Radio Recombination Lines toward the Galactic Center Lobe

0.49 MB·English

by  C. J. Law

Draftversion January23,2009
PreprinttypesetusingLATEXstyleemulateapjv.10/09/06
RADIO RECOMBINATION LINES TOWARD THE GALACTIC CENTER LOBE
C. J. Law1,2,3, D. Backer3, F. Yusef-Zadeh1, and R. Maddalena4
Draft version January 23, 2009
ABSTRACT
TheGalacticCenterlobeisadegree-tallshellseeninradiocontinuumimagesoftheGalacticcenter
(GC)region. IfitisactuallylocatedintheGCregion,formationmodelswouldrequiremassiveenergy
9
input (e.g., starburst or jet) to create it. At present, observations have not strongly constrained the
0
location or physicalconditions of the GC lobe. This paper describes the analysis of new and archival
0
single-dish observations of radio recombination lines toward this enigmatic object. The observations
2
find that the ionized gas has a morphology similar to the radio continuum emission, suggesting that
n
they are associated. We study averagesof severaltransitions from H106α to H191ǫ and find that the
a
lineratiosaremostconsistentwithgasinlocalthermodynamicequilibrium. Theradiorecombination
J
line widths are remarkably narrow, constraining the typical electron temperature to be less than
1
about 4000 K. These observations also find evidence of pressure broadening in the higher electronic
1 states, implying a gas density of n = 910+310 cm−3. The electron temperature, gas pressure, and
e −450
morphology are all consistent with the idea that the GC lobe is located in the GC region. If so, the
]
A ionized gas appears to form a shell surrounding the central 100 parsecs of the galaxy with a mass of
roughly 105 M , similar to ionized outflows seen in dwarf starbursts.
G ⊙
Subject headings: Galaxy: center — radio line: general
.
h
p
- 1. INTRODUCTION The GCL was first noted in a 10 GHz radio contin-
o uum survey of the central few degrees of the Galaxy
Sincetheirpredictionanddiscoverymorethan40years
r (Sofue & Handa 1984). The survey found a degree-tall
t ago (Kardashev 1959; Dravskikh & Dravskikh 1964;
s loopofemissionnorthofthe galacticplane,surrounding
a Sorochenko & Borozich1964), radio recombinationlines thecentraldegree,coveringanareafroml=359◦.2−0◦.2
[ have been useful tools for probing the physical condi-
and b = 0◦.2 − 1◦.2. The resemblance of the GCL
tionsofthe ISM.The propertiesofthermalemissioncan
1 to a wind-driven mass outflow was striking and led to
constrain fundamental parameters such as gas tempera-
v several models for its formation (e.g., Chevalier 1992;
tureanddensity(e.g.,Roelfsema & Goss1992). Further-
0 Melia & Falcke 2001). However, while the first study of
more, the spectral line can provide valuable information
8 the radio spectral index found it to be consistent with
about the kinematics of the emitting gas.
4 thermal emission (Sofue 1985), later work suggested the
In the Galactic Center (GC) region, there have been
1
GCL was nonthermal (Tsuboi et al. 1986; Reich et al.
. extensive observations of radio recombination lines.
1 1987). Bland-Hawthorn & Cohen(2003) noted the mor-
These lines are useful because they are detectable de-
0 phological connection between the radio continuum and
spite heavy extinction and can be used to separate
9 mid-IRemissionandproposedastarburstoutflowmodel.
thermal H II regions from the widespread nonthermal
0 However,theconnectionoftheionizedgastothisfeature
gas (Pauls & Mezger 1975). Separating thermal and
: is still unclear.
v nonthermal emission is especially important in the GC
i region, which is filled with extended ionized gas and Motivated by a desire to understand the nature of the
X emission in the GCL, we have conducted new radio re-
synchrotron-emitting features (e.g., Yusef-Zadeh et al.
r 1984;Morris & Serabyn1996;Lang et al.1997). Recom- combinationline observationsandcollectedarchivedob-
a servations in this region. This paper describes observa-
bination line surveys have found that diffuse ionized gas
tions of the GCL with the Robert C. Byrd 100-mGreen
isstronglycorrelatedwiththemoleculargasdensityand
Bank Telescope5 (GBT) and the Hat Creek Radio Ob-
star formation, suggesting that the GC regionshould be
servatory 26-m telescope (HCRO). Section 2 shows how
filled with ionized gas (Lockman et al. 1996). This gas
theobservationsweremadeandcalibrated. Section3de-
can provide a direct probe of the conditions in the GC
scribestheresultsoftheobservations,includingthemor-
region.
phology, line characteristics, and kinematics of the line
The information available by studying ionized gas
emission in the region. The HCRO observations show
could help address questions surrounding a feature in
that the H109α emission has a morphology similar to
theGCregionknownastheGalacticcenterlobe(GCL).
the radio continuum emission. The GBT observations
1Department of Physics and Astronomy, Northwestern Uni- find that the recombination line emission is this region
versity, Evanston, IL 60208, USA; [email protected], is in local thermodynamic equilibrium (LTE). The un-
[email protected] usually narrow recombination line emission throughout
2AstronomicalInstitute“AntonPannekoek”,UniversityofAm-
the GCL places useful constraints on the gas tempera-
sterdam,Kruislaan403,1098SJAmsterdam,TheNetherlands
3RadioAstronomy Lab, Universityof California,Berkeley, CA
94720,USA;[email protected] 5 The National Radio Astronomy Observatory is a facility of
4National Radio Astronomy Observatory, Green Bank, WV, theNationalScienceFoundationoperatedundercooperativeagree-
24944,USA;[email protected] mentbyAssociatedUniversities,Inc.
2 Law et al.
ture and other properties. In § 4, we discuss the derived The spectrometer was configured with four windows
quantities of the ionized gas, including gas density, tem- of width 200 MHz observing eight Hα and Heα tran-
perature, and the total mass (assuming a geometry and sitions from n =106 to 113. Each pointing produced
distance). Section 5 summarizes the results of this pa- dual-circular polarizations of four 8196 channel spectra.
per. A detailed discussion of the nature of the GCL is Thespectralwindows,centerednear5.37,5.08,4.81,and
left for a later paper (Law, in prep). 4.56 GHz, also covered other transitions, including: Hβ
(7 transitions with n =134–142), Hγ (8; n =152–162),
2. OBSERVATIONSANDDATAREDUCTIONS
Hδ (8; n=167–178),and Hǫ (8; n=180–191). The spec-
2.1. HCRO tra had 24.4 kHz channels, equivalent to 1.5 km s−1 at
In1985,weusedtheHCRO26mtelescopetomapthe 4.9 GHz. All velocities in this work are given in the
H109α recombination line toward the GCL. At 5 GHz, local standard of rest (LSR). Calibration for each scan
the HCRO beam size is 10′ and its Nyquist sample size wasdonebypositionswitchingtoapositiontwodegrees
is 3′.3. Pointings were made over a roughly 1◦ square north in Galactic latitude.
region on a 6′ grid; neighboring points in the grid were Calibrationandanalysis wasdone using tools released
skipped to form a checkerboard pattern. Spectra were withGBTIDLv1.2.1andcustomIDLprograms. Spectra
calibrated by position switching to a distant, fixed posi- were calibrated using the standard (on-off)/off method.
tion, with a typical integration time of 2 hours for each The typical line analysis would apply standard calibra-
spectrum. Severalscatteredpointsoutsidetheradiocon- tion, average both polarizations and lines of the same
tinuum emission associated with the GCL were also ob- transition (e.g., all Hβ transitions; rest frequencies from
served to constrain the background emission. A crude Lilley & Palmer (1968)), subtract a third order polyno-
gain calibration was made by comparing the brightness mial baseline fit, and fit a Gaussian line profile. It is
temperaturesbetweentheHCROandGBTobservations worth noting that the standard error propagation used
described below; this comparison implies a gain of ∼ 13 does not accountfor the uncertainty in fitting a baseline
Jy K−1. to the spectra, and are likely to underestimate the true
The spectrometer at HCRO was tuned to a central errors slightly. By fitting with various assumptions, we
frequency of 5008.923 MHz, 512 channels, and a 20 estimatedthebaseline-induceduncertaintyinthelinefit
MHz bandwith. This bandwidth covered the H109α is always less than the standard statistical errors in the
and H137β transitions, although only the H109α tran- line fit and generally decreases with signal to noise ratio
sitionwas studied. Spectra were Hanning smoothedand of the line.
a seventh-order baseline fit before Gaussian line fitting. All spectra presented here are corrected for the beam
The velocity uncertainty of the fits are about 1 km s−1. efficiency andatmospheric opacity,andthus are inunits
The system temperature wasassumedto be 40 K,based of main beam brightness temperature. The beam effi-
on observations of empty sky. ciency is assumed to be equal to 89%. 6 This work
Figure1showsimagesofthe H109αline antennatem- corrects for mean atmospheric absorption opacity of
perature and velocity from the HCRO data. The line 8.6×10−3, giving a correction factor of 2–5% for these
brightness is not corrected for atmospheric absorption. observations.
Since the checkerboard observing pattern makes visual- The GC region is seen at low elevations from Green
izationawkward,thedatawereinterpolatedontoaregu- Bank, which makes the calibration of the continuum
lar grid. The algorithm calculates a linear interpolation levels difficult. Unfortunately, the standard continuum
for missing pixels with four neighbors, then three, then calibration method could not properly account for at-
two. The order in which this interpolation is done cre- mospheric emission. Since the observations described in
atessomebiasinthe interpolatedpixels,butthe general Law et al. (2008) were made with a calibration method
properties of the emission on spatial scales larger than optimized for the study of continuum levels, we use re-
the 10′-beam are not affected. sults from that observation when continuum fluxes are
needed.
2.2. GBT
3. RESULTS
The goal of the GBT observation was to confirm the
resultsoftheHCROobservationsandstudythegascon- 3.1. HCRO
ditionsinmoredetail. WeusedtheGBTSpectrometerin Figure 1 shows the HCRO line intensity and velocity
a 7 hour sessionin August 2005. The GBT observations compared to 5 GHz continuum emission. The radio re-
were more limited in spatial coverage than the HCRO combination line is brightest along two vertical ridges
survey;thelocationsareshowninFigure2anddescribed near (l,b)=(0◦.0, 0◦.5) and (359◦.4, 0◦.5). There is also
in Table 1. The locations named “GCL3” and “GCL4” a slight increase in brightness at the top of the line-
areatthe twobrightpeaksinthe HCROmapofH109α. emitting region, near (l,b)=(359◦.6, 1◦.0), like a cap to
These regions were mapped in a 3×3, Nyquist-sampled the structure. The longitude of the line emission north
(1′ near5GHz)grid,withon-sourceintegrationtimesof of the plane is centered near l = 359◦.6 and the central
90sperpointing. Diagonalstripswereobservedstarting longitude shifts slightly to the west at higher latitudes.
atthese positionsandgoingtothe Galacticnortheastto The overall morphology of bright limbs with a cap mir-
study gradients in the velocity field. A sparse horizontal rors that of the radio continuum emission.
strip was observed at b = 0◦.45, to sample the changes The right side of Figure 1 shows the best-fit velocity
in the east-west direction. Finally, two positions outside of the H109α line. Two striking characteristics of the
the GCL, “GCL1” and “GCL7”, were observed to mea- line velocities in this region are their small values and
surebackgroundlineemission. Thestripsandindividual
pointings had integration times of 60 s per pointing. 6Seehttp://wwwlocal.gb.nrao.edu/gbtprops/man/GBTpg/GBTpg tf.html.
RRLs toward GCL 3
the absence of any simple pattern or gradient. North line emission has the narrowest widths and vice versa.
of b = 0◦.3 the line velocity does not exceed the range The top panel shows the line is brightest near l = 0◦
of -5 to +5 km s−1. This is consistent with previ- and l = −0◦.7; the bottom panel shows that the line
ous observations of radio recombination lines in the re- is narrowest at these positions. Since the line width is
gion (Lockman & Gordon 1973; Pauls & Mezger 1975; a useful constraint to the gas properties, we needed to
Anish Roshi & Anantharamaiah 1997). understand the origin of this correlation.
There is significant emission at every HCRO pointing Twopotentialexplanationsforthecorrelationarethat
between the two ridges of emission. At the lowest, the (1) the line width is intrinsically tied to its brightness,
linebrightnessrangesfrom10–30mK,whilethebrightest or (2) the line has two components that can affect the
line emissionis 72 mK, at the peak of the westernridge. best-fit line width according to their relative strengths.
Theemissionoutsideoftheridgesofemissionisgenerally Figure 5 shows the average H106−113α profile for four
less than 10 mK. Detailed analysis of the gas properties scans located in the center of the GCL that, separately,
depends on parameters that are not well constrained by have best-fit line widths of 20–30 km s−1. The aver-
theHCROdata. TheGBTresultsaremoreconstraining age spectrum shows a slightly more complex line pro-
andareconsistentwiththe HCROresults,sothe results filethanseennearthebrightestrecombinationlineemis-
are given below in § 3.2.3. sion. IftheprofileisfitwithtwoGaussians,theirbest-fit
components are (T ,v,∆v) = (22±3 mK,5.2±0.5 km
3.2. GBT s−1,9.3±1.5 km sl−1) and (25±2 mK,−5.0±1.3 km
3.2.1. Morphology s−1,30.6±1.9kms−1). Thenarrowcomponentissimilar
Figure 3 shows plots of the average H106−113α line to that observed near the bright ridges of line emission,
brightness for the three strips of observations. The hor- but its amplitude is comparable to the wide component,
izontal strip at b = 0◦.45 is similar to the HCRO data, such that a single-Gaussian fit is moderately wide. In
with two recombination line peaks near the radio con- averages of other scans, there is a tendency for a simi-
tinuum peaks. The eastern peak of the line emission is lar, wide (20–30 km s−1), low-level (10–20 mK) line to
offset west the continuum peak by roughly 0◦.15. The appear. The amplitude of this line is larger than the
western peak of the line and continuum emission fall at ∼ 10 mK seen in the background outside the GCL, so
thesameposition(l ∼359.35−359.4),withinthe5′ spa- the wide line seems to be associated with the GCL. Re-
tial sampling of the line observations. This is consistent gardlessofitsorigin,itseemsthatthechangesinbest-fit,
with the morphology observed by the HCRO and again single-Gaussian line width shown in Figure 4 are likely
showshowthestructureissimilartotheradiocontinuum due to the relative strength of the narrow line and this
emission. widercomponent. Thus,thereseemstobenarrow(9−14
As with the HCRO observations, the line brightness kms−1) recombinationline emissionfor alllines ofsight
between the two emission peaks is significantly brighter through the GCL.
than outside of it. The GCL1 and GCL7 pointings sam- Figure6showsanalternativewayofvisualizingthere-
pled the emission toward lines of sight outside the ra- combination line velocity structure using l–v diagrams.
dio continuum emission of the GCL. The best-fit Gaus- Three strips of observations show the change in the line
sian to the average H106−113α line in the GCL1 po- velocity as a function of Galactic longitude. These plots
sition had (T ,v,∆v) = (0.0103±0.0016 K,−9.5±2.2 show clearly how the brightest emissionis alwayswithin
km s−1,29.2±l 5.2 km s−1). No H106−113α line was 20 km s−1 the rest velocity, and that the lines are typ-
detectedtowardGCL7witha1σ upperlimitof5.9mK. ically about 10 km s−1 wide. The plots also help show
The line brightness for the weakestlines associatedwith how the line profile has multiple components inside the
the GCL have T ∼25 mK. GCL, for l = −0◦.2 to −0◦.45. In particular, the GCL4
TheGCL3andl GCL4diagonalstripsofpointings,like l–v diagram shows two diagonalstructures near l =0◦.4,
theb=0◦.45strip,showadecreaseinpeaklinebrightness whichindicate that there are two narrowcomponents to
astheymoveawayfromthetwolineemissionpeaks. The the line profile with velocities approaching ±10 km s−1
longitudehalf-peak widthofthe eastandwestpeaksare toward the eastern end of the strip.
roughly 0◦.2. To compare the emission at the edges to the center,
an average of the H106−113α spectra for scans at the
3.2.2. Velocity Structure edges of the GCL3 and GCL4 strips was made. Figure
At the brightest parts of the recombination line emis- 7 shows how the averageof three scans from the eastern
sion, the line velocities observedby the GBT are consis- edge of GCL4 confirms the visual impression from the
tent with those of the HCRO. Figure 4 shows how the l–v diagram: thebest-fitspectrumhastwodistinctcom-
average H106−113α line properties change as a func- ponents with (Tl,v,∆v)=(13.8±1.4 mK,13.0±1.6 km
tion of Galactic longitude for the strip at b=0◦.45. One s−1,12.5±1.5 km s−1) and (38.5±1.6 mK,−6.4±0.2
interesting trend in Figure 4 is the east-west asymme- km s−1,8.8±0.4 km s−1). A similar average over the
try in the line velocities, with positive velocities on the four easternmost scans of the GCL3 strip (with l =
east side and negative on the west. Averaging over the 0◦.083−0◦.133) shows two Gaussian components to the
four easternmost and westernmost scans in the b=0◦.45 average profile. The two components have (Tl,v,∆v) =
strip gives spectra with a single Gaussian component (26.9±1.0 mK,7.6±0.3 km s−1,17.2±0.8 km s−1) and
with mean velocities of 2.3±0.2 and −2.4±0.2 km s−1, (11.1±1.4 mK,−18.3±0.5 km s−1,8.6±1.3 km s−1).
respectively. Thisvelocitygradientisconsistentwiththe Bothcomponentsarenarrowandbrighterthantheback-
upper limit provided by the HCRO observations. ground,sotheemissionislikelytobeassociatedwiththe
Comparing the peak line brightness and line widths GCL.
in Figure 4 shows an inverse relationship: the brightest Similar to the HCRO observations, the GBT observa-
4 Law et al.
tions find no emission with velocities |v| > 20 km s−1. the form
Adeepspectrumwasmadebyaveragingobservationsat b T −T β
n e c n
R≈R , (1)
the peak GCL line emission, called GCL3 and GCL4 in LTEb T −T β
m e c m
Table1;thetotalon-sourceintegrationtimeforthisdeep
spectrum was 27 minutes. The rms noise in the baseline where R = n2f /m2f , b and β are
LTE n,n+∆n m,m+∆m
level of the deep GBT spectrum gives an upper limit to the departure coefficients for non-LTE effects, and
the peak line brightness of about 1 mK for v = ±500 T and T are the electron and background contin-
e c
km s−1 and 2 mK for v = ±1500 km s−1The peak line uumbrightnesstemperatures(Dupree & Goldberg1970;
brightnesslimitsare1%and2%ofthepeakatv ∼0km Brocklehurst& Seaton 1972; Roelfsema & Goss 1992).
s−1, respectively. The departure coefficients vary with n , T , and n,
e e
and have been calculated for many transitions un-
der a wide range of gas conditions (Dupree 1972;
3.2.3. Line Widths and Line Ratios Salem & Brocklehurst 1979). To estimate non-LTE ef-
fects, we calculated departure coefficients for T =
e
Tobetter constrainthegasconditions,westudiedsev- 20 − 10000 K and n = 1 − 1000 cm−3. The lowest
e
eral recombination lines in the deep integration toward temperature and density used in the calculations cor-
the peak GCL line emission. The relatively small ve- respond to the conditions that may cause stimulated
locities observed in these integrations makes it possi- emission observed toward the GC region at 1.4 GHz
ble to average them together. The two 3×3, Nyquist- (Lockman & Gordon1973). The LTEline ratiosdepend
sampled patterns near the brightest regions of the GCL ontheoscillatorstrengthsforthetransitions,whichhave
were found to have not only similar line velocities and relative values known to high precision (Dupree 1969).
line widths, but also similar line ratios, suggesting that Table 3 shows the integrated line ratios and the pre-
these regions have similar physical conditions. dictions for LTE and non-LTE conditions. The ratios
Table2liststhelinemeasurementsandFigure8shows show excellent agreement with LTE for all transitions
the highest and lowest significance fits in the deep spec- observed. The predicted line ratios for other models do
trum toward the GCL. A Gaussian line is fit to the av- not agree with the observed line ratios.
erage line profile for all transitions of a given ∆n (e.g., One model, with T = 1000 K and n = 1000 cm−3,
e e
∆n = 1 for “α”), which includes 7 or 8 transitions in predicts line ratios similar to the observed and LTE val-
the range given in Table 2. Generally, these averages ues. The amount of stimulated emission can be calcu-
over several n values are treated like a single transition lated as
withn andν ≈4.95GHz, orthe meanofthe transitions T /TLTE ≈b [1−β (T /T )], (2)
l l n n c e
averaged.
One of the most surprising characteristics of the lines where T is the line brightness, and the backgroundcon-
l
shown in Table 2 is their unusually small widths. The tinuum brightness is T ≈ 1 − 2 K (Law et al. 2008).
c
average H106−113α, H134−142β, H152−162γ, and For T = 1000 K and n = 1000 cm−3, the amplifica-
e e
H167 − 178δ transitions have widths that are simi- tion of the line is only a few percent. A much stronger
lar to each other and have an error-weighted mean of background continuum is required to cause significant
∆v = 13.5 ± 0.2 km s−1. The average H180 − 191ǫ stimulated emission, with 10% amplification for T ∼25
c
line is significantly wider and is discussed in more de- K;this ismore than10times brighterthanthe observed
tail below. Typical H II regions have much larger line continuum in the region. Thus, the line ratios severely
widths, with ∆v ≈ 20 km s−1 (Shaver et al. 1983; constrain the contribution of stimulated emission to the
Afflerbach et al. 1996). A hydrogen recombination line recombination line emission for the brightest emission
width of 13.5±0.2 km s−1 is equivalent to a Doppler observed toward the GCL.
temperature of 3960±120 K (Roelfsema & Goss 1992), Therearetwootherpointsthatsuggestthattherecom-
whichisalsosignificantlylessthanvaluesseenintypical bination line emission is not stimulated. First, if back-
H II regions (TD ≈ 5000−10000K). The Doppler tem- groundradio continuum was stimulating the recombina-
perature represents the combined effects of gas motion tion line emission, the morphology of the line emission
and thermal broadening, so it is an upper limit on the wouldcloselyfollowthe backgroundcontinuumemission
electrontemperature. Thelinewidthisroughlyconstant (Lockman & Gordon1973;Cersosimo & Magnani1990).
for transitions from H106−113α to H167−178δ, which The HCRO and GBT observations show that the peak
representtransitions with mean n ranging from 109.5to lineemission,particularlyintheeasternhalfoftheGCL,
172.5. is significantly offset from the peak continuum emission.
The narrow line widths with central velocities The similar line width, line velocity, and line ratios to-
near 0 km s−1 are highly suggestive of stimu- ward the eastern and western recombination line peaks
lated emission by ionized gas along the line of suggestthattheentirestructureemitsbyasimilarmech-
sight to the GC region (Cersosimo & Magnani 1990; anism. Second,thewidthoftheaverageH180−191ǫline
Anish Roshi & Anantharamaiah 1997). The present ob- is significantly larger than that of the other lines, which
servations can test for the possibility of stimulated line is inconsistent with the idea that it is stimulated. How-
emissionby comparingthe line ratios oftransitions with ever,theH189−191ǫlinerepresentsthehighestnstates,
different n. All transitions are observed with a simi- where stimulated emission is expected to be strongest.
lar beam size, which means that they are probing the These points strengthen the case for an LTE origin for
samevolumeofspaceandcanbemeaningfullycompared the recombination line emission, which implies that the
(Dupree & Goldberg 1970). In this case, the integrated narrow line widths are related to intrinsic properties of
line ratiobetweentransitionsto statesnandm takeson the ionized gas.
RRLs toward GCL 5
UsingtheLTEassumption,thelinepropertiesgiveuse- in the GCL is nonthermal (Law et al. 2008), so the line-
ful constraints on the intrinsic gas properties. As men- to-continuumratioisanupperlimitontheelectrontem-
tionedbefore,thewidthoftheaverageH180−191ǫlineis perature. Forthedeepspectrum,thecontinuumfluxwas
significantlylargerthanthe othertransitions. Widths of estimated from the 5 GHz map described in Law et al.
lines that arenot stimulatedaregenerallydominatedby
(2008) to be T ≈ 0.8 K, which gives T ∆v/T = 2.0
thermal and Doppler broadening, although these effects km s−1, for thec averaged H106−113α linle. Thcis line-
are not expected to change between the different tran-
to-continuum ratio gives to an upper limit on the LTE
sitions observed here. The most common n-dependent
electron temperature of TLTE . 5220 K, which is con-
broadening effect is collisional broadening (a.k.a. “im- e
sistent with the upper limit on T from the narrow line
pact” broadening), in which inelastic electron collisions e
widths. For the largest non-LTE effects that are consis-
preferentiallybroadenlineswithlargen(Lang & Willson
tent with the observed line ratios (n = 100 cm−3 and
1978;Roelfsema & Goss1992). Expressingthetotalline e
width as the quadrature sum of all broadening effects Te = 1000 K), Te = (0.855)0.87TeLTE . 4550 K, but for
gives: higherdensitiesandtemperatures,Teiswithinafewper-
centofTLTE. Alternatively,wecanusetheobservedline
e
∆vtot = 4.31 n 7.4 ne Te −0.1 2+∆v2, wlimiditthtototchoenesltercatirnonthteeemlepcetrraotnurteemgpiveersataulroew. eTrhleimuiptpoenr
km s−1 vuu"∆n “100” 104 cm−3 „104 K« # d the line-to-continuum ratio of Tl∆v/Tc > 2.75 km s−1
t (3) and an upper limit on the thermal continuum of 0.6 K.
where ∆v is the Doppler line width. 7 Thus,atmostabout70%oftheradiocontinuumtoward
d
The strong dependence of collisional broadening on n the recombination line peaks of the GCL is thermal.
canbe usedto constrainn , since ∆v is knownand the Theupperlimittotheelectrontemperatureconstrains
e d
dependence on T is weak. Equation 3 was fit to the theemissionmeasure. Assumingτ ≪1andLTEforthe
e L
distribution of Hydrogen line widths assuming ∆v = averageH106−113αtransition, the continuum emission
d
13.5 km s−1, which is the error-weighted mean of the measure is:
H106−113αtoH167−178δlinewidths8 andT =T =
e d EM =8.5×10−3(1+Y+)∆vTT3/2e−Xnpc cm−6, (6)
3960K.Figure9showsthebest-fitlinewidthmodelwith c l e
ne = 950+−234100 cm−3 (1σ). The fit quality is relatively where ∆v is in units of km s−1and e−Xn ≈ 1, for
poor,withχ2/ν =14.4/4,whichsuggeststhattheerrors T & 1000 K (Dupree & Goldberg 1970). Using the av-
e
may be underestimated somewhat. For a more realistic erage H106− 113α line parameters for the deep spec-
fit,wemodeladditionaluncertaintyinthe linewidthsas trum gives EM ≈ 3850(T /3960 K)3/2 pc cm−6. The
c e
equal to the statistical uncertainty for the H180−191ǫ eastern and western peak line brightnesses are some-
line and scaled by the line peak signal to noise ratio; what different, giving EM ≈ 3080(T /3960 K)3/2 and
c e
thisuncertaintyisconsistentwithvariationinthefitline 4570(T /3960 K)3/2 pc cm−6 for the east and west, re-
width with changes in the baseline fit. Adding this line e
spectively.
width uncertainty in quadrature to the statistical error,
the best-fit electron density is ne = 910+−341500 cm−3 with 4. DISCUSSION
a more realistic χ2ν =8.4/4. 4.1. The Structure of the Ionized Gas
The He106−113α line was detected in the deep spec-
The GBT and HCRO observations have found radio
trum with 18σ significance. The ionized helium abun-
dance is measured by the integrated line ratios Y+ = recombination line emission throughout the radio con-
tinuum emission of the GCL. The strongest emission is
I /I . IntheGCL,thisratiois0.075±0.007,similar
Heα Hα
foundalonganeasternandwesternridge,butsignificant
to the solar value (Lang 1974; Roelfsema & Goss 1992).
emission is found between these ridges. All of this emis-
Typically, the line-to-continuum ratio for a radio re-
sion has an unusually narrow line width and LTE line
combination line is a useful way to measure the electron
ratios, suggesting that it is all a part of the same struc-
temperature of the emitting region. The LTE electron
ture and distinct from background Galactic emission.
temperature is:
The morphology of the ionized gas suggests that it
ν 1.1 T 1 0.87 is associated with the radio continuum emission of the
TeLTE =(cid:20)6943GHz T ∆cv1+Y+(cid:21) K, (4) GCL.Thetworidgesandcapofrecombinationlineemis-
l sion are parallel and adjacent to similar structures in
which is related to the actual electron temperature by: radiocontinuummaps(Law et al.2008). Ifthe radiore-
combination line and continuum emission are colocated,
−0.87
TLTE =T b 1− βnτc , (5) the line emission may be in the GC region, as has been
e e(cid:20) n(cid:18) 2 (cid:19)(cid:21) argued for radio continuum GCL (Lasenby et al. 1989;
Bland-Hawthorn & Cohen 2003). A more detailed ar-
where τ is the optical depth of the background contin-
c gument is left for a future paper; for the results shown
uumemission. However,muchofthecontinuumemission
below, we derive physical parameters for the ionized gas
assuming a distance of 8 kpc.
7Technically,collisionalbroadeningtakestheshapeofaLorentz
Themorphologyofthe line emissionis consistentwith
distribution, which, when convolved with a Gaussian, produces a
Voigt profile (Roelfsema&Goss 1992). However, at the low sig- a three-dimensional shell model. In this model, the two
nal tonoiseratioobserved forthe H180−191ǫ transition, noline ridges of emission are the shell edges, where the col-
wingsareapparentandaGaussianisacloseapproximationtothe
umndensityislargest. Bland-Hawthorn & Cohen(2003)
expected lineshape.
8Allowing∆vdtovarydidnotsignificantlychangethefitvalues model the MSX mid-IR emission from this region as a
oritsquality. “telescope dome” shell specified by a height, radius, and
6 Law et al.
shell thickness. Assuming this geometry, the radius of ening. The best-fit model to the line widths gives
thedomeisroughly0◦.3andtheheightisroughly1◦. The n = 910+310 cm−3. The best-fit density is consis-
e −450
GBT observations measured the ridge half-peak thick- tent with the lack of stimulated emission, which would
nessto be about0◦.2; simplesimulationssuggestthatfor change the line ratios for n . 100 cm−3. Using the
e
a shell geometry, the true shell width is about half the rms and true electron densities, we calculate the volume
apparent thickness, or 0◦.1. Assuming this shell is in the filling factor for the deep spectrum as f = hn2i/n2 =
e e
GC region, the dome radius is 40 pc, the total height is (9+27)×10−5(T /3960K)1.3(50 pc/L) cm−3, where L is
−4 e
140pc,andthewidthis15pc. Thepathlengththrough the path length through the shell, assuming a distance
the shell edge would be about 50 pc. Combining this of 8 kpc. Since the emission measure scales as hn2i, the
e
with the EM measurement, gives an rms electron den- derivedfilling factor applies to the densestgas along the
sity, hn2i≈8.8(T /3960 K)3/4 cm−3. line of sight. This filling factor is much smaller than
e e
Thpe morphology of the GCL is reminiscent of “Galac- that seen in the Galactic disk (f ∼ 10−2; Heiles et al.
ticworms”(Koo et al.1992). Wormsareionizedcavities 1996), but comparable to that seen in Galactic outflows
formed when supernovae and stellar winds from H II re- (f ∼10−3−10−4; Heckman et al. 1990).
gions blow out of the Galactic disk (Heiles et al. 1996). The thermal pressure implied by the constraint on T
e
A difference between the GCL and a worm is that the andmeasurementofn isP/k=7.2×106(T /3960K)K
e e
GCL has a cap at its top. The GCL may be thought cm−3. This pressure is about 100 times larger than the
of as a frustrated worm, an ionized cavity that has not total gas pressure near the Sun (Bloemen 1987), but it
yetblownoutof the Galactic disk. A degree-scalestruc- is not unusual for the GC region (Spergel & Blitz 1992;
ture like this would not have been detected by previous Martin et al. 2004; Koyama et al. 1996; Muno et al.
surveys for worms (Koo et al. 1992). 2004). This suggests that the ionized gas is in the GC
region and in equilibrium with its molecular and hot,
4.2. Properties of the Ionized Gas
x-ray-emitting gas.
It is critical to show that the recombination line emis-
sionisinLTEifwewishtoderiveintrinsicgasconditions. 4.3. Mass and Ionization of the Gas
Previous observations in the GC region found narrow
Assuming a shell geometry and GC distance to the
recombination lines, particularly at lower frequencies, recombination line emission allows us to estimate other
and concluded that the emission was most likely stim- physical properties of the gas. For ionized gas in LTE,
ulated(Lockman & Gordon1973;Pauls & Mezger1975; the mass is parameterized as:
Anish Roshi & Anantharamaiah 1997). However, there
are several characteristics of the line emission observed M =0.419(Te/104K)0.175(S5GHz/Jy)0.5(D/kpc)2.5(θG/arcmin)1.5,
here that show that the gas is in LTE, including well- (7)
constrained line ratios consistent with LTE, the lack of where D is the distance to the gas, θG is the an-
coincidence between background continuum brightness gular size of the emission, and the geometry is as-
andlinebrightness,andevidenceforcollisionalbroaden- sumed to have a cylidrical shape (Pauls & Mezger
ing in the line widths. 1975; Panagia & Walmsley 1978). Since the gas is
Assuming the emission is not stimulated, the Doppler in LTE, the line-to-continuum ratio is Tl∆v/Tc =
widths of these lines provide a strict upper limit to the 2.75(Te/3960 K)−0.87 km s−1, as shown in § 3.2.3. The
electron temperature of 3960±120 K. This limit to the integrated line intensity over the ionized shell of gas in
electrontemperatureisamongthelowestvaluesobserved theHCROobservationsisabout1.2K∗13.5kms−1=16
in the Galaxy, although radio recombination line obser- Kkms−1,forb≥0◦.2andsubtracting10%forthe back-
vations of other H II regions have found electron and ground. Thisgivesa5 GHz continuumflux of50Jyand
Doppler temperatures as low as 4000 K (Shaver et al. a ionized mass M = 2×105(Te/3960 K)0.61 M⊙ in the
1983; Afflerbach et al. 1996, M. Goss 2006,private com- shell. Uncertainties in the mass estimate are probably
munication). Some of the best-fit line widths discussed dominated by incomplete sampling of the HCRO survey
inthisworkareaslowas9kms−1,whichcorrespondsto and are about 20%.
a Doppler temperature of ∼1800 K, much smaller than The mean electron density constrains the number of
observed elsewhere in the Galaxy. Lyman continuum photons required to ionize the gas.
The dominant effect in determining the electron tem- TheLymancontinuumphotonfluxisNLy =VnenHα(2),
perature of ionized gas is the cooling efficiency; high where V is the volume of the emitting region and
metal abundance allows ionized gas to cool quickly α(2) ≈ 6.6 × 10−13 cm3 s−1, for T = 3 × 103 K
e
(Mathis 1985). Studies of recombination line emission (Scheffler & Els¨asser 1988). Using the integrated ther-
from H II regions have used this effect to establish that malcontinuumfluxdensity,wefindthatthegasrequires
thereisametallicitygradientwithdistancefromtheGC an ionizing flux N =7×1049(T /3960 K)1.22) s−1.
Ly e
(Shaver et al. 1983; Afflerbach et al. 1996). The gradi- The estimated ionizing flux is about seven times the
entisseenasatrendinT ,withcoolestandmostmetal- Lyman continuum flux of an O7 V star of about 1049
e
rich gas in the GC region. The relation observed by s−1 (Smith et al.2002). Assumingitisashellinthe GC
Afflerbach et al. (1996) predicts T ≈5500 K in the GC region, the source of ionizing photons would most likely
e
region, while Shaver et al. (1983) predicts T ≈ 3100 K. be stars in the plane. Based on geometry, at most half
e
Regardlessoftheexactvalue,theserelationssuggestthat of the stellar flux would reach the shell, so the required
the ionized gas in the GCL is located near the GC, sup- ionizingfluxisequivalenttomorethan1407V-typestars,
porting the morphologicalconnection. or 1.4×1050 s−1. The Arches, Quintuplet, and Central
The broadening of the line width with increasing elec- star clusters, all located inside the GCL in projection,
tronic state, n, is best explained by collisional broad- havemeasuredLymancontinuumfluxesof1051.0,1050.9,
RRLs toward GCL 7
and 1050.5 s−1 (Figer et al. 1999). Even if extinction is regionisbeingactivelydebated(Yusef-Zadeh et al.1987;
large,theseclusterscouldeasilyionizethegasassociated LaRosa et al. 2005; Boldyrev & Yusef-Zadeh 2006); a
with the GCL. weaker field would decelerate more slowly, but could ul-
timately have the same effect.
4.4. Kinematics
The observations confirm previous observations that
5. CONCLUSIONS
found line velocities near zero toward the GCL
(Lockman & Gordon 1973). However, the GBT l–v di- We have presented analysis of new and archival radio
recombinationlineobservationstowardtheGCL,aradio
agrams show that the line velocities tend to be much
more complex, particularly between the bright ridges of continuumshellbelievedtobeevidenceofamassoutflow
emission, from l = 0◦.0 to −0◦.5. Unusually narrow lines from our GC region. The observations have found that
arebrightestnearthe GCLandhavephysicalconditions the radio line emission has a morphology strikingly sim-
most consistent with the GC region. This narrow line ilar to the radio continuum, strongly arguing that they
observed across the GCL at b = 0◦.45 shows a velocity areassociated. Thelineandcontinuumemissionappears
gradientsimilarto Galacticrotation,indicating that the as a limb-brightened shell. A simple shell model for the
structureisbestfitwitharadiusof0◦.3,aheight1◦,and
gas may be connected to the disk.
The low velocity of the ionized gas has been used a shell width 0◦.1.
Recombination lines are detected in averaged profiles
to argue that it is in the foreground to the GC region
(Pauls & Mezger1975). However,recentworkhasshown covering transitions from H106α up to H191ǫ. Diagnos-
that a significant fraction of gas in the central few hun- tics derived from the detected lines show that the emis-
sion is not stimulated, allowing us to constrain the elec-
dred parsecs has velocities near zero. Oka et al. (2005)
studiedH+ absorptionlinestowardtheGCregion;these tron temperature and gas density from the line widths.
3 The electron temperature is unusually low, with an up-
lines are excited at temperatures and densities found in
perlimitofroughly4000K.Thistemperaturelimit,and
the central few hundred parsecs and have only been ob-
the gas pressure, are consistent with conditions in the
served toward the GC region. They found that roughly
1/3ofH+ columndensityhasvelocitiesnearzero. Thus, GC region.
3 Assuming that the gas is organized as a shell and
low velocity gas is quite common in the GC region.
is in the GC region, we derive a mass of M = 2 ×
Sofue (1996) argued that the GCL is associated with
105(T /3960K)0.61 M . This massissimilarto thatob-
molecular gas rotating about the GC at ± ∼ 100 km e ⊙
s−1. While that association relies on a positional co- servedin dwarfstarburstoutflows (Veilleux et al.2005).
Thefillingfactoroftheionizedgasisalsosimilartothat
incidence in a complex region, it is worth considering
seen in starburst outflows. In the canonical model for
how it would relate to the ionized gas. In particular,
such outflows, the emission comes from the ionized sur-
how can the large difference in velocity of these two
face of clumps of molecular gas entrained in the outflow
components be explained? The motion of this molec-
(Heckman et al. 1990).
ular gas is best explained by the gas dynamics in a
Modest observations with radio interferometers cur-
barred gravitational potential (Binney et al. 1991). Un-
rentlyindevelopmentcouldgreatlyexpandontheresults
der this model, the gas possibly associated with the
presented here. A simple estimate using the filling fac-
GCL is either in or transitioning to the inner “x2” or-
torand the size ofour deep integrationshowsthat these
bits (Contopoulos & Mertzanides 1977). In light of this,
clumps could be resolved at size scales of ∼9′′. Deeper
the low velocity of the ionized gas in the GCL may be
observations of more transitions would strengthen the
tied to the shocked molecular gas on these transitioning
significance of collisional broadening and detect spatial
orbits.
structure in the electron density and gas pressure. Fi-
Alternatively, the low velocity of the ionized gas may
nally, high-spectral resolution observations would map
indicate that it as been decelerated by an ambient mag-
the line velocities and help us understand its dynamics.
netic field. Equating gas ram pressure to a magnetic
pressure allows us to estimate the magnetic field re-
quired to stop the gas on a dynamic time scale. For
the molecular gas associated with the GCL by Sofue We thank Miller Goss for an enlightening discussion
(1996), these pressures are equal for B = 1 mG. The and the anonymous referee for useful comments.
possibility of large-scale, mG-strength fields in the GC Facilities: GBT (), HCRO ()
REFERENCES
Afflerbach,A.Churchwell,E.,Acord,J.M.,Hofner,P.,Kurtz,S., Dravskikh,Z.V.&Dravskikh,A.F.1964,Astron.Tsirk.,158,2
&Dupree,C.G.1996,ApJS,106,423 Dupree,A.K.1972,ApJ,173,293
AnishRoshi,D.&Anantharamaiah,K.R.1997,MNRAS,292,63 Dupree,A.K.1969,ApJS,18,221
Binney,J.,Gerhard,O.E.,Stark,A.A.,Bally,J.,&Uchida,K.I. Depree,C.G.&Goldberg,L.1970, ARA&A,8,231
1991, MNRAS,252,210 Figer,D.F.,McLean,I.S.&Morris,M.1999,ApJ,514,202
Bland-Hawthorn,J&Cohen,M.2003,ApJ,582,246 Heckman,T.M.,Armus,L.,&Miley,G.K.1990,ApJS,74,833
Bloemen,J.B.G.M.1987,ApJ,322,694 Heiles,C.,Reach,W.T.,&Koo,B.-C.1996,ApJ,466,191
Boldyrev,S.&Yusef-Zadeh,F.2006,ApJ,637L,101 Kardashev,N.S.1959,SovietAstr.,3,813
Brocklehurst,M.&Seaton, M.J.1972,MNRAS,157,179 Koo,B.-C.,Heiles,C.,&Reach,W.T.1992,ApJ,390,108
Carigi,L.,Peimbert,M.,Esteban,C.,Garc´ıa-Rojas,J.2005,ApJ, Koyama, K., Maeda, Y., Sonobe, T., Takeshima, T., Tanaka, Y.,
623,213 &Yamauchi,S.1996,PASJ,48,249
Cersosimo,J.C.&Magnani,L.1990,A&A,239,287 Lang,C.C.,Goss,W.M.,&Wood, D.O.S.1997,ApJ,474,275
Chevalier,R.A.1992,ApJ,397,L39 Lang,K.R.1974,AstrophysicalFormulae,Springer-Verlag,Berlin
Contopoulos, G.&Mertzanides,C.1977, A&A,61,477 Lang,K.R.&Willson,R.F.1978,ApJ,219,474
8 Law et al.
LaRosa, T. N., Brogan, C. L., Shore, S. N.,Lazio, T. J., Kassim, Roelfsema,P.R.&Goss,W.M.1992,A&ARev.,4,161
N.E.,&Nord,M.E.2005, ApJ,626L,23 Salem,M.&Brocklehurst,M.1979,ApJS,39,633
Lasenby,J.,Lasenby,A.N.,&Yusef-Zadeh,F.1989,ApJ,343,177 Scheffler, H., & Elsa¨sser, H. 1988, in Physics of the Galaxy and
Law, C. J., Yusef-Zadeh, F., Cotton, W. D., & Maddalena, R. J. InterstellarMatter,Springer-Verlag,pg.271
2008, ApJS,177,255 Shaver, P. A., McGee, R. X., Newton, L. M., Danks, A. C., &
Lilley,A.E.&Palmer,P.1968,ApJS,16,143 Pottasch, S.R.1983,MNRAS,204,53
Lockman,F.J.&Gordon,M.A.1973,ApJ,182,25 Smith, L. J., Norris, R. P. F., & Crowther, P. A. 2002, MNRAS,
Lockman, F. J., Pisano, D. J., & Howard, G. J. 1996, ApJ, 472, 337,1309
173 Sofue,Y.1985, PASJ,37,697
Martin,C.L.,Walsh,W.M.,Xiao,K.,Lane,A.P.,Walker,C.K., ——1996, ApJ,459,L69
&Stark,A.A.2004,ApJS,150,239 Sofue,Y.&Handa,T.1984,Nature,310,568
Mathis,J.S.1985,ApJ,291,247 Sorochenko, R.L.&Borozich,E.V.1964,Rep.oftheSov. Acad.
Melia,F.&Falcke,H.2001,ARA&A,39,309 ofScience,162,3,603
Mezger,P.G.&Henderson,A.P.1967,ApJ,147,471 Spergel,D.N.&Blitz,L.1992,Nature,357,665
Morris,M.&Serabyn, E.1996,ARA&A,34,645 Tsuboi,M.,etal.1986, AJ,92,818
Muno,M.P.etal.2004,ApJ,613,326 Veilleux, S., Cecil, G., & Bland-Hawthorn, J. 2005, ARA&A, 43,
Oka, T.,Geballe, T. R.,Goto, M.,Usuda, T.,& McCall,B. 2005 769
ApJ,632,882 Yusef-Zadeh,F.,Morris,M.,&Chance,D.1984,Nature,310,557
Panagia,N.&Walmsley,C.M.1978,A&A,70,411 Yusef-Zadeh,F.&Morris,M.1987,AJ,94,1178
Pauls,T.&Mezger,P.G.1975,A&A,182,25
Reich,W.,Sofue,Y.,&Fuerst,E.1987,PASJ,39,573
RRLs toward GCL 9
TABLE 1
GBTRadio RecombinationLinePointings
Name l b Maptype Exposure Spacing
(deg) (deg) (s) (arcmin)
GCL3 0.0 0.5 3x3grid 90 1
GCL4 359.4 0.5 3x3grid 90 1
GCL3strip +0.016–0.133 0.516–0.633 diagonalstrip 60 1.4
GCL4strip 359.416–359.666a 0.516–0.766 diagonalstrip 60 1.4
b=0.45strip 0.25–359.25 0.45 horizontalstrip 60 5.4
GCL1 0.3 0.5 singlepoint 60 –
GCL7 359.1 0.8 singlepoint 60 –
a Onepointing,nearl=−0◦.48,wasaccidentallyskipped.
TABLE 2
Detected Linesandtheir Propertiesin DeepIntegration TowardGCL
RecombinationLinePeaks
Transition Tl vLSR ∆v Iobsa Roαbsb
(K) (kms−1) (kms−1) (Kkms−1)
H106−113α 0.1257±0.0024 0.71±0.13 13.45±0.30 1.802±0.052 1
He106−113α 0.0110±0.0006 0.97±0.32 11.50±0.74 0.135±0.012 0.075±0.007
H134−142β 0.0339±0.0008 0.75±0.15 13.72±0.36 0.495±0.017 0.275±0.012
H152−162γ 0.0144±0.0007 0.36±0.35 15.26±0.83 0.233±0.017 0.130±0.010
H167−178δ 0.0091±0.0006 1.98±0.40 12.47±1.05 0.120±0.013 0.067±0.007
H180−191ǫ 0.0030±0.0003 1.67±1.34 24.47±3.15 0.073±0.012 0.041±0.007
a Iobs=RlineTmbdv≈1.065Tl∆vb Roαbs=Ixobs/Iαobs
TABLE 3
ComparingLine Ratiosto LTE andStimulatedEmission Models
Transition Roαbs RLαTE R2α0,1 Rα100,10 Rα100,100 Rα100,1000 Rα1000,100 Rα1000,1000 Rα10000,10 Rα10000,100 Rα10000,1000
H106−113α 1 1 1 1 1 1 1 1 1 1 1
H134−142β 0.275±0.012 0.279 0.322 0.381 0.300 0.295 0.281 0.275 0.417 0.242 0.245
H152−162γ 0.130±0.010 0.127 0.224 0.198 0.137 0.133 0.123 0.124 0.193 0.098 0.103
H167−178δ 0.067±0.007 0.073 0.146 0.118 0.083 0.077 0.071 0.072 0.107 0.053 0.058
H180−191ǫ 0.041±0.007 0.047 0.112 0.077 0.055 0.049 0.053 0.048 0.055 0.051 0.047
Note. —ThelineratiosforLTEconditionscorrespondtothe“mean”transitionoftherangeofnvaluesshown. Oscillatorstrengthsweretakenat
themeannortheaverageofthenearesttwonvalues,forfractionaln. Stimulatedemissionmodelslabeledwith(Te,ne)inKandcm−3. Dupree(1972)
andSalem&Brocklehurst(1979)wereusedforT ≤100Kand>100K,respectively.
10 Law et al.
Fig. 1.— Left: The gray scale shows GBT 5 GHz radio continuum emission toward the GCL with contours of H109α brightness from
theHCROatTa=15,20,30,and40mK.Right: Sameastheleftpanel,butwithacontourofH109αlinevelocityat0kms−1. Theareas
tothefarwestandeasthavenegativevelocities andbetween thelinesaregenerallypositivevelocities.
Fig. 2.—PositionsoftheGBTobservationstowardtheGCLareshownascirclesonaGBT5GHzradiocontinuumsurveyoftheregion
(Lawetal.2008). Thecircleshaveadiameterof2′.5,whichistheFWHMoftheGBTbeamat5GHz.