1: \documentstyle[epsf,times]{mn}
2: \newcommand{\etal}{{et al}\/.}
3: \begin{document}
4: \title[{\it Chandra} detection of 3C\,123]{A {\it Chandra} detection of the radio hotspot of 3C\,123}
5: \author[M.J.~Hardcastle \etal]{M.J.\ Hardcastle, M.\ Birkinshaw and
6: D.M.\ Worrall\\
7: Department of Physics, University of Bristol, Tyndall Avenue,
8: Bristol BS8 1TL}
9: \maketitle
10: \begin{abstract}
11: {\it Chandra X-ray Observatory} observations of the powerful, peculiar
12: radio galaxy 3C\,123 have resulted in an X-ray detection of the bright
13: eastern hotspot, with a 1-keV flux density of $\sim 5$ nJy. The X-ray
14: flux and spectrum of the hotspot are consistent with the X-rays being
15: inverse-Compton scattering of radio synchrotron photons by the
16: population of electrons responsible for the radio emission
17: (`synchrotron self-Compton emission') if the magnetic fields in the
18: hotspot are close to their equipartition values. 3C\,123 is thus the
19: third radio galaxy to show X-ray emission from a hotspot which is
20: consistent with being in equipartition. {\it Chandra} also detects emission
21: from a moderately rich cluster surrounding 3C\,123, with $L_X (2--10{\
22: \rm keV}) = 2 \times 10^{44}$ ergs s$^{-1}$ and $kT \sim 5$ keV, and
23: absorbed emission from the active nucleus, with an inferred intrinsic
24: column density of $1.7 \times 10^{22}$ cm$^{-2}$ and an intrinsic
25: 2--10 keV luminosity of $10^{44}$ ergs s$^{-1}$.
26: \end{abstract}
27: \begin{keywords}
28: galaxies: active -- X-rays: galaxies -- galaxies: individual: 3C\,123
29: -- radiation mechanisms: non-thermal
30: \end{keywords}
31:
32: \section{Introduction}
33:
34: The magnetic field strengths in the extended components of
35: extragalactic radio sources cannot be inferred directly from
36: observations of synchrotron emission, and so the energy densities and
37: pressures in the radio-emitting components are poorly constrained. In
38: order to make progress it is common to estimate `minimum energy' field
39: strengths (Burbidge 1956), which minimise the energy density required
40: for a given synchrotron emissivity. This is roughly equivalent
41: to the assumption that that magnetic and relativistic particle
42: energy densities are equal (`equipartition'). But without measurements
43: of magnetic field strengths these assumptions, for which there is no
44: physical justification, may underestimate the true energy densities by
45: arbitrary factors.
46:
47: The magnetic field strength may be measured by observations of the
48: `synchrotron self-Compton' (SSC) process, in which the
49: synchrotron-emitting electrons inverse-Compton scatter synchrotron
50: photons up to X-ray energies. Because the emissivity from this process
51: depends on the photon number density (which is known from radio
52: observations) and the electron number density as a function of energy,
53: observations of SSC emission allow the electron number density to be
54: inferred, and so determine the magnetic field strength. Such tests
55: require observations of regions with well-measured volume and a
56: well-defined synchrotron spectrum with a high photon energy density;
57: these conditions exist in the hotspots of FRII radio sources. Direct
58: evidence supporting the equipartition/minimum energy assumptions in
59: hotspots has come from only two X-ray observations. Harris, Carilli \&
60: Perley (1994) detected the hotspots of the powerful FRII Cygnus A
61: (3C\,405) with {\it ROSAT} and showed that the X-ray emission could be
62: interpreted as being due to the SSC process, with a magnetic field
63: strength consistent with the equipartition model. [This result was
64: recently confirmed with {\it Chandra} by Wilson, Young \& Shopbell
65: (2000).] {\it ROSAT} was not sensitive enough to detect any other SSC
66: hotspots, though it was used to put lower limits on the field
67: strengths in some sources (Hardcastle, Birkinshaw \& Worrall
68: 1998). More recently, {\it Chandra} verification observations have
69: detected the hotspots of 3C\,295, another powerful radio galaxy, at a
70: level which implies field strengths fairly close to the equipartition
71: values if the emission process is SSC (Harris \etal\ 2000). Here we
72: report a third detection, of the E hotspot of the radio galaxy
73: 3C\,123, based on our {\it Chandra} AO1 guest observer (GO)
74: observations.
75:
76: 3C\,123 is a $z = 0.2177$ radio galaxy, notable for its peculiar radio
77: structure. Like normal classical double sources it has twin hot spots
78: on either side of the active nucleus, but the lobes take the form of
79: diffuse twisted plumes unlike those in any other well-studied object
80: (e.g. Riley \& Pooley 1978; Hardcastle \etal\ 1997, hereafter
81: H97). Like Cygnus A and 3C\,295, its radio luminosity is unusually
82: high for its redshift. For our purposes, its most important feature is
83: its bright eastern double hotspot. With a flux density of $\sim 6$ Jy
84: at 5 GHz, it is the second brightest hotspot complex known (after
85: Cygnus A). The hotspots' structure and synchrotron spectrum are well
86: known (H97; Meisenheimer \etal\ 1989; Meisenheimer, Yates \& R\"oser
87: 1997; Looney \& Hardcastle, 2000).
88:
89: Throughout this letter we use $H_0 = 50$ km s$^{-1}$ Mpc$^{-1}$ and
90: $q_0 = 0$. At the redshift of 3C\,123, 1 arcsec corresponds to
91: 4.74 kpc.
92:
93: \section{Observations}
94:
95: We observed 3C\,123 with the {\it Chandra X-ray Observatory} for 46.7
96: ks on 2000 March 21. The source was near the aim point for the S3 ACIS
97: chip. After filtering for intervals of high background, the usable
98: exposure time was 38.5 ks. We considered events in the energy range
99: 0.5--7.0 keV, as the spectral response of the instrument is uncertain outside
100: this range. Fig.\ \ref{image} shows the exposure-corrected {\it
101: Chandra} image of 3C\,123 in this band. Diffuse cluster emission, an
102: X-ray nucleus and the eastern hotspot are all detected in X-rays. We
103: discuss each component in turn. In each case, spectra were extracted
104: using {\sc ciao}, with the best available responses being constructed for
105: each extraction region, and analysed using {\sc xspec}. Spectra were
106: binned such that every bin had $>20$ net counts.
107:
108: \begin{figure}
109: \epsfxsize 8.4cm
110: \epsfbox{fig1.eps}
111: \caption{Exposure-corrected 0.5--7.0 keV {\it Chandra} image of the
112: central region of 3C\,123. Linear greyscale: black is $4 \times
113: 10^{-7}$ photons cm$^{-2}$ s$^{-1}$ per standard Chandra pixel (0.492
114: arcsec on a side). Superposed is the 3 mJy beam$^{-1}$ contour from an
115: 8.4-GHz Very Large Array (VLA) map with 0.6-arcsec resolution (H97),
116: showing the position of the radio lobes. As discussed in section
117: \ref{nucleus}, the radio map has been shifted by $\sim 3$ arcsec so
118: that the radio core aligns with the X-ray nucleus.}
119: \label{image}
120: \end{figure}
121:
122: \section{The cluster}
123:
124: The X-ray counts from 3C\,123 are dominated by diffuse cluster-scale
125: emission, which is detectable above the background more than an
126: arcminute away from the central source. The source was known to be
127: extended from {\it ROSAT} images (Hardcastle \& Worrall 1999), and
128: there are many faint optical objects in the field of 3C\,123 which may
129: be cluster members (e.g.\ Longair \& Gunn 1975), though strong
130: galactic reddening (the source is at $b \approx -12^\circ$, and see
131: below) means that this cluster has not been studied in detail in the
132: optical. We detect around 5,000 counts in a 75-arcsec radius circle centred
133: on the nucleus, excluding the core and hotspot. The overall spectrum
134: of this region is well fitted with an absorbed MEKAL spectrum with $kT
135: = 5.0^{+0.6}_{-0.4}$ keV, abundance of $0.47^{+0.12}_{-0.11}$ solar
136: and a galactic hydrogen column density of $4.3_{-0.3}^{+0.2} \times
137: 10^{21}$ cm$^{-2}$ (errors are $1\sigma$ for one interesting
138: parameter). 3C\,123 lies behind a well-known molecular cloud system in
139: Taurus (e.g.\ Ungerechts \& Thaddeus 1987), and our derived column
140: density is consistent with the total hydrogen column inferred
141: from radio observations of the molecular cloud system. From HI
142: absorption against 3C\,123, Colgan, Salpeter \& Terzian (1988) derive
143: $N_{\rm HI} = 1.97 \times 10^{21}$ cm$^{-2}$ at the velocity of the
144: cloud system, while the molecular hydrogen column can be inferred from
145: CO measurements [$W_{\rm CO} \approx 10$ K km s$^{-1}$, Megeath
146: (private communication)] to be $N_{\rm H_2} \approx 1.6 \times
147: 10^{21}$ cm$^{-2}$, with a large systematic uncertainty [we use a
148: recent estimate of the conversion factor averaged over the galactic
149: plane, due to Hunter \etal\ (1997), but this may not be appropriate
150: for the Taurus region]. We adopt a galactic absorption column of $4.3
151: \times 10^{21}$ cm$^{-2}$, unless otherwise stated, from now on.
152:
153: The X-ray spectral fit implies a rest-frame 2--10 keV luminosity from
154: the cluster within a radius of 75 arcsec (360 kpc) of $2 \times
155: 10^{44}$ ergs s$^{-1}$, consistent with the fitted temperature on the
156: temperature-luminosity relation (determined largely for Abell
157: clusters) of David \etal\ (1993). Fig.\ \ref{image} shows a plateau
158: of X-ray emission on scales comparable to those of the radio source,
159: with clear structure in the emission (note particularly X-ray
160: voids to the E and SW of the nucleus) although there is no evidence
161: for interaction between the X-ray gas and the radio lobes. The voids
162: may represent large-scale inhomogeneity in the cluster gas; if so,
163: they would help to explain the peculiar radio structure. The central
164: cooling time is a few $\times 10^{9}$ years, so we might expect to see
165: a cooling flow around the source. But there is no strong evidence for
166: cooling in the temperature fits; the best-fitting temperature
167: for the material within 15 arcsec of the nucleus (fixing the abundance
168: to the value obtained for the whole cluster) is $4.4\pm 0.3$
169: keV. Since gas with temperatures below $\sim 3$ keV is not observed
170: even in the centres of well-studied cooling flows (e.g.\ Fabian \etal\
171: 2000, Peterson \etal\ 2000) this is perhaps not surprising.
172: Our preliminary analysis implies particle densities around the lobes which
173: are similar to those reported by Hardcastle \& Worrall (2000) using
174: {\it ROSAT} data, and the measured temperature implies comparable, but
175: slightly larger, external pressures.
176:
177: \section{The nucleus}
178: \label{nucleus}
179:
180: The point-like nucleus contains $517 \pm 36$ 0.5--7.0 keV counts,
181: measured in a 2.5-arcsec region about the centroid, with the
182: background being taken from a 3--5 arcsec concentric annulus. Pileup
183: is not significant. The X-ray core position (J2000.0) is measured to
184: be 04 37 04.30 +29 40 11.2. The core position measured from the VLA
185: radio map of H97 is 04 37 04.375 +29 40 13.86, and this is expected to
186: be accurate to within about 0.05 arcsec. The X-ray core position is
187: therefore offset from the (true) radio position by about 3 arcsec. We
188: attribute this to uncertainties in aspect determination in the
189: early version of the pipeline software (R4CU4UPD7.4) used to process
190: the {\it Chandra} data (see URL:
191: $<$http://asc.harvard.edu/mta/ASPECT/$>$). In Fig.\ \ref{image} we
192: have aligned the radio data with the X-ray core.
193:
194: The spectrum of the nucleus is well fitted with an absorbed,
195: flat-spectrum power law model. Fitting with free galactic absorption,
196: the best-fit values of photon index $\Gamma$ and $N_{\rm H}$ are $1.16
197: \pm 0.14$ and $1.48_{-0.24}^{+0.32} \times 10^{22}$ cm$^{-2}$,
198: respectively. If we fix galactic absorption at the value derived from
199: the cluster fits and require the absorber to be at the redshift of the
200: galaxy, the best-fit $N_{\rm H}$ for the intrinsic absorber is
201: $1.69_{-0.41}^{+0.51} \times 10^{22}$ cm$^{-2}$, with the photon index
202: unchanged. This implies a rest-frame 2--10 keV luminosity (assumed
203: isotropic) of $10^{44}$ ergs s$^{-1}$, comparable to the luminosity of
204: the nuclear component in 3C\,295 (Harris \etal\ 2000). Fits in which
205: the absorbing column is constrained to the galactic value are much
206: poorer and require an inverted nuclear spectrum. The column density
207: required for the intrinsic absorber is considerably lower than the $4 \times
208: 10^{23}$ cm$^{-2}$ inferred for Cygnus A by Ueno \etal\ (1994), and
209: more comparable to that inferred for the much lower luminosity nucleus
210: of Hydra A by Sambruna \etal\ (2000). The best-fit photon index is
211: flatter than might be expected from, for example, the photon indices
212: of radio-loud quasars (e.g.\ Lawson \& Turner 1997) or other radio
213: galaxies (Sambruna, Eracleous \& Mushotzky 1999) although the errors
214: are large; as shown in Fig.\ \ref{speccont}, more reasonable values of
215: $\Gamma$ are allowed in conjunction with somewhat higher intrinsic absorbing
216: columns. The inferred absorbing column in front of the nucleus
217: explains the non-detection of this core component in the {\it ROSAT}
218: HRI image (Hardcastle \& Worrall 1999).
219:
220: \begin{figure}
221: \epsfxsize 8.4cm
222: \epsfbox{fig2.eps}
223: \caption{Joint confidence contours for the spectrum of the nucleus of
224: 3C\,123 using the model described in the text. Intrinsic absorbing
225: column is plotted on the $x$-axis, power-law photon index on the
226: $y$-axis. The contours are at 1, 2 and 3$\sigma$ for two interesting
227: parameters. The cross marks the best-fit values.}
228: \label{speccont}
229: \end{figure}
230:
231: \section{The hotspot}
232:
233: The E hotspot complex is detected with $145 \pm 32$ 0.5--7.0 keV
234: counts, using a 2.5-arcsec source circle and concentric 3--5 arcsec
235: background annulus. The X-ray hotspot is positionally coincident with
236: the larger, `secondary' hotspot of the eastern hotspot pair in the
237: radio (after the X-ray and radio cores have been aligned). The X-ray
238: emission appears to be slightly elongated in an east-west direction,
239: matching the radio. (The X-ray structure will be discussed in more
240: detail in a subsequent paper which will include results of further
241: radio observations now in progress.) The X-ray spectrum of the
242: hotspot is well fitted with a power law with $\Gamma = 1.6 \pm 0.3$,
243: with the absorbing column fixed at the galactic value. The
244: corresponding unabsorbed 1-keV flux density is $4.6 \pm 0.9$ nJy.
245:
246: We used the synchrotron-self-Compton code described by Hardcastle
247: \etal\ (1998) to predict the SSC flux density expected at this
248: frequency from the hotspots. The basic model for the hotspots is
249: described by Looney \& Hardcastle (2000). The larger hotspot is
250: treated as a cylinder of $1.14 \times 0.54$ arcsec (length times
251: radius); the fainter `primary' (more compact) hotspot is a cylinder of
252: $0.74 \times 0.14$ arcsec, based on the MERLIN maps of H97. Radio flux
253: densities of the two components are taken from Looney \&
254: Hardcastle. In addition to these, we have used infra-red and optical
255: upper limits and a 231-GHz data point from Meisenheimer \etal\ (1989,
256: 1997) and archival HST observations, and low-frequency radio data from
257: Readhead \& Hewish (1974) and Stephens (1987). As these data do not
258: resolve the two hotspot components, we have {\it approximately}
259: corrected them by scaling by the appropriate factors measured from the
260: 5-GHz data. Looney \& Hardcastle showed that the radio-to-mm spectra
261: of the two hotspots are well modelled as broken power laws, and we
262: adopt the break energies they found. The apparent low-frequency
263: turnover in the spectrum observed by Stephens (1987) requires a
264: low-energy cutoff in the electron energy spectrum corresponding at
265: equipartition to a minimum Lorentz factor $\gamma_{\rm min} \approx
266: 1000$, and we adopt this value, although Stephens' flux densities are
267: inconsistent with a larger flux at a lower frequency derived from the
268: scintillation measurements of Readhead \& Hewish. If we were to adopt
269: the scintillation measurements as our low-frequency constraint, we
270: would obtain $\gamma_{\rm min} \approx 400$, which is more consistent
271: with the value inferred for the hotspots of Cygnus A by Carilli \etal\
272: (1991); but this would not significantly affect our
273: conclusions. (Scheduled low-frequency VLBA observations should give a
274: definitive answer.) An upper limit on the maximum Lorentz factor is
275: given by the non-detection in the IR, $\gamma_{\rm max} < 3.6 \times
276: 10^5$; a lower limit is given by the detection at 231 GHz,
277: $\gamma_{\rm max} > 8 \times 10^4$. The SSC emissivity at 1 keV turns
278: out to be insensitive to $\gamma_{\rm max}$ if it lies between these
279: two values, and so we fix $\gamma_{\rm max}$ at its largest
280: value. With these parameters, the equipartition field strengths of the
281: two eastern hotspot components, assuming no contribution to the energy
282: density from non-radiating particles such as protons, are 24 nT
283: (primary) and 16 nT (secondary), and the predicted SSC flux densities
284: at 1 keV are respectively 0.44 and 2.6 nJy. The predicted photon index
285: at this frequency is 1.55 (of course, this is simply a function of the
286: electron energy spectrum, and so is true for any inverse-Compton
287: process). Fig.\ \ref{flux} shows the synchrotron fluxes and SSC
288: prediction for the secondary hotspot. The predicted SSC flux density
289: for the much weaker western hotspot pair is negligible, $\sim 0.07$
290: nJy, corresponding to 2 {\it Chandra} counts in this observation.
291:
292: \begin{figure}
293: \epsfxsize 8.4cm
294: \epsfbox{fig3.eps}
295: \caption{The spectrum of the secondary hotspot of 3C\,123 at
296: equipartition. Points are from data as described in the text. Symbols
297: indicate the source of the data, as follows: filled circles, Looney
298: \& Hardcastle (2000); crosses, Meisenheimer \etal\ (1989);
299: squares, Stephens (1987); triangle, Readhead \& Hewish (1974); star,
300: optical limit from HST data; open circle, {\it Chandra} data
301: point. Arrows denote an
302: upper limit. Error bars are smaller than the symbols
303: in most cases. The
304: solid line is the model synchrotron spectrum. The dash-dotted line shows
305: the predicted SSC spectrum at equipartition and the dotted line shows
306: inverse-Compton emission due to scattering of microwave background
307: photons. The two solid bars through the X-ray data point show the 0.5--7.0
308: keV band of the {\it Chandra} data and the $1\sigma$ range of photon
309: indices permitted by the data. Frequencies are plotted in the
310: rest frame of the radio source.}
311: \label{flux}
312: \end{figure}
313:
314: These predictions are relatively insensitive to cosmological
315: parameters; for example, using a cosmology where $\Omega_{\rm matter}
316: = 1.0$ gives a 3 per cent decrease in the expected flux density from
317: the secondary, while using $H_0 = 70$ km s$^{-1}$ Mpc$^{-1}$,
318: $\Omega_{\rm matter} = 0.3$, $\Omega_\Lambda = 0.7$ gives a 2 per cent
319: decrease. The hotspots may be projected, but this also has only a weak
320: effect: if the projection angle is $45^\circ$ then the true long axis
321: is larger by a factor $\sqrt{2}$ and the predicted flux is about 7 per
322: cent lower. One clear systematic error in the calculations comes from
323: an assumption of spherical symmetry for the scattering geometry which
324: is used in the SSC code. Because a cylinder has a higher surface area
325: to volume ratio, the mean photon density is lower in a cylinder than
326: in a sphere for a given volume synchrotron emissivity. We estimate
327: that this effect is less than 10 per cent for the secondary component,
328: but we may be overestimating the SSC flux from the fainter primary by
329: 30 per cent (these values are increased if there is substantial
330: projection).
331:
332: Comparing our predicted flux densities with the data, we see that the
333: magnitude of the observed flux density, its origin in the larger
334: eastern hotspot and its photon index are all in good agreement with
335: the equipartition predictions of the SSC model, as is the non-detection of the
336: western hotspot pair. The observed 1-keV flux density of the E
337: hotspots is somewhat higher than the predicted value for an
338: equipartition field, though only by about 1.5 standard
339: deviations. Most of the changes to the model discussed above have the
340: effect of reducing the SSC flux density; to increase it we must reduce
341: the magnetic field strength or find an additional (external) source of
342: photons. If the magnetic field in the secondary hotspot is reduced by
343: 25 per cent to 12 nT, the secondary can produce all the flux seen in
344: X-rays. Neglecting the small SSC contribution from the primary
345: hotspot, the {\it Chandra} data with their associated uncertainties
346: imply within the SSC model that the magnetic field strength in the secondary
347: hotspot is $12 \pm 2$ nT ($1\sigma$ statistical errors only).
348:
349: One possible external source of photons is the active nucleus of
350: 3C\,123. Inverse-Compton (IC) scattering of photons from the active
351: nucleus will make a significant contribution to the X-ray emission if
352: such photons (at frequencies around $10^{11}$ Hz, since $\gamma_{\rm
353: min} \approx 1000$) are comparable in number density to the
354: synchrotron photons. At this frequency, the number density of
355: synchrotron photons is approximately 0.08 m$^{-3}$ Hz$^{-1}$; since
356: the hotspots are a projected 37 kpc from the nucleus, a similar number
357: density would be produced from the nucleus if its luminosity at this
358: frequency as seen by the hotspot were $\ga 2 \times 10^{27}$ W
359: Hz$^{-1}$ sr$^{-1}$, which would correspond to a flux density of $\ga
360: 100$ Jy at $10^{11}$ Hz. (This condition is equivalent to $S_{\rm
361: core} = (3D/2R)^2 S_{\rm hs}$, where $D$ is the core-hotspot distance
362: and $R$ is the hotspot radius, and the two fluxes are measured at the
363: required frequency.) The observed core flux density of 3C\,123 at
364: $10^{11}$ Hz is about 40 mJy, though it may be variable (Looney \&
365: Hardcastle 2000), so that isotropic radiation from the core cannot
366: provide the required photon density. However, if the core emission at
367: this frequency is beamed, the hotspot will see the core as having a
368: higher luminosity than the one we observe. We assume the most
369: favourable case for this model of no misalignment between the pc- and
370: kpc-scale jet, though such good alignment is not often observed, and
371: we neglect effects due to the finite angle subtended by the hotspot at
372: the nucleus, which may be significant. The ratio of required to
373: observed flux, ${\cal R}$ (${\cal R} \approx 2500$ for rough equality
374: of predicted SSC and IC flux densities), then constrains $\beta$, the
375: bulk speed in the nucleus, and $\theta$, the angle of the core-hotspot
376: vector to the line of sight:
377: \[
378: {{(1-\beta)^{-(2+\alpha)} + (1+\beta)^{-(2+\alpha)}}\over{(1 -
379: \beta \cos \theta)^{-(2+\alpha)} + (1 +
380: \beta \cos \theta)^{-(2+\alpha)}}} =
381: {(1-\cos \theta)^{-(1+\alpha)}{\cal R}\over{\sin^2\theta}}
382: \]
383: where the term $(1-\cos \theta)^{-(1+\alpha)}$ approximately corrects
384: for the anisotropic nature of the resulting IC emission (e.g.\ Jones,
385: O'Dell \& Stein 1974), and the core is treated as a two-sided jet with
386: a power-law spectrum; the $\sin^2\theta$ term incorporates the effects
387: of projection. If we assume a core spectral index $\alpha = 0.5$, then
388: we can obtain ${\cal R} \approx 2500$ for plausible $\beta$,
389: corresponding to bulk Lorentz factors $\sim 4$--$10$, if the source is
390: within 50 degrees of the plane of the sky. (In reality, this value
391: for $\alpha$ is probably an overestimate, since the $10^{11}$ Hz
392: photons seen by the nucleus will be blueshifted from lower frequencies
393: where core spectra are typically flat, $\alpha \approx 0$. Lower
394: $\alpha$ requires higher $\beta$.) As no jet has been detected in
395: 3C\,123 and its optical emission lines are weak, the angle to the line
396: of sight is not constrained, and so we cannot rule out a contribution
397: to the X-ray emission from nuclear IC scattering. However, a
398: contribution at approximately the same level as the SSC emission would
399: not affect the conclusion that the hotspot is close to equipartition;
400: in fact, it might account for some of the difference between the
401: equipartition prediction and the observed 1-keV flux density. But if
402: nuclear IC emission {\it dominates} the observed X-rays, then the
403: hotspot could have a magnetic field higher than the equipartition
404: value, and this cannot be ruled out by the present observations. For
405: example, ${\cal R} = 2.5 \times 10^4$, corresponding to an IC emissivity ten
406: times the SSC value, can be obtained for a bulk Lorentz factor
407: $\approx 6$ if the source is close to the plane of the sky and $\alpha
408: = 0.5$, and would require $B \approx 4B_{\rm eq}$. Such a model requires a
409: coincidence to explain the similarity of the observed emissivity to
410: that predicted by the simple SSC model with a near-equipartition
411: field.
412:
413: In carrying out the SSC calculations we have assumed that the hotspots
414: are homogeneous, that they contain no relativistic protons, and that
415: the small-scale filling factor is unity. From the MERLIN maps of H97
416: we know that the hotspots do have internal structure on 100-pc scales,
417: although the variations in surface brightness are not very large;
418: there is no evidence for filamentary structures of the kind seen in
419: radio lobes. As discussed in Hardcastle \& Worrall
420: (2000), the general effect of a filling factor less than unity is to
421: increase the SSC emissivity, but the results are dependent on the
422: geometry of the synchrotron-emitting regions, particularly if the low
423: filling factor is a result of a spatial variation in electron density
424: in a relatively constant magnetic field. Our present results may be
425: taken as evidence against low filling factors in the hotspots, as such
426: filling factors would require coincidences to produce X-ray emission
427: at the observed levels. Similarly, if the particle population is
428: energetically dominated by non-synchrotron-emitting particles such as
429: relativistic protons, it is a coincidence that the energy density in
430: magnetic field corresponds so closely to that in the
431: synchrotron-emitting electrons.
432:
433: Although SSC emission is a required process, we cannot rule out the
434: possibility that the magnetic field strength is much greater than the
435: equipartition value and that some other process happens to produce
436: X-ray emission at a level consistent with the SSC model. One such
437: process which we have already discussed is the IC
438: scattering of nuclear photons. Some other simple models can be
439: rejected. Thermal bremsstrahlung can be ruled out by the compact
440: structure seen with {\it Chandra} and the flat X-ray spectrum, while,
441: as shown in Fig.\ \ref{flux}, IC scattering of the
442: microwave background is two orders of magnitude too weak to be
443: responsible for the observed emission. But more speculative models
444: remain possible. For example, we cannot rule out the possibility that
445: the X-ray emission is synchrotron radiation from an arbitrarily chosen
446: population of electrons. This model has been invoked to explain some
447: other X-ray hotspots (e.g. Harris \etal\ 1999) in which the
448: IC or SSC models do not seem to work
449: well. In the case of 3C\,123, it requires a coincidence to explain the
450: close similarity between the observed X-ray emission and the
451: predictions of the SSC model.
452:
453: One model which makes quantitative predictions about the origin and
454: properties of this second population of electrons is the
455: proton-induced cascade (PIC) model of Mannheim, Kr\"ulls \& Biermann
456: (1991). In this model, the high-energy electrons which produce X-ray
457: synchrotron emission are the end result of photomeson production on a
458: population of ultra-relativistic protons, through pion decay and pair
459: production. If protons are present in the jet, they should undergo
460: shock acceleration in the hotspot; there is some evidence for
461: high-energy protons in the lobes of FRII sources (Leahy \& Gizani
462: 1999, Hardcastle \& Worrall 2000). Mannheim \etal\ considered the
463: hotspot of 3C\,123 and predicted that the PIC process would dominate
464: over SSC if protons were highly energetically dominant in the hotspot
465: and in equipartition with magnetic fields at the level inferred by
466: Meisenheimer \etal\ (1989), somewhat higher than our equipartition
467: fields. Their predicted X-ray flux at 1 keV is $\sim 30$ nJy, nearly a
468: factor 10 higher than the observed value, so a model with extreme
469: proton dominance does not seem to be consistent with the
470: data. However, if protons have energy densities closer to those of the
471: electrons, we cannot rule out an origin from PIC-generated electrons
472: for some or all of the observed X-rays; the spectra of the two
473: processes are not distinguishable with our data. Once again, though, a
474: moderate contribution from the PIC process would not affect our
475: conclusions regarding the closeness of the hotspot to equipartition,
476: while a model in which PIC was responsible for {\it all} the emission
477: would require fine-tuning of the energy fraction in fields, electrons
478: and protons, and so seems less plausible than the simple SSC model.
479:
480: \section{Conclusions}
481:
482: The X-ray emission from the E hotspot of 3C\,123 is consistent with a
483: SSC model, with the inferred magnetic field strength close to the
484: value predicted from equipartition of energy between the magnetic
485: field and the synchrotron-emitting electrons. Other models are
486: possible, but require coincidences to explain the closeness of the
487: X-ray flux density and (in some cases) the photon index to the SSC
488: prediction. This reinforces a conclusion already drawn from
489: observations of Cygnus A and 3C\,295 that the magnetic field strengths
490: in typical hotspots are near their equipartition values. Although
491: there is still no {\it a priori} reason to expect equipartition, it is
492: now very likely that it is achieved in the hotspots of at least some
493: fraction of the source population. Observations of inverse-Compton
494: scattering of microwave background photons from radio lobes may tell a
495: similar story (Feigelson \etal\ 1995; Tsakiris \etal\ 1996; Tashiro
496: \etal\ 1998). Even the deviations from equipartition reported by
497: Tashiro \etal\ require a magnetic field strength only a factor $\sim
498: 2$ below the equipartition value. However, some X-ray detections of
499: hotspots (e.g. 3C\,120, Harris et al 1999; Pictor A, Wilson, Young \&
500: Shopbell 2001) are at a level much too bright to be consistent either
501: with synchrotron emission (from the electron population responsible
502: for the radio and optical synchrotron radiation) or with SSC at
503: equipartition. More observations are necessary to demonstrate that
504: Cygnus A, 3C\,295 and 3C\,123 are typical of most radio
505: sources. We will report on the results of scheduled {\it Chandra}
506: observations of further hotspot sources in a future paper.
507:
508: \section*{Acknowledgements}
509: We thank all those involved with the design and operation of {\it
510: Chandra} for doing such an excellent job, and particularly the staff
511: of the {\it Chandra} X-ray Center for their help with data
512: analysis. We are grateful to an anonymous referee for a careful
513: reading of the paper and constructive suggestions. The National Radio
514: Astronomy Observatory Very Large Array is operated by Associated
515: Universities Inc., under co-operative agreement with the National
516: Science Foundation.
517:
518: \begin{thebibliography}{}
519: \bibitem[]{126}Burbidge G., 1956, ApJ, 124, 416
520:
521: \bibitem[]{155}Carilli C.L., Perley R.A., Dreher J.W., Leahy J.P., 1991, ApJ, 383, 554
522:
523: \bibitem[]{173}Colgan S.W.J., Salpeter E.E., Terzian Y., 1988, ApJ, 328, 275
524:
525: \bibitem[]{209}David L.P., Slyz A., Jones C., Forman W., Vrtilek S.D., 1993, ApJ, 412, 479
526:
527: \bibitem[]{246}Fabian A.C., et al., 2000, MNRAS in~press, astro-ph/0007456
528:
529: \bibitem[]{324}Hardcastle M.J., Alexander P., Pooley G.G., Riley J.M., 1997, MNRAS, 288, 859 [H97]
530:
531: \bibitem[]{326}Hardcastle M.J., Birkinshaw M., Worrall D.M., 1998, MNRAS, 294, 615
532:
533: \bibitem[]{332}Hardcastle M.J., Worrall D.M., 1999, MNRAS, 309, 969
534:
535: \bibitem[]{333}Hardcastle M.J., Worrall D.M., 2000, MNRAS in~press (astro-ph/0007260)
536:
537: \bibitem[]{340}Harris D.E., Carilli C.L., Perley R.A., 1994, Nat, 367, 713
538:
539: \bibitem[]{345}Harris D.E., Hjorth J., Sadun A.C., Silverman J.D., Vestergaard M., 1999, ApJ, 518, 213
540:
541: \bibitem[]{346}Harris D.E., et al., 2000, ApJ, 530, L81
542:
543: \bibitem[]{374}Hunter S.D., et al., 1997, ApJ, 481, 205
544:
545: \bibitem[]{404}Jones T.W., O'Dell S.L., Stein W.A., 1974, ApJ, 188, 353
546:
547: \bibitem[]{466}Lawson A.J., Turner M.J.L., 1997, MNRAS, 288, 920
548:
549: \bibitem[]{477}Leahy J.P., Gizani N.A.B., 1999, To appear in `Life Cycles of Radio Galaxies', ed. J. Biretta et al., New Astronomy Reviews, astro-ph/9909121
550:
551: \bibitem[]{509}Longair M.S., et al., 1975, MNRAS, 170, 121
552:
553: \bibitem[]{518}Looney L.W., Hardcastle M.J., 2000, ApJ, 534, 172
554:
555: \bibitem[]{536}Mannheim K., Kr\"ulls P.L., Biermann P.L., 1991, A\&A, 251, 723
556:
557: \bibitem[]{551}Meisenheimer K., R\"oser H.-.J., Hiltner P.R., Yates M.G., Longair M.S., Chini R., Perley R.A., 1989, A\&A, 219, 63
558:
559: \bibitem[]{552}Meisenheimer K., Yates M.G., R\"oser H.-.J., 1997, A\&A, 325, 57
560:
561: \bibitem[]{644}Peterson J.R., et al., 2000, A\&A in press, astro-ph/0010658
562:
563: \bibitem[]{669}Readhead A.C.S., Hewish A., 1974, Mem.~RAS, 78, 1
564:
565: \bibitem[]{680}Riley J.M., Pooley G.G., 1978, MNRAS, 183, 245
566:
567: \bibitem[]{705}Sambruna R.M., Chartas G., Eracleous M., Mushotzky R.F., Nousek J.A., 2000, ApJ, 532, L91
568:
569: \bibitem[]{706}Sambruna R.M., Eracleous M., Mushotzky R.F., 1999, ApJ, 526, 60
570:
571: \bibitem[]{765}Stephens P., 1987, PhD thesis, University of Manchester
572:
573: \bibitem[]{785}Tashiro M., et al., 1998, ApJ, 499, 713
574:
575: \bibitem[]{799}Ueno S., Koyama K., Nishida M., Yamauchi S., Ward M.J., 1994, ApJ, 431, L1
576:
577: \bibitem[]{802}Ungerechts H., Thaddeus P., 1987, ApJS, 63, 645
578:
579: \bibitem[]{843}Wilson A.S., Young A.J., Shopbell P.L., 2000, ApJ (Letters) in~press (astro-ph/0009308)
580:
581: \bibitem[]{844}Wilson A.S., Young A.J., Shopbell P.L., 2001, ApJ in~press (astro-ph/0008467)
582:
583: \end{thebibliography}
584:
585: \bsp
586:
587: \end{document}
588: