DIFFUSIVE SHOCK ACCELERATION BY MULTIPLE WEAK SHOCKS
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
Journal of the Korean Astronomical Society https://doi.org/10.5303/JKAS.2021.54.3.103 54: 103 ∼ 112, 2021 June pISSN: 1225-4614 · eISSN: 2288-890X Published under Creative Commons license CC BY-SA 4.0 http://jkas.kas.org D IFFUSIVE S HOCK ACCELERATION BY M ULTIPLE W EAK S HOCKS Hyesung Kang Department of Earth Sciences, Pusan National University, Busan 46241, Korea; hskang@pusan.ac.kr Received May 4, 2021; accepted June 15, 2021 Abstract: The intracluster medium (ICM) is expected to experience on average about three passages of weak shocks with low sonic Mach numbers, M . 3, during the formation of galaxy clusters. Both protons and electrons could be accelerated to become high energy cosmic rays (CRs) at such ICM shocks via diffusive shock acceleration (DSA). We examine the effects of DSA by multiple shocks on the spectrum arXiv:2106.08521v1 [astro-ph.HE] 16 Jun 2021 of accelerated CRs by including in situ injection/acceleration at each shock, followed by repeated re- acceleration at successive shocks in the test-particle regime. For simplicity, the accelerated particles are assumed to undergo adiabatic decompression without energy loss and escape from the system, before they encounter subsequent shocks. We show that in general the CR spectrum is flattened by multiple shock passages, compared to a single episode of DSA, and that the acceleration efficiency increases with successive shock passages. However, the decompression due to the expansion of shocks into the cluster outskirts may reduce the amplification and flattening of the CR spectrum by multiple shock passages. The final CR spectrum behind the last shock is determined by the accumulated effects of repeated re- acceleration by all previous shocks, but it is relatively insensitive to the ordering of the shock Mach numbers. Thus multiple passages of shocks may cause the slope of the CR spectrum to deviate from the canonical DSA power-law slope of the current shock. Key words: acceleration of particles — cosmic rays — galaxies: clusters: general — shock waves 1. I NTRODUCTION instance, Melrose & Pope (1993, MP93 hereafter) as- sumed that mono-energetic particles are injected with Cosmic rays (CRs) are known to be produced primar- the spectrum, φ0 (p) = Kδ(p − p0 ), at each shock, and ily via diffusive shock acceleration (DSA) in a variety that the postshock flow is decompressed by a factor of of astrophysical shocks (Bell 1978; Blandford & Eich- 1/r between each shock. They provided an analytic ex- ler 1987). In the test particle regime, where the CR pression for the distribution function f (p) after multiple pressure is dynamically insignificant, DSA predicts the passages of identical shocks. power-law momentum distribution of accelerated par- ticles, f (p) ∝ p−q , where the slope is q = 3r/(r − 1) In this paper, we revisit the problem of DSA by and r is the shock compression ratio (Drury 1983). For multiple shocks to explore the effects of re-acceleration strong, adiabatic, non-relativistic shocks, this would of CRs by weak shocks that are induced in the in- give a power-law index of q = 4, if nonlinear DSA effects tracluster medium (ICM) of galaxy clusters. Cosmo- were to be ignored. However, plasma simulations of logical hydrodynamic simulations demonstrated that collisionless shocks demonstrated that at strong quasi- the ICM could encounter shocks several times on av- parallel shocks order of 10 % of the shock kinetic energy erage during the formation of the large scale struc- could be transferred to CRs (e.g., Caprioli & Spitkovsky tures in the Universe (e.g., Ryu et al. 2003; Vazza et 2014; Park et al. 2015), possibly leading to nonlinear al. 2009, 2011). Such ICM shocks are expected to pro- back-reactions from CRs to the underlying flow. duce CR protons and electrons via DSA (see Brunetti & Several previous studies considered multiple pas- Jones 2014, for a review). In particular, merger-driven sages of shocks as a possible scenario to achieve a CR shocks have been detected as radio relics in the outskirts spectrum flatter than p−4 (e.g., White 1985; Achterberg of galaxy clusters through radio synchrotron radiation 1990; Schneider 1993; Melrose & Pope 1993; Gieseler & from shock-accelerated CR electrons (e.g. van Weeren Jones 2000). These authors showed that for a large et al. 2019). In a recent study, using cosmological hy- number of identical shocks, the resulting spectrum ap- drodynamic simulation, Ha et al. (2020) have estimated proach to f (p) ∝ p−3 , independent of the shock com- that the slope of the CR proton spectrum flattens by pression ratio (or shock Mach number). In a more re- ∼ 0.05 − 0.1 and the total energy of CR protons in- alistic situation, the injection process at each shock, creases by ∼ 40−80% due to the re-acceleration of three adiabatic decompression in the far-downstream region, passages of weak shocks through the ICM throughout and further accelerations due to magnetic turbulence in the structure formation history. the postshock region should be considered in addition The physics of collisionless shocks depends on vari- to the re-acceleration of incident upstream CRs. For ous shock parameters including the sonic Mach number, M , the plasma beta, β ≡ Pg /PB , and the obliquity an- Corresponding author: H. Kang gle, θBn , between the upstream background magnetic 103
104 Kang field direction and the shock normal (e.g., Marcowith 2. DSA S PECTRUM BY M ULTIPLE S HOCKS et al. 2016). CR protons are known to be acceler- The ICM is expected to experience on average about 3 ated efficiently at quasi-parallel (Qk ) shocks with θBn . passages of shocks with a mean separation, L ∼ 1 Mpc, 45◦ (e.g. Caprioli et al. 2015), while CR electrons are during the formation of galaxy clusters (e.g. Ryu et al. accelerated preferentially at quasi-perpendicular (Q⊥ ) 2003). The average Mach number of shocks formed in shocks with θBn & 45◦ (e.g. Guo et al. 2014). the outskirts of typical galaxy cluster is estimated to be Recent studies have suggested that in the weakly in the range hM i ∼ 2−2.5 (e.g. Vazza et al. 2011; Ha et magnetized ICM plasmas (β ∼ 100) the proton accel- al. 2018a). The mean time between consecutive shock eration is effective only at supercritical Qk -shocks with passages with Vs ∼ 3×103 km s−1 can be approximated M & 2.3 due to the efficient reflection of protons at roughly as tpass ∼ L/Vs ∼ 3 × 108 yrs. the shock (Ha et al. 2018b). By contrast, the electron During the time between consecutive shock pas- acceleration is effective only at supercritical Q⊥ -shocks sages, tpass , CR protons can be accelerated via DSA up with M & 2.3 via the excitation of the electron fire- to the maximum momentum, hose instability (Kang et al. 2019) and the Alfvén ion 2 cyclotron instability (Ha et al. 2021). Based on these pmax Vs tpass BICM ≈ 1.5×109 , plasma simulation studies, Ryu et al. (2019) and Kang mp c 3 × 103 km s−1 108 yrs 1 µG (2020) proposed analytic forms for f (p) for CR protons (1) and electrons, respectively, which are valid for weak where the shock compression ratio, r = 3, is adopted, shocks in the test-particle regime. However, it is known and BICM is the magnetic field strength in the ICM in that magnetic fluctuations on the relevant kinetic scales units of microgauss (Kang 2020). Energy loss processes could facilitate particle injection to DSA (e.g. Guo & such as Coulomb collisions, synchrotron emission, and Giacalone 2015). So the effects of pre-existing magnetic the interactions with background radiation are not con- turbulence in the ICM on DSA at subcritical shocks sidered in the estimation of pmax here. Throughout the (M < 2.3) have yet to be understood. paper, common symbols in physics are used: e.g., mp As discussed in Kang (2020), DSA can oper- for the proton mass, c for the speed of light, and kB for ate in the two different modes: (1) in situ injec- the Boltzmann constant. The gyroradius of CR protons tion/acceleration mode, in which particles are injected is given as directly from the background thermal pool at the shock, −1 and (2) re-acceleration mode, in which pre-existing pmax BICM rg (p) ≈ 1.1 kpc , (2) CR particles are accelerated repeatedly at subsequent 109 mp c 1 µG shocks. In the test-particle regime, the particle spec- trum resulted from these two modes of DSA can be so typically rg (p) L a the CR proton energy of E < found either analytically or semi-analytically. 1018 eV. We defer the exploration of the electron accelera- 2.1. Preambles tion by multiple shocks to a future work, since energy We consider a sequence of consecutive shocks that prop- losses of CR electrons due to synchrotron radiation and agate into the upstream gas of the temperature, T1 , and inverse Compton (iC) scattering introduce additional the hydrogen number density, n1 . Hereafter, the sub- physical scales. However, here we briefly describe a dis- scripts, 1 and 2, denote the preshock and postshock crepancy related to observations of radio relics in or- states, respectively. We adopt the following assump- der to allude a possible implication of electron DSA by tions (see Figure 1): multiple shocks. For some ‘radio relic shocks’, the ra- dio Mach number, Mrad = [(αint + 1))/(αint − 1)]1/2 , is 1. There are Ns consecutive shocks that sweep the higher than the X-ray Mach number, MX , inferred from background medium. The k-th shock is specified X-ray observations (e.g., temperature jump), where by a sonic Mach number, Mk , which determines the αint = (q − 2)/2 is the spectral index of the integrated shock compression ratio, rk = n2,k /n1,k , the slope radio spectrum (e.g., Akamatsu & Kawahara 2013; van of DSA power-law spectrum, qk = 3rk /(rk − 1), Weeren et al. 2019). Possible solutions to explain this and the temperature jump, T2,k /T1,k . For simplic- puzzle suggested so far include re-acceleration of pre- ity, we presume that the hydrogen number density existing fossil CR electrons with a flat spectrum (e.g., of the underlying ICM may decrease by a factor Kang 2016) and in situ acceleration by an ensemble of D ≤ 1, that is, n1,(k+1) = Dn1,k , considering of shocks formed in the turbulent ICM (e.g., Hong that the ICM shocks may expand into the strati- et al. 2015; Roh et al. 2019; Rajpurohit et al. 2020; fied outskirts of the host cluster. In order to limit Domı́nguez-Fernández et al. 2021). the number of free parameters, we fix the preshock In the next section we describe the semi-analytic temperature at T1 . approach to follow DSA by multiple shocks along with 2. Adopting the DSA power-law in the test-particle the underlying assumptions and justification. In Sec- regime for weak shocks with Mk . 3, the postshock tion 3, we apply our approach to a few examples, where spectrum of CR protons, injected and accelerated the re-acceleration by several weak shocks of M ≤ 3 in at the k-th shock, is assumed to have the form the ICM environment is considered. A brief summary of finj,k ∝ p−qk for p ≥ pinj,k , where pinj,k is the will be given in Section 4. injection momentum.
DSA by Multiple Weak Shocks 105 2.2. Injected Spectrum at Each Shock Following Ryu et al. (2019), we assume that suprather- mal protons with p & pinj,k could diffuse across the shock transition layer, and participate in the DSA pro- cess at the k-th shock with the sonic Mach number Mk . Here the injection momentum is defined as pinj,k = Q · pth,k , (3) where the postshock thermal momentum, pth,k = (2mp kB T2,k )1/2 , depends on the postshock tempera- ture, T2,k , and the injection parameter is set to be Q ≈ 3.8 (e.g., Caprioli & Spitkovsky 2014; Caprioli et al. 2015; Ha et al. 2018b). At each shock the test-particle spectrum of CR pro- tons results from particle injection to DSA for p > pinj,k with the following power-law form: −qk Figure 1. Basic concept of DSA by multiple shocks adopted p in this study. The upstream CR spectrum, fu,k , is re- finj,k (p) = fo,k · . (4) pinj,k accelerated (red arrow) at the k-th shock, resulting in the downstream spectrum, fd,k . In addition, particles are in- In this study, thermal protons are assumed to follow a jected and accelerated (green arrow) to form finj,k at each shock. Then the downstream spectrum is the sum of the Maxwellian distribution for p ≤ pinj,k , so the normal- two components, fsum,k = fd,k + finj,k . After decom- ization factor is specified at pinj,k as pression (blue arrow), the particle momentum decrease to n2,k −3 p0 = Rk p, and so the decompressed spectrum becomes fo,k = p exp(−Q2 ), (5) 0 fsum,k (p) = fsum,k (p/Rk ), where Rk = (D/rk )1/3 is the de- π 1.5 th,k compression factor. The upstream spectrum at the subse- 0 where n2,k is the postshock proton number density. The quent shock is taken to be fu,(k+1) = fsum,k . far-downstream spectrum of injected/accelerated CRs, decompressed behind the k-th shock, becomes 3. Even subcritical shocks with M . 2.3 could ac- −qk celerate CRs via DSA, presuming that the ICM 0 p/Rk finj,k (p) = fo,k · . (6) contains pre-existing magnetic turbulence on the pinj,k relevant kinetic scales. Scattering of particles off such turbulent waves is a prerequisite for DSA. Hereafter, the primed distribution function, f 0 (p), represents the decompressed spectrum in the far- 4. The upstream spectrum of incident particles, downstream region of each shock. 0 If the acceleration is limited by a finite size or age, fu,k (p) = fsum,k−1 (p), includes the contributions of CRs injected and re-accelerated by all previ- an exponential cutoff at a maximum momentum, pmax , ous shocks. It becomes the downstream spectrum, should be applied to Equations (4) and (6). We do not fd,k (p), after the re-acceleration at the k-th shock. consider such cases, since pmax mp c. 5. In the postshock region, CRs are transported and 2.3. Re-acceleration by Subsequent Shocks decompressed adiabatically without energy losses The upstream spectrum of the k-th shock, fu,k (p), con- (e.g., radiation) and escape from the system, so tains the particles injected and re-accelerated by all pre- the particle momentum, p, decreases to p0 = Rk p, vious shocks, which are convected downstream and de- where Rk = (D/rk )1/3 is the decompression factor compressed. Then, the downstream spectrum, fd,k , re- of the k-th shock (see MP93). Due to this com- accelerated at the k-th shock, can be calculated by the bined decompression, the far-downstream spec- following integration (Drury 1983): trum becomes, f 0 (p) = f (p/Rk ). As illustrated in Z p Figure 1, the total downstream spectrum fsum,k (p) fd,k (p, pinj,k ) = qk · p −qk tqk −1 fu,k (t)dt, (7) 0 is decompressed to become fsum,k (p) behind the k- pinj,k th shock. where the lower bound is set to be pinj,k , above which 6. The time between the consecutive passages of the the particles can participate in DSA. Then the im- k-th and (k + 1)-th shocks is much longer than the mediate postshock spectrum before decompression is DSA time scale, so pmax,k pinj,k . In other words, fsum,k (p) = finj,k (p) + fd,k (p), where finj,k corresponds the timescale of shock passage can be separated to the freshly injected particles at the current shock from the DSA timescale. (Equation (4)).
106 Kang 0 Figure 2. DSA at five identical shocks with M = 3 (left) and M = 20 (right). Panels (a.1) and (b.1) show fsum,k (p) with D = 1 (solid lines) and D = 0.5 (dotted lines). The dashed lines show the Maxwellian distributions for the cases with 0 D = 0.5. Panels (a.2) and (b.2) show the power-law slope, qsum = −∂(ln fsum,k )/∂ ln p. The injection parameter is set to be Qi = 3.8. The black, red, blue, green, and magenta lines are used for k =1, 2, 3, 4, and 5, respectively. The cases with M = 20 are shown for illustrative purposes only, since the test-particle assumption is not valid for such strong shocks. After decompression in the downstream region, the 3.1. DSA by Multiple Identical Shocks re-acceleration spectrum becomes In this section, we consider the special case of multi- Z p/Rk ple identical shocks with the same values of M , q, pinj , 0 fd,k (p, pinj,k ) = qk · (p/Rk )−qk tqk −1 fu,k (t)dt, and R (without the subscript k), to compare our ap- pinj,k proach with the analytic treatment of MP93. Their (8) assumptions differ from ours in the following aspects: which can be compared with Equation (3) of MP93. (1) They adopted an injection spectrum with mono- Hereafter, we refer Equation (8) as the ‘re-acceleration energetic particles, φ0 (p) = Kδ(p − p0 ). In this case, integration’, which in fact includes both the re- the spectrum of CRs injected and accelerated at the acceleration and decompression steps. We take the sum first shock, followed by the decompression downstream, of the injection and re-acceleration spectra, becomes −q 0 0 0 0 K q p fsum,k (p) = finj,k + fd,k = fu,(k+1) (p), (9) f1 (p) = . (10) p0 Rp0 as the upstream spectrum at the subsequent shock. Fig- Note that this is the same as the decompressed injec- ure 1 illustrates this concept. After decompression, the tion spectrum in Equation (6), if we take K = fo,i p0 /q momentum of some low energy CRs shifts to p < pinj,k . and p0 = pinj . (2) They did not adopt an injection mo- They can no longer participate in DSA and join the mentum. We presume that they set the lower bound thermal distribution (e.g. Gieseler & Jones 2000). of the re-acceleration integration as p0 Rk−1 (see their Figure 1). In fact, Gieseler & Jones (2000) introduced 3. R ESULTS the concept of the injection momentum, p0 , and made a distinction between thermal and CR distributions. So We consider the ICM plasma that consists of fully ion- they took p0 as the lower bound of the re-acceleration ized hydrogen atoms and free electrons with T1 = 5.8 × integration (see their Equation (1)), and considered the 107 K (5 keV) and nH = 10−4 cm−3 , so the preshock CR particles, which are decompressed to p < p0 , ‘re- thermal pressure at the k-th shock is P1,k = 2n1,k kB T1 . thermalized’. (3) MP93 assumed the uniform back- Note that the normalization of f (p) presented in the ground medium, so the preshock gas density remains figures below scale with nH . constant with D = 1.
DSA by Multiple Weak Shocks 107 Figure 3. DSA by a sequence of three shocks with different Mach numbers of M = 3, 2.7, and 2.3. Two models, A and B, 0 each with D = 1 (left panels) and D = 0.5 (right panels) are shown. In Panels (a.1)-(d.1), the dot-dashed lines show finj,k 0 and the Maxwellian distributions at the k-th shock, while the solid lines show fsum,k . In Panels (a.2)-(d.2), the power-law 0 slope, qsum = −∂(ln fsum,k )/∂ ln p is shown. The injection parameter is set to be Qi = 3.8. The black, red, and blue lines are used for k =1, 2, and 3, respectively. Moreover, MP93 followed the consecutive re- tegration’ in Equation (8) repeatedly for (n − 1) times. acceleration of injected particles from the first shock to For p pinj , the small integration constant at each ‘re- the Ns -th shock (see their Equation (4)). Then the con- n−1 acceleration integration’, ln(1/Rn−1 ) , can be ig- tributions from all the shocks were added to obtain the nored, so the resulting spectrum can be approximated final total spectrum (see their Equation (5)). By con- with the following analytic form, trast, we add the particles freshly injected/accelerated −q h at each shock to the upstream spectrum of the next q n−1 i p p n−1 0 shock, as illustrated in Figure 1. fn (p) ≈ fo · ln . (n − 1)! Rn pinj Rn−1 pinj Here we follow the MP93’s approach to derive an (13) approximate analytic expression. The spectrum of CRs This equation differs slightly from Equation (4) of injected/accelerated at the first shock, after the down- MP93, which is an exact expression because the in- stream decompression, is given as tegration constant disappears with their lower bound, p −q p0 Rk−1 , of the re-acceleration integration. 0 f1 (p) = fo · . (11) Note that fn0 (p) represents the particles injected at Rpinj the first shock and re-accelerated by (n − 1) times by Applying the ‘re-acceleration integration’ for the first subsequent shocks. Then, the total contribution due to time at the second shock results in all Ns shocks can be expressed as −q p p Ns f20 (p) = fo · q ln . (12) 0 X R2 pinj Rpinj ftot (p, Ns ) ≈ fn0 (p). (14) n=1 Then, the far-downstream spectrum behind the n-th shock can be found by applying the ‘re-acceleration in- In fact, the first term of this summation, f10 (p), rep-
108 Kang Figure 4. Same as Figure 3, except the models C and D with M = 2.5, 2 and 1.7 are shown. resents CRs injected/accelerated at the Ns -th shock, Gieseler & Jones (2000). 0 while the last term, fN s (p), corresponds to CRs in- For the parameters relevant for weak ICM shocks, jected/accelerated at the first shock and re-accelerated namely, M = 3 and Ns = 3, panel (a.2) of Figure 2 repeatedly by (Ns − 1) consecutive identical shocks. shows that the flattening of the spectrum (blue solid 0 0 Note that ftot (p, Ns ) ≈ fsum,N s (p), since the contri- line) by multiple shocks could leads to a substantial bution due to the small integration constants can be deviation from the canonical DSA slope of qDSA = 4.5. ignored. However, the decompression due to adiabatic expansion (D = 0.5, blue dotted line) somewhat weakens such flat- Figure 2 shows the cases for five identical shocks tening effects of multiple shock passages. On the other with M = 3 (left panels) and M = 20 (right panels). hand, if one does not account for the decompression Panels (a.1) and (b.1) show the downstream spectrum, factor of 1/rk , the resulting spectrum would be much 0 fsum,k (p), for the cases with D = 1 (solid lines) and flatter than shown here. 0 D = 0.5 (dotted lines). Note that fsum,k (p) is calcu- The results shown in Figure 2 could be applied lated numerically, based on Equation (9). Although as well for low energy CR electrons that might not 0 not shown, the analytic expression, ftot (p, Ns ), approx- be affected by synchrotron and IC losses. For ex- 0 imates closely the numerical estimation, fsum,k (p), and ample, for radio emitting electrons with the Lorentz the difference between the two is small for p pinj . factor γe ∼ 104 (p/mp c ∼ 5), the power-law slope, However, the test-particle assumption becomes invalid qsum ≈ 4.25, which is slightly flatter than qDSA = 4.5 after several shock passages even with low Mach num- for a single M = 3 shock. bers (see Figure 6 below). So the analytic approxima- tion in Equation (13) should be used only for a small 3.2. DSA by Multiple Shocks with Different Ms number of shocks, e.g., Ns . 3 − 5. The cases with We now explore a more realistic scenario, in which weak M = 20 are shown for comparisons with the previous ICM shocks with different Mach numbers of M . 3. studies, although the test-particle assumption is not Figure 3 shows the two cases, A (M = 3, 2.7, 2.3 in valid for such strong shocks. For example, the solid upper panels) and B (M = 2.3, 2.7, 3 in lower pan- lines in panel (b.2) can be compared with Figure 1 of els) with D = 1 (left panels) and D = 0.5 (right pan-
DSA by Multiple Weak Shocks 109 Figure 5. DSA by a sequence of three shocks with different Mach numbers shown in Figures 3 and 4. The left (right) panels 0 show the cases with D = 1 (D = 0.5). The upper panels show the total spectrum behind the third shock, fsum,3 (p), while 0 the lower panels show qsum = −∂(ln fsum,3 )/∂ ln p. The black and red lines compare A and B, while the blue and green lines compare C and D. els). Figure 4 compares the two cases, C (M = 2.5, 1983). The momentum gain of a particle through re- 2, 1.7 in upper panels) and D (M = 1.7, 2, 2.5 in peated shock crossings in a sequence of multiple shocks lower panels). In panels (a.1)-(d.1), the solid lines show is cumulative, so it should not depend on the ordering 0 the total spectrum, fsum,k in the far-downstream re- of the shock Mach numbers. On the other hand, the gion of the k-th shock (where the index k runs from 1 amount of injected particles at each shock depends on to 3), while the dashed lines show the injection spec- Mk in our injection scheme, since pinj,k and fo,k depend 0 trum, finj,k , injected/accelerated at the k-th shock and on the postshock temperature T2,k and the density of decompressed far-downstream. Note that for the first the incoming proton density, n1,(k+1) = Dn1,k , may de- 0 0 shock, fsum,1 = finj,1 , so the black dashed line overlap crease at subsequent shocks. For example, case A with with the black solid line. M1 = 3 injects more particles at the first shock than The comparison of A and B (or C and D) illustrates case B with M1 = 2.3, and hence the resulting spec- that the final spectrum (blue solid lines) is determined trum has a slightly higher amplitude. As illustrated in by the accumulated effects of repeated re-acceleration Figure 5, the final CR spectrum produced from DSA by by multiple shocks, but almost independent of the or- multiple shocks is almost independent of the ordering dering of Mach numbers. The black and red lines in of M ’s, other than small differences in the amplitude Figure 5 compare the final spectrum at the third shock, due to different injection rates at each shock. 0 fsum,3 for A and B, while the blue and green lines com- Figure 5 also shows that, in the cases with D = 0.5 0 0 pare C and D. The amplitudes of fsum,3 differ slightly with additional decompression, the amplitudes of fsum,3 due to the small difference in pinj,k and fo,k at each are smaller and the slopes, qsum , are flattened slightly shock. Otherwise, the slopes, qsum , are almost the same less than the cases with D = 1. Thus, the additional for p/mp c > 1 in the two cases, A and B (or C and D). decompression parameterized with D weakens the flat- In the DSA theory in the test-particle regime, the tening effects of multiple shocks. mean momentum gain of a particle at each shock cross- ing is h∆pi = (2/3)p(u1 − u2 )/v, where u1 and u2 are the preshock and postshock flow speeds, respectively, in 3.3. DSA Efficiency the shock rest frame, and v is the particle speed. The probability of the particle returning from downstream In this section, we examine how multiple passages of is Pret = (1−4u2 /v). These two factors, h∆pi, and Pret , shocks affect the DSA efficiency. First, we have calcu- determine the DSA power-spectrum, f (p) ∝ p−q (Drury lated the CR energy density in the immediate postshock
110 Kang Figure 6. CR pressure, ECR,k /ρ0 u20 , the DSA efficiency, η, and the ratio, ECR,k /Eth,k , due to a sequence of three shocks, where ECR,k and Eth,k are the CR and thermal energy densities in the immediate postshock region at the k-th shock. The constant normalization factor, ρ0 u20 , corresponds to the shock kinetic energy density for the preshock gas density, ρ0 , and the shock speed, u0 , with M = 3. Five cases, A-E, with different combinations of M ’s are shown. The upper (lower) panels show the cases with D = 1 (D = 0.5). region behind the k-th shock, corresponds to the shock kinetic energy density for Z q the preshock gas, ρ0 , and the shock speed, u0 , with ECR,k = 4πc ( p2 + (mp c)2 −mp c)fsum,k (p)p2 dp, M = 3. The three quantities increase with the suc- pinj,k cessive passages of shocks, as expected. The compar- (15) ison of A (black) and B (red) or the comparison of C where fsum,k (p) = finj,k +fd,k is the postshock spectrum (blue) and D (green) demonstrate that the final effi- before decompression. The DSA efficiency is commonly ciencies at the third shock do depend slightly on the defined in terms of the shock kinetic energy flux (Ryu ordering of the three values of Mk ’s. This may seem et al. 2003), inconsistent with Figure 5. However, we note that ECR ECR,k u2,k is estimated from the immediate postshock spectrum, η≡ 3 , (16) fsum,k , before decompression, while Figure 5 shows the 0.5ρ1,k Vs,k 0 far-downstream spectrum, fsum,3 decompressed in the where u2,k is the postshock flow speed and the denom- downstream region. In addition, both the denomina- inator is the incident kinetic energy flux of the k-th 3 tor in η, 0.5ρ1,k Vs,k , and the denominator in the ratio, shock. Another measure for the DSA efficiency can be Eth,k , do depend on Mk of the k-th shock. So, in the given in terms of the ratio, ECR,k /Eth,k , where Eth,k is cases with higher M3 (B and D), η and ECR,3 /Eth,3 the postshock thermal energy density behind the k-th are smaller at the third shock, compared with the cases shock. with lower M3 (A and C). Figure 6 shows ECR,k /ρ0 u20 (left column), η (mid- dle column) and the ratio of ECR,k /Eth,k (right column) Obviously, case E (magenta) with three identical at each shock for the five cases, A-E, with D = 1 (up- shocks with M = 3 and D = 1 produces the highest per panels) and D = 0.5 (lower panels). Note that amount of CRs. So η increases as 7.1 × 10−3 , 3.7 × the normalization factor, ρ0 u20 , is a constant, which 10−2 , and 0.13 for k = 1, 2, and 3, respectively, while
DSA by Multiple Weak Shocks 111 ECR,k /Eth,k increases as 9.6 × 10−3 , 5.0 × 10−2 , and CR electron spectrum injected and re-accelerated at 0.17. The other four cases with different M ’s, A-D, multiple shocks, if the energy losses due to synchrotron η . 0.1 after three shock passages, so the test-particle emission and iC scattering are taken into account. For assumption remains valid. example, if the timescale of shock passage is compa- The cases with D = 0.5 with additional decom- rable to or greater than the synchroton cooling time, pression (lower panels) produce smaller ECR than the tsync ≈ 108 yrs(Beff /5 µG)−2 (γe /104 )−1 , the flatten- cases with D = 1 (upper panels). Note that the ratio, ing and amplifying effects of multiple shock passages ECR,k /ρ0 u20 , could decrease at subsequent shocks be- would be reduced substantially. The effective magnetic cause of the successively lower preshock density, n1,k . field strength, Beff , takes into account both synchrotron and iC cooling, and γe is the Lorentz factor of radio- 4. S UMMARY emitting electrons. We leave such calculations including We have revisited the problem of DSA by multi- treatments of the energy losses to a upcoming paper. ple shocks, considering weak ICM shocks in the test- We recognize the possibility that the re- particle regime. Suprathermal particles with p ≥ pinj ∼ acceleration by multiple shocks may explain the 3.8pth are assumed to be injected to the DSA pro- discrepancies, Mx . Mradio , found in some radio relics cess at each shock, while incident upstream CRs are (Akamatsu & Kawahara 2013; Hong et al. 2015; van re-accelerated by subsequent shocks (Ryu et al. 2019; Weeren et al. 2019), as discussed in the Introduction. Kang 2020). Based on Liouville’s theorem, the accel- In other words, Mx , inferred from X-ray observations erated CRs are assumed to undergo adiabatic decom- (e.g., temperature jump) may represent the Mach pression by a factor of R = (D/r)1/3 , which takes into number of the most recent shock, while Mradio , inferred account the expansion of the background medium and from radio observations (e.g., ratio spectral index), the decompression of the postshock flow behind each may reflect the accumulated effects of all past shocks. shock (see MP93). The decompression causes the par- ticle momentum, p, to decrease to p0 = Rp. ACKNOWLEDGMENTS We have considered the several examples with three shocks with the sonic Mach numbers, M = 1.7 − 3. The The author thanks the anonymous referee for construc- main findings can be summarized as follows: tive comments. This work was supported by a 2-Year Research Grant of Pusan National University. 1. Multiple passages of identical shocks with the same Mach number enhance the amplitude of the mo- 0 R EFERENCES mentum distribution, fsum,k (p), and flatten the 0 slope, qsum = −∂(ln fsum,k )/∂ ln p (see Figure 2). Akamatsu, H., & Kawahara, H. 2013, Systematic X-Ray This is in a good agreement with the previous stud- Analysis of Radio Relic Clusters with Suzaku, PASJ, 65, ies (e.g., Schneider 1993; Melrose & Pope 1993; 16 Gieseler & Jones 2000). Moreover, the DSA effi- Achterberg, A. 1990, Particle Acceleration by an Ensemble ciency, η and the ratio ECR,k /Eth,k increase during of Shocks, A&A, 231, 251 each shock passage (see Figure 6). Bell, A. R. 1978, The Acceleration of Cosmic Rays in Shock Fronts. I, MNRAS, 182, 147 2. For the cases with multiple shocks with different Blandford, R. D., & Eichler, D. 1987, Particle Acceleration Mach numbers, the differences in the injection mo- at Astrophysical Shocks - a Theory of Cosmic-Ray Origin, mentum, pinj,k , and the ensuing normalization fac- Phys. Rept., 154, 1 tor, fo,k , at each shock lead to small differences Brunetti, G., & Jones, T. W. 2014, Cosmic Rays in Galaxy 0 in the final CR spectrum, fsum,3 (p), depending on Clusters and Their Nonthermal Emission, Int. J. of Mod- the ordering of M ’s (see Figures 3 and 4). The fi- ern Physics D. 23, 30007 nal spectrum of CRs at the last shock turns out to Caprioli, D., & Sptikovsky, A. 2014, Simulations of Ion Ac- be almost independent of the ordering of M ’s (see celeration at Non-relativistic Shocks. I. Acceleration Effi- Figure 5). Likewise, the power-law slope, qsum (p), ciency, ApJ, 783, 91 remains almost unchanged as long as p/mp c 1, Caprioli, D., Pop, A. R., & Sptikovsky, A. 2015, Simulations regardless of the order of M ’s. and Theory of Ion Injection at Non-relativistic Collision- less Shocks, ApJ, 798, 28 3. The power-law slope of final CR spectrum is gov- Domı́nguez-Fernández, P., Brüggen, M., Vazza, F. et al. erned by the accumulated effects of repeated re- 2021, Morphology of Radio Relics – I.What Causes the acceleration by past shocks, meaning it can be dif- Substructure of Synchrotron Emission?, MNRAS, 500, ferent from the DSA power-law slope, qDSA , of the 795 current shock. Drury, L. O’C. 1983, An Introduction to the Theory of Diffu- 4. The decompression in the downstream region of sive Shock Acceleration of Energetic Particles in Tenuous each shock due to the expansion of the background Plasmas, Rep. Prog. Phys., 46, 973 medium weakens the amplification and flattening Gieseler, U. D. J. & Jones, T. W. 2002, First order Fermi acceleration at multiple oblique shocks, A&A, 357, 1133 of the resulting CR spectrum. Guo, F., & Giacalone, J. 2015, The Acceleration of Elec- Although we have considered DSA of CR protons trons at Collisionless Shocks Moving Through a Turbu- in this study, the same approach can be applied to the lent Magnetic Field, ApJ, 802, 97
112 Kang Guo, X., Sironi, L., & Narayan, R. 2014, Non-thermal Park, J., Caprioli, D., & Spitkovsky, A. 2015, Simultaneous Electron Acceleration in Low Mach Number Collision- Acceleration of Protons and Electrons at Nonrelativistic less Shocks. I. Particle Energy Spectra and Acceleration Quasiparallel Collisionless Shocks, PRL, 114, 085003 Mechanism, ApJ, 793, 153 Rajpurohit, K., Hoeft, M., Vazza, F. et al., 2020, New Mys- Ha, J.-H., Ryu, D., & Kang, H. 2018, Properties of Merger teries and Challenges from the Toothbrush Relic: Wide- Shocks in Merging Galaxy Clusters, ApJ, 857, 26 band Observations from 550MHz to 8GHz, A&A, 636, Ha, J.-H., Ryu, D., Kang, H., & van Marle, A. J. 2018, A30 Proton Acceleration in Weak Quasi-parallel Intracluster Shocks: Injection and Early Acceleration, ApJ, 864, 105 Roh, S., Ryu, D., Kang, H., Ha, S., & Jang, H. 2019, Turbu- Ha, J.-H., Ryu, D., & Kang, H. 2020, Gamma-Ray and Neu- lence Dynamo in the Stratified Medium of Galaxy Clus- trino Emissions due to Cosmic-Ray Protons Accelerated ters, ApJ, 883, 138 at Intracluster Shocks in Galaxy Clusters, ApJ, 892, 86 Ryu, D., Kang, H., Hallman, E., & Jones, T. W. 2003, Cos- Ha, J.-H., Kim, S., Ryu, D., & Kang, H. 2021, Effects of mological Shock Waves and Their Role in the Large-Scale Multi-scale Plasma Waves on Electron Preacceleration at Structure of the Universe, ApJ, 593, 599 Weak Quasi-perpendicular Intracluster Shocks, arXiv e- Ryu, D., Kang, H., & Ha, J.-H. 2019, A Diffusive Shock prints, arXiv:2102.03042. Acceleration Model for Protons in Weak Quasi-parallel Hong, S., Kang, H., & Ryu, D. 2015, Radio and X-ray Intracluster Shocks, ApJ, 883, 60 Shocks in Clusters of Galaxies, ApJ, 812, 49 Kang, H. 2016, Reacceleration Model for the ‘Toothbrush’ Schneider, P. 1993, Diffusive particle acceleration by an en- Radio Relic, JKAS, 49, 83 semble of shock waves, A&A, 278, 315 Kang, H. 2020, Semi-analytic Models for Electron Acceler- van Weeren, R. J., de Gasperin, F., Akamatsu, H., et ation in Weak ICM Shocks, JKAS, 53, 59 al. 2019, Diffuse Radio Emission from Galaxy Clusters, Kang, H., & Ryu, D., 2010, Diffusive Shock Acceleration in Space Science Reviews, 215, 16 Test-particle Regime, ApJ, 721, 886 Vazza, F., Brunetti, G., & Gheller, C. 2009, Shock Waves in Kang, H., Ryu, D., & Ha, J.-H. 2019, Electron Pre- Eulerian Cosmological Simulations: Main Properties and acceleration in Weak Quasi-perpendicular Shocks in High- Acceleration of Cosmic Rays, MNRAS, 395, 1333 beta Intracluster Medium, ApJ, 876, 79 Marcowith, A., Bret, A., Bykov, A., et al. 2016, The Micro- Vazza, F., Dolag, K., Ryu, D., et al. 2011, A comparison of physics of Collisionless Shock Waves, RPPh, 79, 046901 cosmological codes: properties of thermal gas and shock Melrose D.B., & Pope M.H., 1993, Diffusive Shock Accel- waves in large-scale structures, MNRAS, 418, 960 eration by Multiple Shocks, Proc. Astron. Soc. Aust. 10, White, R. L, 1985, Synchrotron Emission from Chaotic Stel- 222 (MP93) lar Winds, ApJ, 289, 698
You can also read