astro-ph0204355/ms.tex
1: \documentclass[12pt,preprint]{aastex}
2: \newcommand{\vdag}{(v)^\dagger}
3: \newcommand{\myemail}{skywalker@galaxy.far.far.away}
4: 
5: \def\physica{Physica \,}
6: \def\spjetp{Sov.Phys.JETP \,}
7: \def\casp{Comments.Astrophys.Space Phys. \,}
8: \def\al{Astrophysical Letters \,}
9: \def\physrep{Phys.Rep. \,}
10: \def\asroparticle{Astroparticle Physics \,}
11: \def\cl{\centerline}
12: \def\ni{\noindent}
13: \def\ss{\smallskip}
14: \def\bs{\bigskip}
15: \def\ms{\medskip}
16: \def\ea{et al. \,}
17: \def\eg{{\it e.g.\,}}
18: \def\be{\begin{equation}}
19: \def\ee{\end{equation}}
20: \def\rel{relativistic \,}
21: \def\nrel{nonrelativistic \,}
22: \def\sz{Sunyaev \& Zeldovich \,}
23:  
24: \begin{document}
25: \bs
26: \bs
27: 
28: \title{CMB Comptonization by Energetic Nonthermal Electrons \\in 
29: Clusters of Galaxies}
30: \vspace{2cm}
31: \bs
32: 
33: \author{\bf Meir Shimon}
34: \affil{School of Physics and Astronomy, Tel Aviv University, Tel 
35: Aviv, 69978, Israel}
36: 
37: \email{meirs@ccsg.tau.ac.il}
38: 
39: \vspace{2cm}
40: 
41: \and
42: 
43: \vspace{2cm}
44: 
45: \author{\bf Yoel Rephaeli}
46: 
47: \affil{School of Physics and Astronomy, Tel Aviv University, Tel 
48: Aviv, 69978, Israel, \\and\\ Center for Astrophysics and Space 
49: Sciences, University of California, San Diego, La Jolla, 
50: CA\,92093-0424}
51: 
52: \email{yoelr@noga.tau.ac.il}
53: 
54: \begin{abstract}
55: 
56: Use of the Sunyaev-Zeldovich effect as a precise cosmological probe
57: necessitates a realistic assessment of all possible contributions to
58: Comptonization of the cosmic microwave background in clusters of 
59: galaxies. We have calculated the additional intensity change due to 
60: various possible populations of energetic electrons that have been 
61: proposed in order to account for measurements of intracluster radio,
62: nonthermal X-ray and (possibly also) EUV emission. Our properly 
63: normalized estimates of (the highly model dependent value of) the 
64: predicted intensity change due to these electrons is well below 
65: $\sim 6\%$ and $\sim 35\%$ of the usual Sunyaev-Zeldovich effect due 
66: to electrons in the hot gas in Coma and A2199, respectively. These 
67: levels constitute high upper limits since they are based on energetic 
68: electron populations whose energy densities are {\it comparable} to 
69: those of the thermal gas. The main impact of nonthermal 
70: Comptonization is a shift of the crossover frequency (where the thermal 
71: effect vanishes) to higher values. Such a shift would have important 
72: consequences for our ability to measure cluster peculiar velocities 
73: from the kinematic component of the Sunyaev-Zeldovich effect.
74: \end{abstract}
75: 
76: \section{Introduction}
77: 
78: The Sunyaev-Zeldovich (S-Z) effect is a small intensity change that 
79: results from Compton scattering of the cosmic microwave background 
80: (CMB) radiation by electrons in the hot gas in clusters of galaxies 
81: (Zeldovich \& Sunyaev 1969, \sz 1972). The effect constitutes a unique 
82: cluster and cosmological probe (for reviews, see Rephaeli 1995a, 
83: Birkinshaw 1999), whose great potential is beginning to be realized in 
84: recent years following the major observational progress in obtaining 
85: sensitive images of the effect by (mostly) interferometric arrays 
86: (Jones et al. 1993, Carlstrom et al. 2000, Udomprasert, Mason, \&
87: Readhead 2000, and the review by Carlstrom \ea 2001)
88: and the increasingly more precise values of the Hubble constant 
89: ($H_0$) that have been deduced from S-Z and X-ray measurements. For 
90: example, a fit to data from 33 cluster distances 
91: yields $H_0 = 58$ km s$^{-1}$ Mpc$^{-1}$ (in a flat cosmological 
92: model), with direct observational errors of $\pm 5\%$ (Carlstrom \ea 
93: 2001), but with a much larger level of systematic uncertainties. Among 
94: others, the latter include simplified modeling of the properties of 
95: the hot intracluster (IC) gas, and cluster morphology. These are the 
96: main limitations to the use of the S-Z effect as a precise cosmological 
97: probe, and as such give further motivation for in-depth studies of 
98: the cluster environment. 
99: 
100: Energetic non-thermal (NT) electrons whose pressure is not negligible 
101: compared to the thermal gas pressure constitute an aspect of IC 
102: phenomena with possibly appreciable ramifications for precision S-Z 
103: measurements. The presence of significant energetic electron 
104: populations in many clusters has been known from measurements of 
105: diffuse IC radio emission, and recently also from NT X-ray emission 
106: in a few clusters (Rephaeli, Gruber \& Blanco 1999, hereafter RGB, 
107: Fusco-Femiano \ea 1999, Kaastra et al. 1999, Gruber \& Rephaeli 2002). NT 
108: electrons produce an additional degree of Comptonization which 
109: amounts to a small intensity change ($\Delta I_{nt}$) that must be 
110: accounted for, particularly in the measurement of $H_0$ from the thermal 
111: component, and peculiar cluster velocities from the kinematic component 
112: of the S-Z effect. Relativistic generalizations (Rephaeli 1995b, Challinor 
113: \& Lasenby 1998, Sazonov \& Sunyaev 1998) of the original non-relativistic 
114: calculations of \sz (1972, 1980) have now been performed to a sufficiently 
115: high level of accuracy, including also terms of order $\tau^2$
116: (Nozawa, Itoh, \& Kohyama 1998, Itoh \ea 2000, Shimon \& Rephaeli
117: 2002), where $\tau$ is the Thomson optical depth of the cluster.
118: In calculating the effect of multiple scatterings the finite size of the 
119: cluster has to be explicitly accounted for; this has been done in the 
120: Monte-Carlo simulations of Molnar \& Birkinshaw (1999).
121: The \rel treatment provides the theoretical basis for calculation also 
122: of $\Delta I_{nt}$.
123: 
124: 
125: CMB Comptonization by NT electrons was first assessed for conditions 
126: in lobes (McKinnon, Owen, \& Eilek 1991) and cocoons (Yamada \& Sugiyama 1999)
127: of radio galaxies, with its possible use to measure their electron 
128: pressure. The effect of such electrons in clusters was recently 
129: considered in some more detail (Blasi \& Colafrancesco 1999, Ensslin 
130: \& Kaiser 2000, Blasi, Olinto, \& Stebbins 2000).
131: Clearly, the higher the electron
132: pressure, the higher is the degree of Comptonization induced by the 
133: electrons, and because models of NT electrons vary greatly in energy 
134: density, estimates of their impact on the CMB range from a negligible 
135: level, to a very substantial fraction of the thermal S-Z effect due to 
136: the hot gas. It is quite essential to study NT electron 
137: populations in order to {\it realistically} determine the spectral 
138: and spatial profiles of $\Delta I_{nt}$ in clusters in which extended 
139: radio emission has been measured. This may lead to ways by which their 
140: impact on S-Z work can be minimized, and eventually the Comptonization 
141: induced by NT electrons could perhaps even be used as a diagnostic tool 
142: of these electrons. 
143: 
144: In this paper we perform an exact calculation of the degree of 
145: Comptonization predicted in a range of NT electron models that have 
146: been proposed to explain radio, EUV, and hard X-ray emission in 
147: clusters. Our estimates of $\Delta I_{nt}$ in Coma and A2199 
148: are based on recent results on the electron populations in 
149: these clusters, and the main features of the resulting spectral change 
150: are contrasted with those of the thermal component of the S-Z effect.
151: 
152: \section{Non-thermal Electron Populations}
153: 
154: The main evidence for \rel electrons and magnetic fields in clusters 
155: comes from measurements of extended IC regions of radio emission (in 
156: the frequency range $\sim 0.04-4$ GHz) with spectral indices and 
157: luminosities in the range $\sim 1-2$, and $10^{40.5}$ -- $10^{42}$ 
158: erg/s ($H_0$ = 50 km s$^{-1}$ Mpc$^{-1}$), respectively. Many of 
159: the radio emitting regions detected in over 30 clusters (Giovannini,
160: Tordi, \& Feretti 1999, Giovannini \& Feretti 2000) are central, with
161: sizes in the range $\sim 1-3$ Mpc. 
162: The energy range of the emitting electrons depends on the value of the 
163: mean, volume-averaged magnetic field, a quantity which is not known 
164: very well, but is likely to be in the range $\sim 0.1 - 1$ $\mu$G (for 
165: a recent review, see Rephaeli 2001). The range of electron energies 
166: implied from these measurements is roughly $1-100$ GeV, but electrons 
167: with energies both below and above this range are also expected on 
168: theoretical grounds.
169: 
170: Relativistic electrons produce X-ray emission in a wide spectral 
171: range by Compton scattering off the CMB (Rephaeli 1979) and by NT 
172: bremsstrahlung. This emission has quite possibly been measured already 
173: in Coma (Rephaeli, Gruber \& Blanco 1999, Fusco-Femiano \ea 1999), 
174: A2199 (Kaastra \ea 1999), A2256 (Molendi, De Grandi, \& Fusco-Femiano
175: 2000), and A2319 (Gruber \& Rephaeli 2002) by the RXTE and BeppoSAX
176: satellites. These were spectral measurements; 
177: the PCA, HEXTE (both aboard RXTE) and PDS (aboard BeppoSAX) experiments 
178: do not have the adequate spatial resolution. Thus, we have no spatial 
179: information on NT electrons from X-ray measurements, and only rudimentary 
180: knowledge of the morphology of the radio emission (which, however, involves 
181: also the unknown spatial distribution of the magnetic field). Therefore, 
182: in order to avoid the need for introducing unknown parameters, we 
183: characterize electron populations in terms of the spectral distribution 
184: of their total number, ignoring their spatial profiles in the central 
185: $\sim 1$ Mpc region where they mostly reside.   
186: 
187: Of the various proposed NT IC electron models (see, \eg, Sarazin 1999) 
188: we focus here on those that have been contrasted with actual observational 
189: data, with the electron distribution appropriately normalized by the 
190: determination of both their total number and the power-law index, $q$. 
191: Some of the proposed electron models were motivated by the presumption 
192: that NT X-ray emission could also originate from NT 
193: bremsstrahlung by energetic -- though not necessarily highly relativistic 
194: -- electrons, and by an attempt to account also for observed EUV emission 
195: from a few clusters. This emission is said to be NT (\eg, Sarazin \& Lieu 
196: 1998, Bowyer \ea 1999), possibly by a population of low energy electrons. 
197: The full electron distribution could be described either as a sum of two
198: separate parts (thermal plus NT), or by a `super Maxwellian', consisting
199: of a truncated Maxwellian with a power law tail (Blasi, Olinto, \&
200: Stebbins 2000) `sawn' 
201: together at a given energy (\eg, $\sim 3kT_e$, where $T_e$ is the electron 
202: [and gas] temperature). We consider here four specific models whose 
203: parameters in Coma and A2199 have been determined from radio, EUV, and 
204: NT X-ray measurements. 
205: 
206: The simplest model for the electron momentum distribution is a power-law 
207: over a sufficiently wide range so as to explain both the observed radio 
208: and possibly Compton-produced X-ray emission. We express the electron 
209: spectrum in terms of the (differential) number, $N(p)$, per unit 
210: dimensionless momentum, $p=\beta\gamma$, where $\gamma$ and $\beta$ are 
211: the Lorentz factor and dimensionless velocity, $\beta=v/c$,
212: respectively,
213: \begin{eqnarray}
214: N(p)&=&A p^{-q},    \quad   p_1 \leq p  \leq p_2 \cr
215: \cr
216: A&=&{N_{12}(q-1) \over p_1 ^{-(q-1)} - p_2 ^{-(q-1)}}.
217: \end{eqnarray}
218: The limiting momenta $p_1$ and $p_2$ are the lower and higher values of 
219: $p$ that correspond to the {\it observed radio frequency range}, and 
220: $N_{12}$ is the total number of electrons with energies in this 
221: specific interval, $[p_{1}, p_{2}]$. Details of such a model were 
222: worked out long ago (Rephaeli 1977, 1979) and will not be repeated here. 
223: Suffice it to say that the {\it full} momentum range is substantially 
224: uncertain, especially its low end which is of particular relevance to 
225: our discussion here. 
226: 
227: A simple way to obtain the lower energy extension of the above 
228: distribution is to assume that steady state is attained, whereby 
229: electrons are continually accelerated to compensate for radiative 
230: energy losses -- defined here in terms of $dp/dt$ -- and a population 
231: of lower energy electrons is built up. The latter can be determined by 
232: taking into account the dominant rate of energy loss at energies well 
233: below $\sim 150$ MeV, electronic (Coulomb) excitations, given by 
234: (Rephaeli 1979)
235: \begin{eqnarray}
236: b_{ee}=1.2\times
237: 10^{-12}n_{e}\left[1.0+\frac{\ln\left(\sqrt{1+p^{2}}/n_{e}\right)}{75}
238: \right]\rm{s}^{-1},
239: \end{eqnarray}
240: where $n_{e}$ is the (thermal) electron number density. At higher 
241: energies Compton-synchrotron losses dominate; these occur at a combined 
242: rate (\eg, Rephaeli 1979)
243: \be
244: b_{cs}=1.37\times 10^{-20}[(1+z)^{4} + 9.5\times 10^{-2} (B/1\mu\rm{G})^{2}] 
245: \,  \rm{s}^{-1},
246: \ee
247: where $z$ is the cluster redshift, and $B$ is the mean, volume-averaged
248: value of the magnetic field. The steady state solution is then (Rephaeli 
249: 1979) 
250: \begin{eqnarray}
251: N_{I} \left(p\right)&=&\frac{Ab_{cs}}{b\left(p\right)}
252: \left(1+p^{2}\right)^{-\frac{\left(q-2\right)}{2}}\qquad\left(p_{1}\leq
253:   p\leq
254:   p_{2}\right)\nonumber\\&=&\frac{Ab_{cs}}{b\left(p\right)}
255: \left(1+p_{1}^{2}\right)^{-\frac{\left(q-2\right)}{2}}\qquad\left(p\leq
256: p_{1}\right),\nonumber\\
257: \end{eqnarray}
258: where
259: $b \equiv dp/dt = b_{ee}+b_{cs}\left(1+p^{2}\right)$.
260: Clearly, the observed radio frequency range implies that $p_{1}\gg 1$.
261: (Note that another possible energy loss mechanism is scattering by Alfven 
262: waves, which the electrons themselves can excite if their spatial 
263: distribution is somewhat anisotropic. Because of the substantial uncertainty 
264: in estimating this loss mechanism [\eg, Rephaeli 1979] it will not be 
265: considered here.) Parameters of this model were determined from RXTE 
266: measurements of the Coma cluster, assuming the emission is from Compton 
267: scattering of \rel electrons off the CMB (Rephaeli, Gruber, \& Blanco 1999), 
268: and from BeppoSAX measurements of A2199 (Kaastra \ea 1999). This 
269: appropriately extended electron population which has a power-law form 
270: at low and high energies, but with indices whose values differ by unity, 
271: is our first model. The parameters of all the models considered here are 
272: given in Table 1. 
273: 
274: \begin{table*}[h]
275: \small
276: \begin{center}
277: \begin{tabular}{llllll}
278: \tableline
279: \tableline
280: Cluster&Model &Model
281: details&$\qquad N_{0}$&$\quad N_{nt}^{tot}$&$N_{nt}^{tot}/N_{th}^{tot}$\cr
282: \tableline
283: Coma & && &\cr
284:      &I&Extended power law &$2.275\times 10^{72}$&$1.09\times
285:      10^{65}$&$4.70\times 10^{-7}$\cr
286:      &II&Shock accelerated &$4.270\times
287:      10^{67}$&$2.98\times 10^{68}$&$1.28\times 10^{-3}$\cr
288:      &III&Cooling electrons &$5.238\times 10^{62}$&$3.13\times
289:      10^{65}$&$1.35\times 10^{-6}$\cr
290:      &IV&Simple power law q=3.68&$8.970\times 10^{68}$&$9.27\times
291:      10^{69}$&$0.04$\cr
292: &&\cr
293: A2199 & && &\cr
294:      &I&Extended power law&$2.306\times 10^{68}$&$1.66\times
295:      10^{64}$&$2.10\times 10^{-7}$\cr
296:      &II&Shock accelerated&$4.240\times
297:      10^{67}$&$6.22\times 10^{68}$&$7.90\times 10^{-3}$\cr
298:      &III&Cooling electrons &$5.238\times 10^{62}$&$3.13\times
299:      10^{65}$&$3.95\times 10^{-6}$\cr
300:      &IV&Power law q=3.33&$5.420\times 10^{68}$&$6.47\times
301:      10^{69}$&$0.082$\cr
302: \tableline
303: \end{tabular}
304: \end{center}
305: \caption{Models for NT electrons in the Coma and A2199 clusters.}
306: \end{table*}
307: 
308: Two other forms for $N(p)$ are discussed by Sarazin \& Kempner (2000). In 
309: the first model electrons are accelerated to relativistic energies by 
310: shell-type supernova remnants (Baring \ea 1999). When the back reaction of 
311: the accelerating electrons on the shock structure is taken into account, 
312: then this presumably leads to a modified power-law form at small momenta, 
313: resulting in the following distribution, 
314: \begin{eqnarray}
315: N_{II} \left(p\right)=\frac{N_{0}p^{-2}}{1+p_{c}^{2}}\left[1+
316: \left(\frac{p_{c}}{p}\right)^{2}\right].
317: \end{eqnarray}
318: We assume $p_{c}=1$ in normalizing the model.
319: Explicit account of Coulomb losses yields the third distribution 
320: \begin{eqnarray}
321: N_{III}\left(p\right)=\frac{2N_{0}p^{2}}{1+p^{2}}.
322: \end{eqnarray}
323: $N_{0}$ is derived from comparison of the models with observations 
324: and assuming that the hard X-ray emission mechanism is NT bremsstrahlung. 
325: The electron distribution in Equation (6) has an upper momentum cutoff 
326: at $p=300$. 
327: 
328: Sarazin \& Kempner (2000) also propose a simple power law model which 
329: extends to lower energies {\it with no change in the index}. This is the 
330: fourth (IV) model explored here. The observationally deduced parameters for 
331: the above four models are listed in Table 1; in these models, the lower 
332: momentum cutoff is $p_{1}=\sqrt{(1+3\Theta)^{2}-1}$, where 
333: $\Theta=kT_{e}/mc^{2}$. 
334: 
335: 
336: \section{Comptonization}
337: 
338: A detailed description of the Comptonized spectrum resulting from 
339: scattering of the CMB by a thermal, non-\rel population of electrons 
340: was given by \sz (1972, 1980). A more accurate treatment of the process 
341: requires \rel generalization due to the rapid motion of electrons in 
342: the hot ($kT_{e} \leq 15$ keV) IC gas, as has been shown explicitly by 
343: Rephaeli (1995a). 
344: A fully relativistic treatment (but still in the Thomson limit, and 
345: with electron recoil and induced scattering safely ignored)
346: is obviously required in order to calculate the Comptonized spectrum 
347: resulting from scattering of the radiation by a NT energetic electron 
348: population. We use the approach adopted in the latter paper (see also 
349: Rephaeli \& Yankovitch 1997).
350: 
351: The probability of scattering of an incoming photon moving originally  
352: in the direction specified by $\mu_{0}=\cos\theta_{0}$, to the 
353: direction $\mu'_{0}=\cos\theta'_{0}$, is (Chandrasekhar 1950)
354: \begin{eqnarray}
355: f\left(\mu_{0},\mu'_{0}\right)=\frac{3}{8}\left[1+
356: \mu_{0}^{2}\mu'^{2}_{0}+\frac{1}{2}\left(1-\mu_{0}^{2}\right)\left(1-
357: \mu'^{2}_{0}\right)\right],
358: \end{eqnarray}
359: where the subscript $0$ denotes quantities in the electron rest
360: frame.
361: The logarithmic frequency shift in the scattering is 
362: \begin{eqnarray}
363: s\equiv\ln\left(\nu'/\nu\right)=\ln\left(\frac{1+\beta\mu'_{0}}
364: {1+\beta\mu_{0}}\right) \,,
365: \end{eqnarray}
366: where $\beta$ is the dimensionless electron velocity in the CMB frame. 
367: The probability for scattering is conveniently written in 
368: terms of the variables $\beta$ and $t=e^{s}$ (Wright 1979), 
369: \begin{eqnarray}
370: \mathcal{P} \left(s,\beta\right)=\frac{1}{2\gamma^{4}\beta}
371: \int_{\mu_{1}}^{\mu_{2}}\frac{e^{s}f\left(\mu_{0},
372: \mu'_{0}\right)}{\left(1+\beta\mu_{0}\right)^{2}}d\mu_{0},
373: \end{eqnarray}
374: where $\mu_{1}$ and $\mu_{2}$ are given by
375: \begin{eqnarray}
376: \mu_{1}=\left\{
377: \matrix{\frac{e^{-s}\left(1-\beta\right)-1}{\beta}&s\leq 0
378: \cr
379:    -1&s\geq 0}\right.
380: \end{eqnarray}
381: \begin{eqnarray}
382: \mu_{2}=\left\{
383: \matrix{1&s\leq 0
384: \cr
385:   \frac{e^{-s}\left(1+\beta\right)-1}{\beta}&s\geq 0 }\right..
386: \end{eqnarray}
387: Integration over $\mu_{0}$ in Equation (9) yields
388: \begin{eqnarray}
389: \mathcal{P} \left(t,p\right)&=&-\frac{3|1-t|}{32p^{6}t}\left[1+
390: \left(10+8p^{2}+4p^{4}\right)t+t^{2}\right]\nonumber\\
391: &+&\frac{3\left(1+t\right)}{8p^{5}}\left[\frac{3+3p^{2}+p^{4}}{\sqrt{1
392: +p^{2}}}-\frac{3+2p^{2}}{2p}\left(2 \sinh^{-1}\left(p\right)
393: -|\ln\left(t\right)|\right)\right],
394: \end{eqnarray}
395: where
396: \begin{eqnarray}
397: |\ln\left(t\right)|\leq2 \sinh^{-1}\left(p\right).
398: \end{eqnarray}
399: 
400: Compton scattering results in a frequency shift, $x\rightarrow x'$, where $x$ 
401: is the dimensionless frequency $x=h\nu/kT$, and $T$ is the CMB
402: temperature. The corresponding change of the photon occupation number through 
403: a pathlength along the cluster, $\Delta n(x)$, is obtained by integrations 
404: over the scattering probability and electron momentum distributions,
405: \begin{eqnarray}
406: \Delta n\left(x\right)=\tau\int_{p_{1}}^{p_{2}}\int_{t_{min}}^{t_{max}}
407: dpdt \frac{N(p)}{N_{tot}}\mathcal{P} \left(t,p\right)\left(\frac{1}{e^{xt}-1}
408: -\frac{1}{e^{x}-1}\right),
409: \end{eqnarray}
410: where $\tau$ is the optical depth of the cluster to Compton scattering, 
411: and $N_{tot}$ is the total number of electrons in the cluster.
412: The measured quantity is the change in intensity, 
413: $\Delta I(x)/i_{0} = x^{3}\Delta n(x)$, where $i_{0}=2(kT)^{3}/(hc)^{2}$. 
414: We use the latter equation to calculate the additional intensity change 
415: induced by energetic electrons, $\Delta I_{nt}(x)$, in the electron models 
416: described in the previous section, with parameter values sampling the 
417: observationally deduced ranges, as specified in Table 1 for all the four 
418: electron models.
419: 
420: If Comptonization by NT electrons is not taken into account it could 
421: affect the observationally deduced value of the crossover frequency, 
422: $x_{0}$, defined as the frequency at which the {\it purely thermal} 
423: effect vanishes. The exact value of $x_{0}$ has practical
424: significance, since observations near this frequency are optimal for
425: the determination of cluster peculiar velocities (Rephaeli \& Lahav 1991)
426: from measurements of the kinematic component of the S-Z effect.
427: We have calculated the modified value of the frequency where now the
428: {\it sum} of the thermal and NT intensity changes vanishes. For this we
429: used the analytic expression of Nozawa, Itoh, \& Kohyama(1998)
430: for the thermal and
431: kinematic components of the S-Z effect, and computed $x_{0}$ using the
432: measured values of the temperatures in Coma and A2199.
433: 
434: \section{Results}
435: 
436: The impact of each of the above four NT electron models depends very much 
437: on the total number of these electrons as determined by radio, NT EUV 
438: and X-ray emission. The thermal gas parameters are based on X-ray 
439: measurements. For Coma, recent XMM measurements yield $kT_{e}=8.2 \pm 0.4$ 
440: keV for the gas temperature (Arnaud \ea 2001), and (since 
441: quantitative results for the gas density profile from either XMM or 
442: {\it Chandra} are not yet available) we have taken the ROSAT deduced 
443: values (Mohr, Mathiesen, \& Evrard 1999), $n_{e0} \simeq (3.12. \pm 0.04) 
444: \times 10^{-3}$ cm$^{-3}$, $r_{c}=0.366$ Mpc, and $\beta_{n} \simeq 0.705$, 
445: for the central electron density, core radius, and index in the expression 
446: for (the commonly used) density profile, 
447: $n_{e}(r) = n_{e0}(1+r^{2}/r_{c}^{2})^{-3\beta_{n}/2}$, respectively. 
448: The corresponding values in A2199 are $kT_{e}=4.7$ keV, central
449: electron density $n_{e0}=7.21\times 10^{-3}\rm{cm}^{-3}$,
450: $r_{c}=0.196$ Mpc, and $\beta_{n}=0.78$ (Kaastra et al. 1999). These
451: values are based on $H_0 = 50$ km s$^{-1}$ Mpc$^{-1}$. 
452: 
453: Clearly, most of the intensity change $\Delta I_{nt}(x)$ is due to the 
454: more numerous low energy electrons. More specifically, a typical CMB 
455: photon has dimensionless frequency $\bar{x}=2.701$, and the scattered 
456: photon frequency is on average
457: \begin{eqnarray}
458: \bar{\frac{x'}{x}}=1+\frac{4}{3}p^{2}.
459: \end{eqnarray}
460: The observationally relevant range of values of $x$ is at most $x \leq 20$, 
461: and since the (the 68\% likelihood) interval of rms values of $x'$, 
462: $\Delta x' \simeq x\sqrt{2 p^{2}/3 + 46p^{4}/45}$, is very wide for 
463: $p>1$, it follows that electrons with momenta much larger than at most 
464: a few tens are largely irrelevant for boosting photons to frequencies
465: such that $x \leq 20$. 
466: (Obviously, {\it all} electrons contribute to the value of the Thomson 
467: optical depth and the decrement on the Rayleigh-Jeans side.)
468: 
469: We have calculated $\Delta I_{nt}(x)/i_{0}$ and total electron energies 
470: in the extended power-law model described by equation (4), using the
471: observationally deduced model parameters in Coma and A2199. We emphasize 
472: that the basic model is a power law at high energies ($\geq 1$ GeV) with 
473: the index determined from radio measurements. The distribution is then 
474: appropriately extended to lower energies (by taking electronic excitations 
475: losses into account)
476: resulting in a change in the value of the index at 
477: low energies. The power law indices and overall normalizations were taken 
478: from Rephaeli, Gruber \& Blanco (1999) for Coma, and Kaastra \ea (1999) 
479: for A2199. Based on these parameters, the deduced values of 
480: $\Delta I_{nt}(x)/i_{0}$ are very small in comparison with the
481: magnitude of the 
482: thermal S-Z effect in these clusters as predicted from the measured 
483: values of the gas temperature and density. In Coma the S-Z effect was 
484: actually measured by Herbig et al. (1995) using the 
485: OVRO 5.5 m telescope, and more recently by De Petris \ea (2002) using 
486: the MITO telescope. In the former paper the temperature change in the 
487: center of Coma was reported to be $\Delta T = -505 \pm 92$ $\mu$K at 
488: $32$ GHz ($x \simeq 0.56$), a value which is consistent with that 
489: predicted based on the X-ray deduced parameters. A somewhat lower 
490: value was deduced by De Petris \ea (2002). Our calculated values 
491: of $\Delta I_{nt}(x)/i_{0}$ due to NT electrons are smaller than
492: 0.001\% of the predicted thermal effect in Coma and A2199. Clearly, 
493: the implied shift in the value of the crossover frequency is also 
494: negligible. The very small impact of NT electrons in this model is not 
495: surprising given the relatively low energy content of these electrons,
496: $\sim 0.8\%$
497: of the thermal electron energy in Coma and A2199.
498: 
499: Next we calculated the degree of Comptonization in the shock accelerated 
500: and cooling electrons models (II and III) described in equations (5) and 
501: (6). We have normalized these models to match the observed EUV emission, 
502: at a luminosity level of roughly $5\times 10^{43}\ \rm{erg}\ \rm{s}^{-1}$ over the 
503: band 0.065-0.245 keV in both Coma and A2199. In the shock accelerated 
504: electron model the calculated levels of $\Delta I_{nt}(x)$ are $<0.5\%$ 
505: and $\leq 3\%$ of the corresponding intensity change due to thermal 
506: electrons, and the shift in the value of the crossover frequency is 
507: also quite small. Calculated values of the energy in NT electrons in 
508: this model are $\sim$3\% and 13\%-16\% (for $p_{c}=0.3-1.0$) of the 
509: energy in thermal electrons in Coma and A2199, respectively. In the 
510: cooling electron model, the degree of NT Comptonization is negligible, 
511: less than $0.003\%$ of the magnitude of the thermal S-Z effect in both 
512: Coma and A2199. This is mainly due to the small number of NT electrons 
513: in this model, $\sim$ 0.1\% of that in the shock accelerated electron 
514: model. The NT electron energy constitutes 0.9\% and 4.2\% of 
515: the electron thermal energy in the two clusters. Our calculated 
516: quantities are listed in Table 2. The first column is the cluster name 
517: (or Abell number), the second is the model number, third is the ratio 
518: of (total) energy in NT to energy in thermal electrons. In the 
519: fourth column we list the net energy deposition rate (in keV/Gyr) by 
520: NT electrons, and in the
521: fifth the value of the crossover frequency; 
522: values of the error in the peculiar velocity and the ratio 
523: $\Delta I_{nt}(x)/ \Delta I$ are listed in last two columns.
524: 
525: \begin{table*}[h]
526: \small
527: \begin{center}
528: \begin{tabular}{llllllllll}
529: \tableline
530: \tableline
531: Cluster & Model & $E_{nt}/E_{th}$ & $\quad\rm{dE/dt}$ & $x_{nt}$ & 
532: $\quad\Delta v>$ & $\Delta I_{nt}/ \Delta I$ \cr
533: &&\%&(\rm{keV/Gyr})& &(\rm{km/sec})&\qquad\%\cr
534: \tableline
535: Coma &&&&&&\cr
536:      &Thermal& & &3.9000&&&\cr
537:      &I&0.79&$-0.06$&3.9000&-&$1.4\times 10^{-4}$\cr
538:      &II&3.18&$10.80$&3.9020&-&0.41\cr
539:      &III&0.90&$-0.04$&3.9000&-&$4.7\times 10^{-4}$\cr
540:      &IV&21.59&$296.86$&3.9216&-110&6.8\cr
541: &&&&&&\cr
542: A2199&      &    &    &&&\cr
543:      &Thermal& & &3.8700&&&\cr
544:      &I&0.74&$-0.15$&3.8701&-&$1.2\times 10^{-4}$\cr
545:      &II&15.87&$58.87$&3.8803&-&2.97\cr
546:      &III&4.23&$-0.06$&3.8702&-&$2.5\times 10^{-3}$\cr
547:      &IV&64.17&$626.39$&3.9385&-210&34.5\cr
548: \tableline
549: \end{tabular}
550: \end{center}
551: \caption{The impact of the four energetic electron models (see the text for 
552: definitions of the listed quantities).}
553: \end{table*}
554: 
555: The much higher NT electron energy content in the single power law (model 
556: IV) suggested by Sarazin \& Kempner (2000), $\sim 20\% - 22\%$ (for $q=2.92 - 
557: 3.68$) in Coma, and $\sim 64\% - 184\%$ (for $q=2.2 - 3.33$) in A2199, 
558: result in substantial degrees of additional Comptonization. Using values 
559: of the parameters as deduced by Sarazin \& Kempner (2000), listed in 
560: Table 1, we calculate the change of intensity, $\Delta I_{nt}(x)/i_{0}$, 
561: shown in Figure 1, and listed in Table 2. It is clear from this figure 
562: that $\Delta I_{nt}(x)$ can reach an appreciable fraction of $\Delta I(x)$. 
563: For example, near the peak of the Comptonized Planckian, at $x \sim 6.5$, 
564: $\Delta I_{nt} / \Delta I$ is $\leq 13\%$ and $\leq 35\%$ for Coma and A2199, 
565: respectively. The implied shift in the value of the crossover frequency in 
566: this model, $\sim 2$ GHz in Coma, and $\geq 4$ GHz in A2199, would also 
567: be observationally relevant. Such a systematic shift would introduce an 
568: error in the deduced value of the peculiar velocity. The implied 
569: error (which does not depend on the velocity, if the very small quadratic 
570: dependence of $\Delta I(x)$ on the velocity is ignored) amounts to 
571: $\sim 110$ km/s for Coma ($q= 2.92 - 3.68$), and $\sim 210$ km/s 
572: ($q=3.33$) and $320$ km/s ($q=2.2$) for A2199 (values lower than 
573: 100 km/s are not shown in Table 2).
574: 
575: \begin{figure}[h]
576: \plottwo{f1.eps}{f2.eps}
577: \caption{The spectrum of the additional Comptonization due to energetic 
578: electrons in the models proposed by Sarazin \& Kempner (2000) for the 
579: Coma (left panel) and A2199 (right panel).}
580: \end{figure}
581: 
582: 
583: Finally, we have calculated the heating rate of the hot gas by the NT 
584: electrons through Coulomb interactions (Rephaeli 1979, Rephaeli \& Silk 
585: 1995) in the four models considered in this paper. Energy deposition rates 
586: into the gas were estimated by taking a mean gas density in the central 
587: 1 Mpc region of Coma and A2199. Subtraction of the observed cooling rate 
588: due to thermal bremsstrahlung (X-ray) emission then yields the net rate of 
589: temperature change. Values of the latter quantity (in keV/Gyr) are listed 
590: in Table 2. 
591: 
592: \section{Discussion}
593: 
594: We have briefly discussed models for NT energetic electrons and 
595: their predicted effect on the CMB based on parameters that were 
596: deduced from radio, EUV, and X-ray measurements of Coma and A2199. 
597: Our immediate objective has been the exact calculation of the additional
598: degree of Comptonization due to these electrons, and its possibly relevant 
599: observational consequences. Only electron models that have direct
600: observational motivation were considered. Since all these models have
601: basically a power-law form, it is clear that their impact on the CMB is 
602: largely due to the more abundant lower energy electrons (except for the 
603: cooling electrons model). However, the interpretation of the EUV and X-ray 
604: emissions as NT bremsstrahlung and Compton scattering are not secure, so 
605: there is considerable uncertainty in quantifying electron densities. 
606: Moreover, since low energy electrons lose energy mostly by non-radiative 
607: coupling to the thermal gas, the uncertainty in the estimation of their 
608: density is particularly substantial. Only the presence of \rel electrons is 
609: well established from measurements of extended regions of radio emission 
610: in many clusters (including Coma, but not A2199). Clearly, radio 
611: measurements yield the electron density only if we have an independent 
612: estimate of the mean, volume-averaged value of the magnetic field in the 
613: emitting region. Estimates of the mean IC field are few and uncertain 
614: (Newman, Newman \& Rephaeli 2002).
615: 
616: Due to the lack of detailed information on IC energetic electron 
617: populations, relevant theoretical considerations are of particular 
618: interest. First, since it is virtually always the case that particle 
619: acceleration mechanisms tap only a small fraction of the energy at the 
620: source, energy density in NT electrons is likely to be only a small 
621: fraction of the gas (which is a significant mass component of clusters) 
622: thermal energy density. Realistically, therefore, the thermal energy 
623: density in clusters is expected to set an absolute upper limit to the 
624: energy density in NT electrons, a limit which is very unlikely to be 
625: reached. Second, the energetic electron population is linked to the 
626: gas mainly by Coulomb coupling that results in energy transfer from 
627: energetic electrons (and protons) to the gas. This coupling and the 
628: resulting heating set stringent constraints on the attainment of 
629: steady state in general, and the density of low energy NT electron 
630: models in particular. 
631: 
632: We have calculated the ratio of total energy in NT electrons to that 
633: in thermal electrons in the four energetic electron models 
634: considered in this paper; for the first three models the calculated 
635: values (in Table 2) are low. For the power law model of Sarazin \& Kempner 
636: (2000), the corresponding values are very high ($\sim 22\%$ in Coma, and 
637: $\geq 64\%$ in A2199) and quite unrealistic. Such large energy contents 
638: would also imply a high rate of energy transfer to the gas, and heating at 
639: a rate higher than the rate of cooling by emission of thermal radiation. 
640: This without even considering the additional heating by IC energetic 
641: protons (whose energy density in the Galaxy is higher than that of \rel 
642: electrons). 
643: 
644: In fact, the single power law model of Sarazin \& Kempner (2000) is not 
645: only unappealing from an energetic point of view, but is also questionable 
646: on a more basic physical ground: As we have stated in Section 2 -- based 
647: on the original work of Rephaeli (1979) -- correct extrapolation of the 
648: \rel electron energy spectrum to low ($<100$ MeV) energies must account 
649: for electronic Coulomb excitations, the dominant energy loss mechanism 
650: at such energies for typical gas densities in the central regions of 
651: clusters. The very weak energy dependence of this mechanism (in contrast 
652: with the quadratic dependence of the Compton-synchrotron energy loss rate) 
653: flattens the spectrum at low energies, resulting in a much smaller number 
654: of electrons than would have been predicted otherwise. We conclude that 
655: this model is non-viable; therefore, the high degree of implied 
656: Comptonization and appreciable shift in the value of $x_0$ that 
657: are predicted in this model are at best very high upper limits to 
658: the impact of NT electrons on measurements of the S-Z effect.
659: 
660: Our basic result in this paper is that exact calculation of the impact 
661: of realistically normalized models of NT electrons in Coma and A2199 yields 
662: levels of the degree of Comptonization by such electrons that are only a 
663: negligible fraction of the corresponding S-Z effect due to the hot IC 
664: gas. It is clear from our discussion that this result is generally valid 
665: since the energy density in NT electrons is not likely to be a significant 
666: fraction of the thermal energy density. In clusters with a significant NT 
667: electron population the added Comptonization due to these electrons 
668: would clearly constitute a source of confusion in the analysis of S-Z 
669: measurements. Since there are little or no differences in the spectral 
670: CMB signatures of thermal and NT electrons, high spatial resolution X-ray, 
671: S-Z and radio measurements of these clusters would be needed in order 
672: to minimize this confusion when (as expected) the spatial profiles of the 
673: two electron populations are found to be detectably different.
674: 
675: \acknowledgments
676: Useful comments made by the referee on an earlier version of the paper
677: are gratefully acknowledged. This research has been supported by the
678: Israeli Science Foundation grant 729$\backslash$00-02 at Tel Aviv University. 
679: \begin{thebibliography}{99}
680: \bibitem{1} Arnaud, M.~et al.\ 2001, \aap, 365, L67
681: \bibitem{2} Baring, M.~G., Ellison, D.~C., Reynolds, S.~P., Grenier, 
682: I.~A., \& Goret, P.\ 1999, \apj, 513, 311 
683: \bibitem{3} Birkinshaw, M.\ 1999, \physrep, 310, 97
684: \bibitem{4} Blasi, P.~\& Colafrancesco, S.\ 1999, Astroparticle Physics, 
685: 12, 169 
686: \bibitem{5} Blasi, P., Olinto, A.~V., \& Stebbins, A.\ 2000, \apjl, 535, L71 
687: \bibitem{6} Bowyer, S., Bergh{\" o}fer, T.~W., \& Korpela, E.~J.\ 1999, 
688: \apj, 526, 592 
689: \bibitem{61} Carlstrom, J.~E.~\& et al.\ 2000, IAU Symposium, 201,
690:   E48
691: \bibitem{7} Carlstrom, J.E. \ea 2001, astro-ph/0103480
692: \bibitem{8} Challinor, A.~\& Lasenby, A.\ 1998, \apj, 499, 1 
693: \bibitem{9} Chandrasekhar, S.\ 1950, Oxford, Clarendon Press, 1950
694: \bibitem{10} De Petris, M. \ea 2002, astro-ph/0203303
695: \bibitem{11} Ensslin, T.~A.~\& Kaiser, C.~R.\ 2000, \aap, 360, 417 
696: \bibitem{12} Fusco-Femiano, R., dal Fiume, D., Feretti, L., Giovannini, G., 
697: Grandi, P., Matt, G., Molendi, S., \& Santangelo, A.\ 1999, \apjl, 513, L21 
698: \bibitem{13} Giovannini, G., Tordi, M., \& Feretti, L.\ 1999, New 
699: Astronomy, 4, 141 
700: \bibitem{14} Giovannini, G.~\& Feretti, L.\ 2000, New Astronomy, 5, 335 
701: \bibitem{15} Gruber D.E., \& Rephaeli Y. 2002, \apj, in press 
702: (astro-ph/0110512) 
703: \bibitem{16} Herbig, T., Lawrence, C.~R., Readhead, A.~C.~S., \& Gulkis, 
704: S.\ 1995, \apjl, 449, L5 
705: \bibitem{17} Itoh N., Kawana Y., Nozawa S., Kohyama Y.,
706:   astro-ph/0005390
707: \bibitem{151} Jones, M., et al. 1993, Naure, 365, 320
708: \bibitem{18} Kaastra, J.~S., Lieu, R., Mittaz, J.~P.~D., Bleeker, J.~A.~M., 
709: Mewe, R., Colafrancesco, S., \& Lockman, F.~J.\ 1999, \apjl, 519, L119 
710: \bibitem{19} McKinnon, M.~M., Owen, F.~N., \& Eilek, J.~A.\ 1991, \aj, 101, 
711: 2026 
712: \bibitem{20} Mohr, J.~J., Mathiesen, B., \& Evrard, A.~E.\ 1999, \apj, 517, 
713: 627 
714: \bibitem{21} Molendi, S., De Grandi, S., \& Fusco-Femiano, R.\ 2000, \apjl, 
715: 534, L43 
716: \bibitem{211} Molnar, S.~M.~\& Birkinshaw, M.\ 1999, \apj, 523, 78. 
717: \bibitem{22} Newman, W.I., Newman, A.L., \& Rephaeli, Y. 2002, \apj, in press
718: \bibitem{23} Nozawa, S., Itoh, N., \& Kohyama, Y.\ 1998, \apj, 508, 17 
719: \bibitem{24} Rephaeli, Y.\ 1977, \apj, 212, 608 
720: \bibitem{25} Rephaeli, Y.\ 1979, \apj, 227, 364 
721: \bibitem{26} Rephaeli, Y.~\& Lahav, O.\ 1991, \apj, 372, 21 
722: \bibitem{27} Rephaeli, Y.\ 1995a, \araa, 33, 541 
723: \bibitem{28} Rephaeli, Y.\ 1995b, \apj, 445, 33 
724: \bibitem{29} Rephaeli, Y.~\& Silk, J.\ 1995, \apj, 442, 91 
725: \bibitem{30} Rephaeli, Y.~\& Yankovitch, D.\ 1997, \apjl, 481, L55 
726: \bibitem{31} Rephaeli, Y., Gruber, D., \& Blanco, P.\ 1999, \apjl, 511, L21 
727: \bibitem{32} Rephaeli Y. 2001, {\it Astrophysical Sources of High Energy 
728: Particles \& Radiation}, Kluwer, Dordrecht (astro-ph/0105265) 
729: \bibitem{33} Sarazin, C.~L.~\& Lieu, R.\ 1998, \apjl, 494, L177
730: \bibitem{331} Sarazin, C.~L.\ 1999, \apj, 520, 529.  
731: \bibitem{34} Sarazin, C.~L.~\& Kempner, J.~C.\ 2000, \apj, 533, 73 
732: \bibitem{35} Sazonov, S.~Y.~\& Sunyaev, R.~A.\ 1998, \apj, 508, 1
733: \bibitem{36} Shimon M., Rephaeli Y.\ 2002, in preparation 
734: \bibitem{37} Sunyaev, R.~A.~\& Zeldovich Y.~B.\ 1972, \casp, 4, 173
735: \bibitem{38} Sunyaev R.~A.~\& Zel'dovich Y.~B.\ 1980, \mnras, 190, 413
736: \bibitem{381} Udomprasert, P.~S., Mason, B.~S., \& Readhead, A.~C.~S.\
737:   2000, American Astronomical Society Meeting, 197, 110401 
738: \bibitem{39} Wright, E.~L.\ 1979, \apj, 232, 348 
739: \bibitem{40} Yamada, M., Sugiyama, N., \& Silk, J.\ 1999, \apj, 522, 66 
740: \bibitem{41} Zeldovich, Y.~B.~\& Syunaev, R.~A.\ 1969, \apss, 4, 301 
741: \end{thebibliography}
742: 
743: \end{document}
744: 
745: 
746: 
747: 
748: 
749: 
750: 
751: 
752: 
753: