorly
sampled extended emission. The same problem affects
the 0.83′′ resolution CARMA image shown in Figure 1;
the peak intensity atthe positionofBNinthis lowreso-
Figure 2. ImagesofBNandSrcIat229GHzfromtheCARMA lution image is ∼125 mJy/beam. The higher resolution
A-arraydata. Eachboxis0.6′′ onaside. Thelowestcontourand CARMAmapsarefreeofdeepnegativesidelobes,allow-
thecontour intervalis15mJybeam−1. Thermsnoiselevelis2.6 ing more reliable flux density measurements of compact
mJy beam−1. The 0.15′′×0.13′′ synthesized beam is shown by
sources.
a hatched ellipse in the upper left corner of each image. SrcI is
extendedatPA140◦,asin43GHzVLAimages(Reidetal.2007;
Goddietal.2011).
4. DISCUSSION
4 Plambeck et al.
Table 1
Sourceparameters forSrcIandBNat229GHz
source RA DEC peakfluxdensity integratedfluxdensity deconvolved size
(hms) (◦ ′ ′′) (mJy) (mJy)
SrcI 053514.514 -052230.59 170±25 310±45 0.20′′×0.03′′ atPA140◦
BN 053514.109 -052222.73 215±30 240±35 0.06′′×0.04′′ atPA90◦
Note. —Allparametersweremeasuredfromthe0.14′′ resolutionmap. Positionsareforepoch2009.1.
Figure 3. Radiospectra of BNand SrcI. Squares indicate continuum fluxdensities; circles,flux densities at the peakof recombination
lines. CARMA results presented in this paper are shown as red in the online edition. Solid and dashed lines show, respectively, the
continuum andrecombinationlinefluxdensitiespredictedbythemodelsdescribedinsections4.1.1and4.2.1.
Figure3showstheradiocontinuumspectraofBNand brightness temperatures at this projected radius in the
SrcI from cm to submillimeter wavelengths. Numerical continuum and at the peak of a recombination line are
valuesaregiveninTable2. ForBNwealsoplotthepeak then T (x) = T [1 − exp(−τ )] and T (x) = T [1 −
C e C L e
intensitiesoftheH30αandH41/42αrecombinationlines exp(−τ −τ )], where the continuum and line opacities
C L
measured with CARMA, and the H53α line measured are approximated as (Rohlfs & Wilson 2000)
with the VLA (Rodr´ıguez et al. 2009).
τ (x)=8.23×10−2T−1.35ν−2.1EM(x) (1)
C e GHz
4.1. BN τ (x)=1.92×103 T−2.5 EM(x)∆ν−1 . (2)
L e kHz
The continuum spectrum of BN is typical of free-free
emissionfromahypercompactHII region. Thefluxden- We integrate the brightness temperatures over circular
sityscalesasν1.3,indicatingthatopticallythickemission annuli to obtain flux densities, assuming a distance to
has a larger angular extent at lower frequencies, either the source of 415 pc.
because the ionized gas is clumpy (Ignace & Churchwell The model BN spectra shown by smooth curves in
2004) or because the electron density declines smoothly Figure 3 assumed Te = 8000 K, n0 = 5 × 107 cm−3,
r = 7.4 AU, α = −3.5, and recombination line veloc-
with radius (Wright & Barlow 1975). The spectrum be- c
gins to flatten at 229 GHz, indicating that the source is ity widths of 30 km s−1 for all transitions. The model
becoming optically thin at higher frequencies. does not take into account systematic velocity gradients
or pressure broadening of the recombination lines. The
4.1.1. Model spectra model parameters were determined by trial and error,
andare not a formalfit to the data. To first order,how-
To model the spectrum, we assume that the emission ever, the turnover frequency sets n , the spectral slope
0
region is spherically symmetric, with a uniform electron sets α, and the absolute flux densities set r .
c
temperature T andanelectrondensity that is constant,
e
ne(r)=n0,insidecoreradiusrc,thendeclinesasne(r)= 4.1.2. Excitation parameter
n (r/r )α atlargerradii. Foreachprojectedradiusxwe
0 c
numerically integrate along the line of sight to compute The model parameters above yield an excitation pa-
the emission measure EM(x) = Rn2dz pc cm−6. The rameter U = r (n n )1/3 = 5 pc cm−2, which serves as
e c e H
Ionized Envelopes of Orion SrcI and BN 5
Figure 4 displays the recombination line profiles to-
Table 2 ward BN. H30α and H41/42α (the average of the H41α
Integrated fluxdensitiesforSrcIandBN
andH42αlines)wereobservedwithCARMA,H53αwith
the VLA (Rodr´ıguez et al. 2009). The H30α spectrum
frequency S(SrcI) S(BN) epoch References
was measured using the 250 MHz wide correlator win-
(GHz) (mJy) (mJy)
dowwith4kms−1 channelspacing;ithasbeenHanning
smoothed to 8 km s−1 resolution. Line parameters and
4.8 ··· 2.2±0.8 1990 1
8.4 1.1±0.2 4.5±0.7 1994.3 2 uncertainties derived from Gaussian fits to the spectra
8.4 1.2±0.1 4.8±0.1 2006.4 3 are summarized in Table 3. The H53α line is at slightly
15 ··· 9.1±0.6 1981.6 4 lower velocity and is noticeably wider than the higher
15 ··· 9.4±1.0 1983.7 4
frequency lines.
1155 11.5.64±±00..418 66.1.89±±10..519 11998960.3 15 For the electron densities ne > 107 cm−3 that we
22 ··· 13.6±1.1 1983.7 4 infer for BN, collisional broadening of recombination
22 5.7±0.9 16.57±0.7 1991.5 6 lines can be significant. Collisional broadening is most
43 13±2 31±4.7 1994.3 2
43 10.8±0.6 28.0±0.6 1994.9 7 important for high-n transitions because the diameter
43 13 26.4±0.7 2000.9 3,8 of a hydrogen atom is proportional to n2 – an atom
43 14.5±0.7 28.6±0.6 2007.9 9 in the n=53 level is 0.3 µm across! The collision-
43 11±2 23±2 2009.0 10 broadened linewidths are given by the approximate for-
86 34±5 84±10 1995.0 11
89 50±5 91±9 2009.0 12 mula (Brocklehurst & Seaton 1972)
229 310±45 240±35 2009.1 13
348 320±48 ··· 2004.1 14 ∆vc/2=4.7c/ν(n/100)4.4(104/Te)0.1ne. (3)
690 6700±3200 ··· 2005.1 15
For n = 5 × 107 cm−3 and T = 8000 K, ∆v ∼
e e c
200kms−1 forH53α,30kms−1 forH41α,and3kms−1
References. — (1) Fellietal. 1993b; (2) Menten&Reid
1995; (3) G´omezetal. 2008; (4) Garayetal. 1987; for H30α. These widths must be convolved with the ex-
(5) Fellietal. 1993a; (6) Forbrichetal. 2008; pected ∼ 20 km s−1 wide thermal profiles, plus any ad-
(7) Chandler&Wood 1997; (8) Reidetal. 2007; ditional Doppler broadening from turbulent motions.
(9) Rodr´ıguezetal. 2009; (10) Goddietal. 2011;
Since the observedwidth ofthe H53αline is muchless
(11) Plambecketal. 1995; (12) Friedel&WidicusWeaver
2011 (assuming 10% absolute calibration accuracy); (13) this than 200 km s−1, most of the H53α emission that is de-
paper;(14)Beuther etal.2004;(15)Beuther etal.2006 tected must originatemore than ∼12 AU fromthe star,
inthe lowerdensityhalowheren <107 cm−3. The line
e
a measure of the flux of ionizing photons from the cen- isslightlyblueshiftedrelativetothe othertransitions,as
tral star. This corresponds to the excitation parameter expected if this gas is expanding. The H30α line is little
expectedforamainsequenceBstarwithLymancontin- affectedbypressurebroadeningorbycontinuumopacity,
uumflux5×1045photonss−1andtotalluminosity(0.5− so its central velocity, 23.2 km s−1, is the best indicator
1)×104 L⊙ (Panagia1973). The ionizing flux is compa- oftheVLSR ofBN.Weattributetheanomalouslynarrow
rable to the value derived by Scoville et al. (1983) from widthobservedfortheH41/42αtransitiontopoorsignal
infrared data, and the luminosity is comparable with to noise; more sensitive spectra of all these recombina-
the bolometric luminosity obtained by De Buizer et al. tionlines wouldbe useful infurther constrainingmodels
(2012) from 6–37 µm SOFIA observations. of the electron density in BN.
We do not detect the 231.995 GHz helium recombi-
4.1.3. Recombination lines nation line toward BN. The absence of this line is con-
sistent with the identification of BN as a B-star. In the
30kms−1resolutionBNspectruminFigure1,the2σup-
per limit on the intensity of the He30α line is ∼12 mJy,
while the intensity of the H30α line is 270 mJy; thus
τ(He30α)/τ(H30α)<0.045. If helium were ionized over
thefullvolumeoftheHIIregion,theopacityratiowould
be comparable to the He/H abundance ratio, ∼ 0.1.
Thus we can say that the He+ zone, if it exists, fills less
thanhalfthevolumeoftheHIIregion. Thisplacesanup-
perlimitofroughly0.05ontheratioofhelium-ionizingto
hydrogen-ionizing photons from the star (Mezger et al.
1974, Fig. 2), consistent with the ratio expected for B-
stars,butnotforO9ormoremassivestars(Mezger et al.
1974; Vacca et al. 1996).
4.2. SrcI
According to the data in Figure 3, the flux density
Figure 4. Hydrogen recombination lines observed toward BN. of SrcI scales approximately as ν2 from 43 to 229 GHz,
H30αandH41/42α(theaverageofthetwospectra)wereobserved
suggesting that over this frequency range the source is
withCARMA;H53α, withtheVLA(Rodr´ıguezetal.2009). The
H53αfluxdensitiesarescaledupby4forbettervisibility. Smooth anopticallythickblackbodywithconstantangularsize.
curvesshowGaussianfitstothespectra. Even if the absolute calibrations are wrong, the absence
ofadetectableH30αlinerulesoutthepossibilitythatthe
6 Plambeck et al.
diameter of 15AU is, however,inconsistentwith the ob-
Table 3 servedsizeofSrcI,whichisroughly0.2′′,or80AU,along
Recombination LineParametersforBN
its major axis. This suggests that the emission region is
line freq Speaka VLSR FWHM clumpy or filamentaryso that it does not fill the synthe-
(GHz) (mJy) (kms−1) (kms−1) sizedbeam. Butitisimplausiblethatalltheseclumpsor
filaments are perfectly sharp-edged, such that every one
of them is optically thick even at 229 GHz. Generally,
H53αb 42.952 10.4±1.1 20.1±2.1 39.0±4.9
oneexpectsadistributionofclumpstoproduceemission
HH4412αα 9825..063848 47±5c 24.2±1.1c 20.0±2.7c with a shallower spectral slope (Ignace & Churchwell
H30α 231.901 291±10 23.2±0.5 31.8±1.3 2004).
Onealternativeisthattheelectrontemperatureisless
a Peakfluxdensityofthelineabovethecontinuum level. than 8000 K. The highest resolution measurements of
b Rodr´ıguezetal.2009. SrcI that are available, 43 GHz VLA observations with
c fittoaverageoftheH41αandH42αlineprofiles.
a 34 mas synthesized beam, imply that the source has a
emission is optically thin p+/e− bremsstrahlung, unless brightnesstemperatureofonly1500K(Reid et al.2007).
theH30αlineisunexpectedlybroad(>300kms−1). For This is much cooler than is possible in an HII region,
however; hydrogen is almost entirely neutral in dense
alinewidthof30kms−1,comparabletothevelocityspan
gas at T<4000 K (Reid & Menten 1997).
of the SrcI SiO masers, the line to continuum opacity
ratio τ /τ ∼3 from equations (1) and (2), and the line
shouldLeasCily have been detected if τ <2. 4.2.2. H− free-free emission from a massive disk?
C
Ourdatacannoteasilybereconciledwiththe348GHz
The absence of hydrogen recombination lines, 1500 K
flux density of 320±48 mJy measured with the SMA
apparent brightness temperature, and spectral index of
(Beuther et al.2004),whichfallsbelowplausibleextrap- ∼2 all are difficult to explain if SrcI is a hypercom-
olations from the lower frequency points. Such a low
pact HII region. The data more naturally fit an alter-
348GHzfluxdensitywouldimplythatthefree-freeemis-
native hypothesis, first considered by Reid et al. (2007),
sionfromSrcIis somewhatopticallythinat229GHz, in
that the emission from SrcI originates in a T< 4500 K
which case we should have detected the H30α line. Pos-
disk, similar to the radio photospheres of Mira vari-
sibly the 348 GHz SrcI measurement was corrupted by
ables (Reid & Menten 1997). The infrared spectrum of
negativesidelobes fromthe brightHotCoreclumps 1-2′′
SrcI,seen via light reflected off the neighboring nebulos-
east of SrcI, or possibly SrcI is time-variable (cf. Sec-
ity, suggests such an interpretation as well – Testi et al.
tion4.2.3). The690GHzfluxdensitymeasuredwiththe
(2010) find that the spectrum is characteristic of a low-
SMA, 6.7±3.2 Jy (Beuther et al. 2006), lies above the
gravity photosphere with effective temperature 3500–
ν2 extrapolationfromlowerfrequencies,butthisiseasily 4500 K. The close association of SrcI with SiO masers,
explained if there is thermal emission from dust within also characteristic of Mira variables, further strengthens
the synthesized beam. the case for this hypothesis.
As discussedby Reid & Menten (1997), free-freeemis-
4.2.1. An optically thick hypercompact HII region?
sion at temperatures of about 1500 K in such a photo-
If SrcI is a conventional hypercompact HII region, sphere arises via the H− opacity, the scattering of elec-
it must have an extraordinarily high emission measure, tronsbyneutralhydrogenatomsormolecules. Theelec-
> 3×1011 pc cm−6, in order to remain optically thick trons are produced by collisional ionization of Na, K,
at229GHz. We areable reproducethe observedcontin- and other metals. The H− opacity is roughly 103 times
uumfluxdensitieswiththesimplesphericallysymmetric smallerthanthep+/e−opacity,sohighdensities(>1011
model described in section 4.1.1 if we assume that the cm−3) typically are required to obtain optically thick
ionized region has radius r = 7.5 AU, electron tem- emission.
c
perature T = 8000 K, and uniform electron density Reid et al.(2007)wereabletofitthe43GHzVLAob-
e
n = 1.5 × 108 cm−3. The resulting continuum spec- servations of SrcI with models of H− emission from an
e
trumis shownby the solidcurveinFigure 3; the dashed 80 AU diameter, 3 M⊙ disk. However, they found that
curve shows the recombination line intensities predicted such a large disk cannot be maintained at T ≥ 1500 K
by the model. The HII region is constrained to have if it is heated solely by a central star – a stellar lumi-
a sharp outer edge – any halo of lower density plasma nosity > 105 L⊙, greater than the luminosity of the
would lead to excess emission at cm wavelengths. Col- entire Orion-KL region, would be required. They pro-
lisional broadening of the H30α line is predicted to be posed that accretion processes might heat the disk lo-
of order 10 km s−1, not enough to smear out the line. cally. Testi et al. (2010) also find that the infraredspec-
The excitation parameter U ∼ 10 pc cm−2 computed trum of SrcI can be explained by emission from a disk
from the radius and electron density corresponds to an around a ∼ 10 M⊙ protostar that is accreting at a
ionizing flux of 4×1046 photons s−1, consistent with a few × 10−3 M⊙ yr−1. However, their models show that
B0 to B1 main sequence star with luminosity ∼ 104 L⊙ thedisktemperaturefallsbelow1500Kforr>10AU,so
(Panagia 1973; Smith et al. 2002). Since the continuum cannot easily explain the H− free-free emission at larger
emission is optically thick, this should be interpreted as radii.
a lower limit to the ionizing flux. In summary, the H− hypothesis explains the observa-
The source diameter of 15 AU used in the model is tional results in a natural way. The principal barrier
dictatedbytheassumedelectrontemperatureof8000K, to its acceptance is uncertainty over the mechanism by
themeasuredfluxdensities,andthedistancetoOrion. A whichsucha largedisk canbe maintainedatsufficiently
Ionized Envelopes of Orion SrcI and BN 7
line also was recognized by Hirota et al. (2012) in the
Orion Band 6 ALMA Science Verification data. With
0.6 km s−1 velocity resolution, the ALMA data show
thattheH Olineisdouble-peaked,likethe22GHzH O
2 2
masers or the 43 and 86 GHz SiO masers toward SrcI.
In the 1.5′′ resolution ALMA maps the H O line
2
is blended with a HCOOCH transition at 232.684
3
GHz, whereas in the 0.25′′ resolution CARMA map the
HCOOCH emission is resolved out. The H O emission
3 2
isunresolvedbytheCARMAbeam,soitisclearthatthe
lineoriginatesfromthecompactSiOmaserzonecloseto
SrcI, rather than the extended 2′′×0.5′′ strip where 22
GHz H O masers are found (Gaume et al. 1998). The
2
SiO v=2 masers also originate in energy levels 3500 K
above the ground state; Goddi et al. (2009) find that
Figure 5. SrcI/BNfluxdensityratiovs. year,at43GHz(filled
these masers can be excited in gas with kinetic temper-
squares)andat86GHz(circles),derivedfromthedatainTable2.
This ratio should be immune to errors in the absolute flux scale, ature >2000 K.
since the SrcI and BN are just 10′′ apart on the sky and are ob- The same transition of vibrationally excited H O was
2
served simultaneously. These data provide possible evidence that detected toward the evolved stars VY CMa and W Hya
SrcIhasbrightenedrelativetoBNsince1995.
by Menten & Melnick (1989). Like SrcI, both of these
stars have SiO and 22 GHz H O masers.
high temperature. Whether the disk could be heated by 2
an anomalously high accretion rate, a merging binary, 5. SUMMARY
magneticeffects,orsomeothermechanismremainsakey CARMA229GHz maps ofOrion-KLwith upto 0.14′′
question. angular resolution clearly distinguish the ionized en-
velopes of SrcI and BN from the surrounding molecular
4.2.3. Is SrcI brightening? cloud. The principal results are as follows:
IfaccretionisresponsibleformuchofSrcI’sluminosity,
1. TheintegratedfluxdensitiesofSrcIandBNat229
thenonemightexpecttoseechangesinitsfluxdensityon
GHzare310and240mJyrespectively,withaprob-
timescales of years. Although Goddi et al. (2011) found ableuncertaintyof±15%. The BNflux densitywe
no evidence of flux variations at 43 GHz over 4 epochs measure with 0.14′′ resolution is a factor of two
from 2000 to 2009, we are struck by the discrepancy in
greater than that derived by Galv´an-Madrid et al.
the86GHzfluxesmeasuredatBIMAin1995(34±5mJy,
(2012)fromALMAScienceVerificationdatawitha
Plambeck et al. 1995) and at CARMA in 2009 (50±5 1.3′′×0.6′′ synthesizedbeam. We suspectthat the
mJy, Friedel & Widicus Weaver 2011).
flux density of BN is depressed in the ALMA map
Generally, uncertainties in absolute flux densities are
because of negative sidelobes from poorly sampled
dominatedbysystematiceffects–e.g.,variationsintele-
extendedemission. Theselargescalestructuresare
scope gains as a function of elevation, or uncertainties
filteredoutmoreeffectivelyinthehigherresolution
in the brightness temperature models of planets used as
CARMA data.
primaryfluxstandards. Theratioofthe fluxdensitiesof
SrcI and BN should be immune to such effects because 2. SrcI’s flux density is proportional to ν2 from 43 to
the two sources are just 10′′ apart and can be observed 229GHz–itappearstobeanopticallythickblack
simultaneously. In Figure 5 we plot this ratio as a func- body. By comparison, BN’s flux density scales as
tionoftimeforthe43and86GHzdatainTable2. Note ν1.3uptoabout100GHz,thenflattens;itsfree-free
that the error bars in this figure were calculated from continuum is becoming optically thin at 229 GHz.
the thermal noise in the originalmaps, not from the ab-
3. SrcIiselongatedatPA140degreesintheCARMA
solute flux uncertainties listed in Table 2. From these 7
images. Its deconvolved size is similar at 229 and
independent datasets it appears that the ratio increased
43GHz–furtherevidencethatitisopticallythick.
from about 0.4 in 1995 to 0.5 in 2009.
This result should be interpreted cautiously. Al- 4. The H30α recombination line is not detected to-
though the flux density ratio is unaffected by the ab- ward SrcI, though it is bright toward BN. The ab-
solute calibration scale, it is sensitive to the sampling sence of this recombination line is a third piece of
of visibility data in the (u,v) plane. It is also unclear evidence that SrcI is optically thick, or that it is
whether SrcI or BN, or both, are varying; periodic vari- not an HII region at all.
ationsinthe infraredluminosityofBNwerereportedby
5. The v =1, 5(5,0)-6(4,3) transition of H O at
Hillenbrand et al. (2001). Future careful monitoring of 2 2
232.687 GHz, 3450 K above the ground state, is
the flux densities of both SrcI and BN with the VLA
prominent toward SrcI. The line emission origi-
and ALMA will be required to find conclusive evidence
nates close to the central star, not in the larger
for source brightness variations.
zone where 22 GHz H O masers are found.
2
4.2.4. Vibrationally excited H O
2 6. While ourobservationalresultsdo notrule outthe
The bright emission line in the spectrum of SrcI in possibility that SrcI is an exceptionally dense hy-
Figure1istheH Ov =1,5(5,0)-6(4,3)transitionofH O percompact HII region, they are more easily ex-
2 2 2
at 232.687 GHz, 3450 K above the ground state. This plained by models of free-free emission via the
8 Plambeck et al.
H− opacity in a T< 4500 K disk, as modeled by ence Foundation under a cooperative agreement, and by
Reid et al. (2007). the CARMA partner universities.
Facilities: CARMA.
7. A model of a spherically symmetric hypercompact
HII region with a diameter of about 15 AU and REFERENCES
electron density n ∼ 5×107 cm−3 matches both
e
the continuum and the recombination line intensi- Allen,D.A.,&Burton,M.G.1993, Nature,363,54
ties toward BN. Bally,J.,Cunningham,N.J.,Moeckel,N.,Burton,M.G.,Smith,
N.,Frank,A.,&Nordlund,A.2011, ApJ,727,113
Beuther,H.,etal.2004,ApJ,616,L31
8. The excitation parameter U ∼ 5 pc cm−2 derived —.2006, ApJ,636,323
for BN implies that its Lyman continuum flux is Brocklehurst,M.,&Seaton, M.J.1972,MNRAS,157,179
5×1045photonss−1,consistentwiththatexpected Chandler,C.J.,&Wood,D.O.S.1997, MNRAS,287,445
from a zero age main sequence B star with a total DeBuizer,J.M.,Morris,M.R.,Becklin,E.E.,Zinnecker, H.,
luminosity of (0.5−1)×104 L⊙. H20e1r2te,rA,pTJ.,L7.4,9A,dLa2m3s,J.D.,Shuping,R.Y.,&Vacca, W.D.
Doi,T.,O’Dell,C.R.,&Hartigan,P.2002,AJ,124,445
9. The H53α recombination line in BN is predicted Eisner,J.A.,Plambeck,R.L.,Carpenter,J.M.,Corder,S.A.,
to have a collisionally broadened linewidth of Qi,C.,&Wilner,D.2008,ApJ,683,304
200 km s−1. Presumably the H53α emission ob- Felli,M.,Churchwell,E.,Wilson,T.L.,&Taylor,G.B.1993a,
A&AS,98,137
servedbyRodr´ıguez et al.(2009),withalinewidth
Felli,M.,Taylor,G.B.,Catarzi,M.,Churchwell,E.,&Kurtz,S.
ofabout40kms−1,originatesprimarilyfromlower 1993b, A&AS,101,127
density ionized gas farther from the central star. Forbrich,J.,Menten,K.M.,&Reid,M.J.2008,A&A,477,267
Friedel,D.N.,&WidicusWeaver, S.L.2011, ApJ,742,64
10. The V of BN is 23.2±0.5 km s−1, based on Galva´n-Madrid,R.,Goddi,C.,&Rodr´ıguez,L.F.2012, A&A,
LSR 547,L3
the central velocity of the 231.9 GHz H30α line, Garay,G.,Moran,J.M.,&Reid,M.J.1987,ApJ,314,535
which is little affected by pressure broadening or Gaume,R.A.,Wilson,T.L.,Vrba,F.J.,Johnston,K.J.,&
continuum opacity. Schmid-Burgk,J.1998,ApJ,493,940
Goddi,C.,Greenhill,L.J.,Chandler,C.J.,Humphreys,
E.M.L.,Matthews,L.D.,&Gray,M.D.2009,ApJ,698,1165
11. Publishedfluxdensitiesatboth43and86GHzpro-
Goddi,C.,Humphreys,E.M.L.,Greenhill,L.J.,Chandler,
vide tentative evidence that SrcI brightened with C.J.,&Matthews,L.D.2011,ApJ,728,15
respect to BN between 1994 and 2009. G´omez,L.,Rodr´ıguez,L.F.,Loinard,L.,Lizano,S.,Allen,C.,
Poveda, A.,&Menten,K.M.2008,ApJ,685,333
Hillenbrand,L.A.,Carpenter,J.M.,&Skrutskie,M.F.2001,
Future ALMA observations of Orion-KL at submil-
ApJ,547,L53
limeter wavelengths will provide a definitive test of the Hirota,T.,Kim,M.K.,&Honma,M.2012,ApJ,757,L1
emission mechanism in SrcI. If SrcI is an HII region, it Ignace, R.,&Churchwell,E.2004,ApJ,610,351
must become optically thin at sufficiently high frequen- Kim,M.K.,etal.2008,PASJ,60,991
Matthews,L.D.,Greenhill,L.J.,Goddi,C.,Chandler,C.J.,
cies. If the source were still optically thick at 650 GHz,
Humphreys,E.M.L.,&Kunz,M.W.2010,ApJ,708,80
for example, an O-star with a Lyman continuum flux
Menten,K.M.,&Melnick,G.J.1989,ApJ,341,L91
> 1049 photons s−1 would be required to maintain its Menten,K.M.,&Reid,M.J.1995,ApJ,445,L157
ionization, but the luminosity of such a star, > 105 L⊙ Menten,K.M.,Reid,M.J.,Forbrich,J.,&Brunthaler,A.2007,
(Panagia1973),wouldexceedthemeasuredluminosityof A&A,474,515
Mezger,P.G.,Smith,L.F.,&Churchwell,E.1974,A&A,32,269
the entire Orion-KL nebula. Then, at frequencies where
Morino,J.-I.,Yamashita,T.,Hasegawa, T.,&Nakano,T.1998,
the continuumis optically thin, hydrogenrecombination
Nature,393,340
lines must be detectable if hydrogen is ionized. If these Panagia,N.1973,AJ,78,929
recombinationlinesaredetected,thefree-freehypothesis P´erez,L.M.,etal.2010,ApJ,724,493
isconfirmed. Ifnot,orifthecontinuumfluxdensitycon- Plambeck,R.L.,etal.2009,ApJ,704,L25
tinues to increase following a ν2 law, the emission must Plambeck,R.L.,Wright,M.C.H.,Mundy,L.G.,&Looney,
L.W.1995,ApJ,455,L189+
be attributable to the H− mechanism.
Reid,M.J.,&Menten,K.M.1997,ApJ,476,327
SubmillimeterobservationsofBN,aswellasmoresen- Reid,M.J.,Menten,K.M.,Greenhill,L.J.,&Chandler,C.J.
sitive observations of its cm and mm wavelength recom- 2007,ApJ,664,950
bination lines, will be useful in constraining models of Rodr´ıguez,L.F.,Poveda, A.,Lizano,S.,&Allen,C.2005,ApJ,
627,L65
the hypercompact HII region around this star, and in
Rodr´ıguez,L.F.,Zapata, L.A.,&Ho,P.T.P.2009,ApJ,692,
tracking small changes in the SrcI/BN flux density ra-
162
tiothatcouldbe anindicationoftime variableaccretion Rohlfs,K.,&Wilson,T.L.2000, ToolsofRadioAstronomy(3rd
onto these objects. edition;NewYork:Springer)
Schaller,G.,Schaerer,D.,Meynet,G.,&Maeder,A.1992,
A&AS,96,269
Scoville,N.,Kleinmann,S.G.,Hall,D.N.B.,&Ridgway,S.T.
Support for CARMA construction was derived from
1983,ApJ,275,201
the states of California, Illinois, and Maryland, the
Smith,L.J.,Norris,R.P.F.,&Crowther,P.A.2002,MNRAS,
James S. McDonnell Foundation, the Gordon and Betty 337,1309
Moore Foundation, the Kenneth T. andEileen L.Norris Testi,L.,Tan,J.C.,&Palla,F.2010,A&A,522,A44
Foundation, the University of Chicago, the Associates Vacca,W.D.,Garmany,C.D.,&Shull,J.M.1996,ApJ,460,914
Wright,A.E.,&Barlow,M.J.1975, MNRAS,170,41
of the California Institute of Technology, and the Na-
Zapata,L.A.,Loinard,L.,Schmid-Burgk,J.,Rodr´ıguez,L.F.,
tional Science Foundation. Ongoing CARMA develop-
Ho,P.T.P.,&Patel,N.A.2011,ApJ,726,L12
ment and operations are supported by the National Sci-