Non-uniform particle injection into black hole jets by radiative magnetic reconnection Rin Oikawa, Kenji Toma, Shigeo S. Kimura Abstract Active galactic nuclei often exhibit highly collimated relativistic plasma outflows launched from the vicinity of their central black holes. One of the key theoretical challenges in understanding black hole jet formation is the origin of the plasma that feeds the jet, which remains poorly understood, particularly in explaining the observed jet emission. In this study, we focus on electron–positron pair production generated by high-energy photons from non-axisymmetric magnetic reconnection near the black hole, as suggested by recent three-dimensional general-relativistic magnetohydrodynamics simulations. By employing general relativistic ray tracing, we calculate the spatial distribution of the pair production rate in the jet, taking into account photon propagation and collision angles in curved spacetime. We find that our scenario can naturally supply a sufficient amount of plasma to explain the observed radio emission from the M87 jet, even when photon anisotropy is considered. Furthermore, we show that a spinning black hole plays a crucial role in shaping the spartial dsitribution of the pairs, which in turn affects jet acceleration and very-high-energy emission from the jet base. show]oikawa.rin@astr.tohoku.ac.jp I Introduction Black holes in accreting systems are often observed as some of the most luminous objects in the Universe. In some cases, highly collimated relativistic jets are launched from the vicinity of the black hole ( V. S. Beskin 2010 ; R. Blandford et al. 2019 ) . A prominent example is the supermassive black hole at the center of M87, which has a relativistic jet extending over kiloparsec scales ( K. Hada et al. 2024 ) . The Blandford–Znajek (BZ) process, which extracts rotational energy from a spinning black hole, is widely regarded as one of the leading mechanisms for powering relativistic jets ( R. D. Blandford & R. L. Znajek 1977 ; S. S. Komissarov 2004 ; K. Toma et al. 2025 ) . However, the origin of the jet plasma remains poorly understood and constitutes a key unresolved problem in black hole astrophysics. The composition of black hole jets plays a crucial role in determining their radiative properties (see M. Cerruti 2020 for a review). In particular, the lepton number density directly affects the emission strength of synchrotron radiation and inverse Compton scattering, which are responsible for the observed radio emission from radio-loud AGN jets and gamma-ray emission from blazars ( M. Kino et al. 2015 ; G. Ghisellini & F. Tavecchio 2015 ) . The plasma content also has important dynamical implications: The amount of plasma loaded into the jet determines its inertia, thereby setting an upper limit on the achievable bulk Lorentz factor ( K. Toma & F. Takahara 2012 ; A. Chow et al. 2026 ) . Since BZ jets are constructed by large-scale ordered magnetic fields threading the horizon, diffusive transport of charged particles from the disk and wind into the magnetized jet region is strongly suppressed ( J. M. Mehlhaff et al. 2026 ; A. Chow et al. 2026 ) . It has therefore been widely argued that photon–photon pair production ( \gamma\gamma\rightarrow e^{+}e^{-} ) can inject plasma into the magnetized jet region ( A. Levinson & F. Rieger 2011 ; M. Mościbrodzka et al. 2011 ; S. S. Kimura & K. Toma 2020 ; S. Kisaka et al. 2020 ; G. N. Wong et al. 2021 ; K. Kin et al. 2024 ) . In particular, pair production associated with magnetic reconnection in the vicinity of the black hole is a promising source of high-energy photons that may account for the observed jet brightness. 1 1 1 The pair production by high-energy photons from the accretion disk can supply at most \sim 100 times the Goldreich–Julian density ( S. S. Kimura & K. Toma 2020 ) , and spark gaps keep only the Goldreich–Julian density ( S. Kisaka et al. 2020 , e.g.) (see also Section V.2 ). Recent three-dimensional general-relativistic magnetohydrodynamic (GRMHD) simulations, together with two-dimensional general-relativistic particle-in-cell (GRPIC) simulations, suggest that an equatorial current sheet naturally forms and magnetic reconnection is triggered near the black hole in the magnetically arrested disk (MAD) state ( B. Ripperda et al. 2022 ; E. Figueiredo et al. 2026 ; J. M. Mehlhaff et al. 2026 ) . E. Figueiredo et al. 2026 argued that, in GRPIC simulations of the MAD state, the accretion flow undergoes intermittent magnetic reconnection near the horizon, which regulates the net accumulation of magnetic flux. When reconnection-driven flux diffusion dominates over flux accumulation, large-scale flux eruptions occur, similar to those reported in GRMHD simulations by B. Ripperda et al. 2022 . Such reconnection events efficiently accelerate particles and produce high-energy photons via synchrotron radiation. These photons can create e^{\pm} pairs, providing an efficient plasma supply to the BZ jet with densities reaching \sim 10^{8} – 10^{9} times the Goldreich–Julian density ( B. Ripperda et al. 2022 ; S. S. Kimura et al. 2022 ; H. Hakobyan et al. 2023 ; A. Y. Chen et al. 2023 ; R. Kuze et al. 2024 ) . However, previous studies of pair injection via magnetic reconnection have not fully incorporated general relativistic effects. In particular, photon trajectories have often been treated without accounting for null geodesics in curved spacetime, and general relativistic Doppler shifts have not been consistently included. Moreover, the effect of photon collision angles has not been examined in detail. General relativistic light bending may increase the frequency of head-on photon collisions, which could substantially enhance the pair-production rate and modify the spatial distribution of plasma injected into the BZ jet region. In this paper, we construct a model for plasma injection into the BZ jet driven by radiative magnetic reconnection, explicitly incorporating general relativistic effects. We perform general relativistic ray-tracing calculations to follow photon trajectories and collision angles in curved spacetime. Based on this approach, we derive the pair-production rate and its spatial distribution in the jet. We also take into account the anisotropy of synchrotron photons from reconnection as suggested by recent local PIC simulations ( A. Chernoglazov et al. 2023 ) . In deriving the pair-production rate, we further consider the physical conditions required to sustain the electric current that drives magnetic reconnection. This paper is organized as follows. In Section II , we describe the radiative reconnection model including anisotropy, and the formulation of the pair production rate in curved spacetime. In Section III , we present the numerical method used to compute the spatial distribution of the injected plasma based on general relativistic ray tracing. Section IV contains our main findings: Section IV.1 shows the results applied to M87 and discusses whether the injected plasma is sufficient to account for the observed jet emission; Section IV.2 investigates the influence of black hole spacetime on the spatial distribution; and Section IV.3 examines the spatial distribution and total plasma supply when photon anisotropy is taken into account. In Section V , we discuss the implications of the injected plasma distribution for jet acceleration and very-high-energy emission from the jet base. Throughout this paper, we use the notation of Q_{x}=Q/10^{x} in cgs units, except for the black hole mass, M_{x}=M/(10^{x}M_{\odot}) . II Model Figure 1 : Schematic picture of particle injection into the black hole magnetosphere driven by magnetic reconnection. Magnetic reconnection releases magnetic energy and converts it into the energy of nonthermal particles. These nonthermal particles subsequently cool via synchrotron radiation, efficiently producing high-energy photons. The resulting photon field leads to electron–positron pair production. The gray-shaded region represents the transition layer between the magnetosphere and the matter-dominated region (see Section II.5 for details). Our matter-loading scenario for the BZ jet is shown in Figure 1 . As introduced in Section I , recent 2D global GRPIC simulations by E. Figueiredo et al. 2026 have revealed a mini-flare phase, referred to as the reconnection-regulated phase, in which the net magnetic flux near the black hole increases over time while undergoing intermittent magnetic reconnection in the MAD state. Similar mini-flares have also been reported in 3D high-resolution GRMHD simulations ( B. Ripperda et al. 2022 ) . These mini-flare releases a large amount of magnetic energy (Section II.1 ). The reconnection-driven electric field efficiently accelerates particles, converting the released magnetic energy into nonthermal particles ( S. Zenitani & M. Hoshino 2001 ; M. Hoshino 2023 ; M. Hoshino 2024 ; L. Sironi et al. 2025 ) . These particles emit high-energy photons via synchrotron cooling (Section II.3 ), which then propagate along geodesics in the curved spacetime and undergo pair production ( \gamma\gamma\rightarrow e^{+}e^{-} ) in the black hole magnetosphere. This process steadily loads a substantial amount of plasma into the jet (Section II.4 ). We assume that the reconnection current sheet has a rectangular shape with a characteristic length scale l_{\rm rec}=f_{l}r_{g} , where r_{g}=GM/c^{2} is the gravitational radius. Here, G is the gravitational constant, M is the black hole mass, and c is the speed of light. We adopt f_{l}\approx 2 , corresponding to the typical size of the reconnection current sheet in a mini-flare ( B. Ripperda et al. 2022 ) . Hereafter, we refer to the current sheet as the ‘reconnection region’. In this section, we introduce models for the background electromagnetic field and the photon field produced by radiative magnetic reconnection, and formulate the pair-production rate in general relativity. As the background spacetime, we adopt the Kerr metric, characterized by the black hole mass M and spin parameter |a|<1 (explicit expressions for the metric components are given in Appendix A ). II.1 Background electromagnetic field model Magnetic reconnection in the vicinity of a black hole releases a large amount of magnetic energy, {L}_{\rm rec}\approx{\beta}_{\rm rec}c\frac{{B}^{2}l_{\rm rec}^{2}}{4\pi}\sim 5\times 10^{42}\,f_{l}^{2}{{\beta}}_{\rm rec,-1}{B}_{3}^{2}M_{9}^{2}\;{\rm erg\,s^{-1}}, (1) where {\beta}_{\rm rec}\approx 0.1 is the reconnection speed in the kinetic regime ( S. Zenitani & M. Hoshino 2001 ; Y. E. Lyubarsky 2005 ; G. R. Werner et al. 2018 ; F. Guo et al. 2020 , e.g.,) , and {B} is the upstream magnetic field strength. A significant fraction of this energy is converted into nonthermal particles (see Section II.3 ), depending on the background magnetic field. In this study, we adopt the analytic model proposed by S. S. Kimura et al. 2022 for the background magnetic field in the black hole magnetosphere, in which the poloidal ( B_{r} ) and toroidal ( B_{\varphi} ) components are roughly B_{r}\sim B_{\varphi}\propto r^{-2} in the reconnection region considered in this paper (see Appendix B ). II.2 Local frame of the reconnection layer The energy released by magnetic reconnection is converted into nonthermal particles, which subsequently undergo synchrotron cooling and are transformed into radiation energy. To describe the physical quantities associated with these microscopic processes, we introduce a local (i.e., locally Minkowskian) frame in the reconnection region. A spinning black hole induces an electric field (see Appendix B and Eq. ( B1 )). However, such a background electric field is typically not included in local PIC simulations that investigate magnetic reconnection. To construct a microscopic model based on PIC simulations, we should consider the local frame in which the background electric field vanishes. The electric field vanishes locally in the frame moving with the drift velocity given by ( S. S. Komissarov 2004 ; K. Toma & F. Takahara 2014 ) \hat{\bm{v}}_{d}=\frac{\hat{\bm{D}}\times\hat{\bm{B}}}{\hat{B}^{2}}, (2) (see Appendix C ), where \bm{D} and \bm{B} denote the electric and magnetic fields measured by the zero-angular-momentum-observer (ZAMO) (see Appendix A ), and \hat{\bm{v}}_{d} is expressed in the ZAMO orthonormal basis. In this study, we adopt the drift frame as the reference frame in which particle acceleration and radiation processes associated with magnetic reconnection are described. Throughout this paper, quantities expressed in the orthonormal basis of the ZAMO frame are denoted by a hat (e.g., A^{\hat{\mu}} or \hat{A} ), while those expressed in the orthonormal basis of the drift frame are denoted by a tilde (e.g., A^{\tilde{\mu}} or \tilde{A} ). II.3 Radiative reconnection model In considering radiative magnetic reconnection, the ratio of the following two physical quantities—the magnetization parameter \sigma and the synchrotron burnoff limit \gamma_{\rm syn} are important for determining the strength of radiative cooling ( D. A. Uzdensky 2011 ; A. Chernoglazov et al. 2023 ; L. Sironi et al. 2025 ) . The two quantities are defined as follows: \sigma=\frac{\tilde{B}^{2}}{4\pi\tilde{\rho}c^{2}},\quad\gamma_{\rm syn}=\sqrt{\frac{6\pi e\tilde{\beta}_{\rm rec}}{\sigma_{T}\tilde{B}}}, (3) where \tilde{\rho} is the mass density of the upstream plasma and \sigma_{T} is the Thomson cross section. The Lorentz factor \gamma_{\rm syn} corresponds to the maximum particle energy for which the acceleration timescale due to the reconnection electric field \tilde{t}_{\rm acc}={\tilde{\gamma}mc}/({e\tilde{\beta}_{\rm rec}\tilde{B}}) balances the synchrotron cooling timescale \tilde{t}_{\rm syn}={\tilde{\gamma}mc^{2}}/{P_{\rm syn}}={6\pi mc}/({\sigma_{T}\tilde{B}^{2}\tilde{\gamma}}) . Here, m is the rest mass of the charged particle and P_{\rm syn} denotes the synchrotron power emitted by a single particle. When \gamma_{\rm syn}>\sigma , radiative cooling can be regarded as negligible, and particles accelerated in the reconnection region can potentially reach Lorentz factors of \tilde{\gamma}\sim\sigma . We refer to this regime as the ‘weak’ cooling regime. In contrast, when synchrotron cooling becomes sufficiently strong ( \gamma_{\rm syn}<\sigma ), most particles can be accelerated only up to Lorentz factors of order \tilde{\gamma}\sim\gamma_{\rm syn} . We refer to this regime as the ‘strong’ cooling regime. Recent local 3D PIC simulations by A. Chernoglazov et al. 2023 have demonstrated that the energy distribution of accelerated particles is significantly modified by secondary acceleration processes. In particular, a fraction of particles can be accelerated up to \gamma_{\rm syn} even in the weak cooling regime (and up to \sigma in the strong cooling regime), resulting in a broken power-law distribution. In addition, A. Chernoglazov et al. 2023 reported the anisotropy of synchrotron radiation produced in magnetic reconnection. Their simulations show that the emission is beamed along the direction of the large-scale reconnection-driven electric field in the weak cooling regime. In contrast the emission is beamed along the direction of the upstream magnetic field in the strong cooling regime. Based on the simulations of A. Chernoglazov et al. 2023 , we model the particle energy distribution in radiative magnetic reconnection as follows: \frac{dN_{e}}{d\tilde{\gamma}}\propto\begin{cases}\tilde{\gamma}^{-2}&(\tilde{\gamma}<\tilde{\gamma_{b}}),\\[6.0pt] \tilde{\gamma}^{-p_{\mathrm{tail}}}&(\tilde{\gamma_{b}}<\tilde{\gamma}<\tilde{\gamma}_{\mathrm{cut}}),\end{cases} (4) where \tilde{\gamma}_{b}=\min(\sigma,\gamma_{\rm syn}) and \tilde{\gamma}_{\rm cut}=\max(\sigma,\gamma_{\rm syn}) . Here p_{\rm tail} is the power-law index of the high-energy tail, and we adopt p_{\rm tail}=2.7 , as suggested by A. Chernoglazov et al. 2023 . Note that, in the weak-cooling regime, the power-law index adopted in Eq. ( 4 ) is taken to be smaller by 1 than that obtained by A. Chernoglazov et al. 2023 . This is because their result describes the particle spectrum in a local region, whereas over a dynamical timescale particles experience cooling, which effectively shifts the spectrum by -1 in the power-law index. In contrast, the strong-cooling spectrum in A. Chernoglazov et al. 2023 already incorporates the effect of cooling and can be used without modification. We model the anisotropy of the photon distribution using a functional form (cf. A. M. Beloborodov et al. 2011 : F({\theta_{s}})\propto\left(\xi^{2}+\sin^{2}{\theta_{s}}\right)^{-3}, (5) where \theta_{s} is the angle between the photon momentum vector ( \tilde{\bm{k}} ) and the electric field in the drift frame in the weak-cooling regime (or the magnetic field in the strong-cooling regime). The parameter \xi represents the degree of anisotropy (see Figure 2 for the functional form). Figure 2 : Plot of Equation ( 5 ) for the anisotropy parameter \xi\in(0.1,\,0.25,\,1.0,\,1000) . Smaller values of \xi correspond to stronger beaming, while larger values indicate that the radiation becomes more isotropic (i.e., F(\theta_{s})\approx 1 ). Non-thermal particles rapidly experience synchrotron cooling, efficiently producing a photon field: G(\tilde{E_{\gamma}})=\tilde{E_{\gamma}}\frac{dN_{\gamma}}{d\tilde{E_{\gamma}}}\propto\begin{cases}\tilde{E}_{\gamma}^{-\frac{1}{2}}&(\tilde{E}_{\gamma}<\tilde{E}_{\gamma,b}),\\[6.0pt] \tilde{E}_{\gamma}^{\frac{1-p_{\mathrm{tail}}}{2}}&(\tilde{E}_{\gamma,b}<\tilde{E}_{\gamma}<\tilde{E}_{\gamma,\rm cut}),\end{cases} (6) where \tilde{E}_{\gamma}={he\tilde{B}\tilde{\gamma}^{2}}/({2\pi m_{e}c}) is the energy of synchrotron photons, and \tilde{E}_{\gamma,b} and \tilde{E}_{\gamma,\rm cut} correspond to emission from electrons with Lorentz factors \tilde{\gamma}_{b} and \tilde{\gamma}_{\rm cut} . We can write the intensity as \tilde{I}_{\tilde{E}_{\gamma}}=\tilde{E}_{\gamma}\frac{dN_{\gamma}}{d\tilde{t}d\tilde{\Omega}d\tilde{E}_{\gamma}d\tilde{A}}=C_{0}F(\theta_{s})G(\tilde{E}_{\gamma}), (7) where d\tilde{A} is the surface element, d\tilde{\Omega}=\sin{\tilde{\theta}}\,d\tilde{\theta}\,d\tilde{\varphi} is the solid angle, and \theta_{s} is related to \tilde{\theta} and \tilde{\varphi} through \cos\theta_{s}=\sin{\tilde{\theta}}\cos{\tilde{\varphi}} . Here C_{0} is a normalization factor, and we calculate it based on energy conservation as \tilde{L}_{\rm rec}=2l_{\rm rec}^{2}\int_{2\pi}d\tilde{\Omega}\int d\tilde{E}_{\gamma}\;\tilde{I}_{\tilde{E}_{\gamma}}\cos{\tilde{\theta}}, (8) where the factor of 2 accounts for the contributions from both sides of the reconnection current sheet. We then define the photon distribution function, i.e., the coordinate-invariant six-dimensional phase-space density in the reconnection region, as \mathcal{F}(x_{E},k_{E})=\frac{c^{2}}{\tilde{E}_{\gamma}^{3}}\tilde{I}_{\tilde{E}_{\gamma}}, (9) where x_{E} and k_{E} denote the spacetime position and four-momentum of the photon at the emission point, respectively. II.4 e^{\pm} pair production in the BZ jet Photons emitted from the reconnection region propagate through curved spacetime. These trajectories are described by the Hamiltonian formulation of the geodesic ( S. V. Fuerst & K. Wu 2004 ; H.-Y. Pu et al. 2016 ) . The Hamiltonian is given by \mathscr{H}=\frac{1}{2}g^{\mu\nu}k_{\mu}k_{\nu}, (10) where g_{\mu\nu} is the metric tensor (see Appendix A for the explicit form of the Kerr metric), and k^{\mu} is the four-momentum of the photon. The equations of motion which we solve are obtained from Hamilton’s equations, \frac{dx^{\mu}(\lambda)}{d\lambda}=\frac{\partial\mathscr{H}}{\partial k_{\mu}},\>\>\frac{dk_{\mu}(\lambda)}{d\lambda}=-\frac{\partial\mathscr{H}}{\partial x^{\mu}}, (11) with g_{\mu\nu}k^{\mu}k^{\nu}=0 . Here \lambda is an affine parameter along the photon geodesic, and x^{\mu}(\lambda) represents the geodesic of the photon. To evaluate the local e^{\pm} pair production rate, we need the photon distribution function \mathcal{F}(x(\lambda),k(\lambda)) at the local point where pair annihilation occurs. We assume that photons emitted from the reconnection region propagate through the magnetosphere along geodesics without undergoing interactions (e.g., with electrons/positrons or other MeV photons) before reaching the annihilation point. This assumption is justified because the optical depths for interactions between photons produced in the reconnection region and the particles and photons in the black hole magnetosphere are much smaller than unity. Their propagation is described by the collisionless Boltzmann equation ( R. Birkl et al. 2007 ; I. Zalamea & A. M. Beloborodov 2011 ) : \dfrac{d\mathcal{F}(x(\lambda),k(\lambda))}{d\lambda}=0. (12) Eq. ( 12 ) shows that \mathcal{F}(x(\lambda),k(\lambda)) is conserved along geodesics. Therefore, the photon distribution function at the local point is equal to that given by Eq. ( 9 ), i.e., \mathcal{F}(x(\lambda),k(\lambda))=\mathcal{F}(x_{E},k_{E}). (13) The local production rate of electron–positron pairs per unit time and unit volume due to photon–photon annihilation is given by ( M. Mościbrodzka et al. 2011 ; G. N. Wong et al. 2021 ) \begin{split}\dot{n}&=\frac{1}{\sqrt{-g}}\frac{dN}{d^{3}x\,dt}\\ &=\frac{1}{2}\iint\frac{d^{3}k}{\sqrt{-g}k^{t}}\frac{d^{3}k^{\prime}}{\sqrt{-g}k^{\prime t}}\mathcal{F}(x,k)\mathcal{F}(x,k^{\prime})\varepsilon_{\rm CM}^{2}\sigma_{\gamma\gamma}c,\end{split} (14) where g is the determinant of g_{\mu\nu} . Here d^{3}k\equiv dk_{1}\,dk_{2}\,dk_{3} denotes the volume element in photon momentum space, where the indices 1,2,3 denote the spatial components. \sigma_{\gamma\gamma} is the cross section for \gamma\gamma\rightarrow e^{+}e^{-} ( G. Breit & J. A. Wheeler 1934 ) : \begin{split}\frac{\sigma_{\gamma\gamma}}{\sigma_{T}}&=\frac{3}{8\varepsilon_{\rm CM}^{6}}\left[(2\varepsilon_{\rm CM}^{4}+2\varepsilon_{\rm CM}^{2}-1)\cosh^{-1}(\varepsilon_{\rm CM})\right.\\ &\qquad\left.-\varepsilon_{\rm CM}(\varepsilon_{\rm CM}^{2}+1)\sqrt{\varepsilon_{\rm CM}^{2}-1}\right],\end{split} (15) and \varepsilon_{\rm CM} is the photon energy measured in the center-of-momentum (CM) frame: \varepsilon_{\rm CM}=-u_{\mu,{\rm CM}}k^{\mu}=-u_{\mu,{\rm CM}}k^{\prime\mu}=\left(\frac{-k_{\mu}k^{\prime\mu}}{2}\right)^{1/2}, (16) where u^{\mu}_{\rm CM} is the four-velocity of the CM frame. Eq. ( 14 ) is invariant under general coordinate transformations. This is because \sqrt{-g}\,d^{3}x\,dt , d^{3}k/(\sqrt{-g}\,k^{t}) , and the distribution function \mathcal{F}(k) are coordinate-invariant, while the photon energy measured by CM and the cross section are scalars. Therefore, the pair production rate \dot{n} is a scalar. In addition, the local energy–momentum deposition rate due to the pair production is given by (see e.g., M. Mościbrodzka et al. 2011 ) \begin{split}\dot{Q}^{\mu}&=\frac{1}{2}\iint\frac{d^{3}k}{\sqrt{-g}\,k^{t}}\frac{d^{3}k^{\prime}}{\sqrt{-g}\,k^{\prime t}}\,(k^{\mu}+k^{\prime\mu})\\ &\qquad\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\>\times\mathcal{F}(x,k)\mathcal{F}(x,k^{\prime})\varepsilon_{\rm CM}^{2}\sigma_{\gamma\gamma}c.\end{split} (17) II.5 A quasi-steady self-regulated state in the reconnection region At the onset of reconnection, a current sheet forms near the equatorial plane of the accretion disk. The current sheet is initially supplied with plasma from the accretion flow, and subsequently from the transition region between the accretion flow ( \sigma_{\rm acc}\sim 1 ) and the magnetosphere ( \sigma_{\rm mag}\gg 1 ) ( A. Chow et al. 2026 ) . Once reconnection begins, high-energy synchrotron photons are gradually produced and can create e^{\pm} pairs in the vicinity of the current sheet. This process self-regulates the upstream magnetization parameter \sigma ( H. Hakobyan et al. 2019 ; H. Hakobyan et al. 2023 ; A. Y. Chen et al. 2023 ; R. Kuze et al. 2024 ) . Then we need to check if this value of \sigma is consistent with the condition of steady magnetic reconnection. The condition of steady magnetic reconnection is determined by whether the current sheet can support the required electric current, which is set by Ampère’s law. This condition strongly depends on the thickness of the reconnection current sheet. Recent local PIC simulations suggest that the current sheet thickness is roughly of the order of the Larmor radius in the kinetic regime ( A. Chernoglazov et al. 2023 ; H. Hakobyan et al. 2023 ; H. Hakobyan et al. 2025 ) r_{\rm L}=\frac{\langle\tilde{\gamma}\rangle m_{e}c^{2}}{e\tilde{B}}, (18) where \langle\tilde{\gamma}\rangle={\int d\tilde{\gamma}\,\tilde{\gamma}(dN_{e}/d\tilde{\gamma})}/{\int d\tilde{\gamma}\,(dN_{e}/d\tilde{\gamma})} denotes the average Lorentz factor of electrons. This is roughly \langle\tilde{\gamma}\rangle\sim\sigma for the weak cooling regime and \langle\tilde{\gamma}\rangle\sim\gamma_{\rm syn} for the strong cooling regime. From Ampère’s law, the current density during magnetic reconnection can be approximated as \tilde{J}=(c/4\pi)\tilde{\nabla}\times\tilde{B}\approx(c/4\pi)\tilde{B}/r_{L} . Equating this with \tilde{J}\approx\tilde{n}_{e}ec , we obtain the number density of charged particles required to carry the electric current in the kinetic regime: \tilde{n}_{e}\gtrsim\frac{\tilde{B}^{2}}{4\pi\langle\tilde{\gamma}\rangle m_{e}c^{2}}. (19) When e^{\pm} pairs are produced by high-energy photons in the vicinity of the reconnection region, i.e., \tilde{n}_{e}\approx\dot{n}r_{g}/c , the condition under which the magnetization parameter self-regulates while supplying the electric current required for magnetic reconnection can be expressed as \langle\tilde{\gamma}\rangle\gtrsim\frac{\tilde{B}^{2}}{4\pi\dot{n}\dfrac{r_{g}}{c}m_{e}c^{2}}=\sigma. (20) This equation suggests that quasi-steady magnetic reconnection is difficult to sustain in the strong cooling regime. 2 2 2 A. Chernoglazov et al. 2023 shows that the current sheet in the strong regime ( \gamma_{\rm syn}=0.2\sigma ) becomes thinner than in the weak cooling regime, although they do not discuss whether the magnetic reconnection can be sustained as a quasi-steady state under condition of even stronger cooling. If \sigma satisfies Eq. ( 20 ), a quasi-steady reconnection state is expected to be established. In this study, we determine the value of \sigma near the reconnection region, by which we can calculate the pair creation in the entire magnetosphere. II.6 Model parameters We summarize our model parameters. Our model is characterized by four parameters; the black hole mass M , the mass accretion rate \dot{M} , the spin parameter a , and the anisotropy parameter \xi . The magnetic field strength is determined by the black hole mass and the accretion rate under the MAD condition (see Appendix B ) ,and is therefore not a free parameter. The spin parameter a governs the structure of the spacetime and the anisotropy parameter \xi determines the strength of the photon anisotropy (see Section II.3 ). Note that we fix the size parameter of the reconnection region to f_{l}=2 corresponding to the mini-flare phase, although it may vary between the mini-flare and large-flare phases ( B. Ripperda et al. 2022 ) . We adopt values appropriate for M87, namely M=6.5\times 10^{9}\,M_{\odot} and \dot{m}=\dot{M}/\dot{M}_{E}=5.0\times 10^{-5} ( Event Horizon Telescope Collaboration et al. 2019a ; Event Horizon Telescope Collaboration et al. 2019b ; Event Horizon Telescope Collaboration et al. 2021 ; S. S. Kimura & K. Toma 2020 ; K. Hada et al. 2024 ) , where \dot{M}_{E}=1.4\times 10^{26}\,M_{9}\;{\rm g\;s^{-1}} denotes the Eddington accretion rate. Below we investigate how a and \xi affect the spatial distribution of injected pairs. The dependence of the pair production rate on M and \dot{M} is presented in a separate paper. III Method Figure 3 : Null geodesic ray-tracing in Kerr spacetime. From each point where pair creation occurs, 7,200 photons are isotropically emitted in the ZAMO frame and traced backward in time (red lines). The photon trajectories are followed until they either intersect the reconnection region (cyan region), fall into the black hole horizon, or escape to large radii ( \sim 30\,r_{g} ). We compute the pair production rate in each grid cell using a ray-tracing method similar to that of R. Birkl et al. 2007 ; I. Zalamea & A. M. Beloborodov 2011 . The spatial grid is constructed in spherical coordinates (r,\theta,\varphi) over the domain r_{h}r_{\rm stag} ). The stagnation surface corresponds to the separation point between inflow and outflow under the steady cold MHD condition; thus, plasma supplied at r>r_{\rm stag} is expected to be ejected and contribute to the jet. The corresponding injection rate to the jet is \dot{N}=\int_{r>r_{\rm stag}}\dot{n}\sqrt{-g}\;d^{3}x\approx 1.0\times 10^{43}\ {\rm s^{-1}}. (23) This estimate is conservative, as it neglects pairs that may be produced with outward momentum and thus contribute to the outflow even if they are created inside the stagnation surface (see Section V.1 ). The synchrotron emission power from the jet can be written as L_{\rm syn}\approx\dot{N}\,T_{\rm dur}\,P_{\rm syn}, (24) where T_{\rm dur}\approx r_{g}/(\beta_{\rm rec}c)\simeq 5.0\times 10^{4}\,M_{9}\beta_{\rm rec,-1}^{-1}\ {\rm s} is the reconnection duration time ( L. Sironi et al. 2025 ) , and P_{\rm syn}=B_{\rm dis}\sigma_{T}\nu_{\rm syn}m_{e}c^{2}/(3e) is the synchrotron power emitted by a single electron. Here, \nu_{\rm syn} is the synchrotron photon frequency, and B_{\rm dis} denotes the magnetic field strength at the dissipation radius r_{\rm dis} . Assuming that the toroidal magnetic field dominates in the distant jet region, it can be approximated as B_{\rm dis}\approx B_{\rm mad}(r_{\rm dis}\sin\theta_{j}/r_{h})^{-1} , where B_{\rm mad} is the magnetic field strength at the horizon (see Appendix B ) and \theta_{j}\approx 0.1\,\mathrm{rad} is the jet opening angle. Figure 5 : Comparison of the spatial distribution of pair injection into the magnetosphere on the meridional plane at \varphi=0 for different spacetime geometries (the equatorial-plane distribution is shown in Appendix D , Figure 11 ). Panels (a)–(d) show the density distributions for Minkowski, Schwarzschild, Kerr metric with a=0.5 , and a=0.9375 , respectively. The Goldreich–Julian density is evaluated using the Kerr black hole with a=0.9375 . In our scenario, the pairs injected at the jet base produce synchrotron emission at \nu_{\rm syn}=43\,\mathrm{GHz} with a luminosity of L_{\rm syn,43\,GHz}\approx 10^{38}\,\mathrm{erg\ s^{-1}} at a dissipation radius of r_{\rm dis}=10^{3}r_{g} in M87, which is consistent with the previous analytical estimate of S. S. Kimura et al. 2022 . VLBI observations at 43 GHz detect radio emission from the jet at r_{\rm dis}\sim 10^{2} – 10^{3}r_{g} with luminosities of roughly 10^{37} – 10^{38}\,\mathrm{erg\ s^{-1}} ( R. C. Walker et al. 2018 ) . This indicates that our scenario can supply sufficient matter loading to account for the observed radio jet. Finally, we compare the energy injection rate with the BZ power. The energy injection rate into the magnetosphere is ( I. Zalamea & A. M. Beloborodov 2011 ) L_{\pm}=-\int_{r>r_{h}}\left(g_{tt}\dot{Q}^{t}+g_{t\varphi}\dot{Q}^{\varphi}\right)\sqrt{-g}\;d^{3}x, (25) where we obtain L_{\pm}\approx 3.2\times 10^{38}\,\mathrm{erg\,s^{-1}} in our numerical calculations. The BZ power can be expressed as L_{\rm BZ}=\frac{\kappa}{4\pi}\left(\frac{\Omega_{H}r_{g}}{c}\right)^{2}\Phi_{\rm BH}^{2}f(\Omega_{H})\dot{M}c^{2}, (26) where f(\Omega_{H})=1+1.38(\Omega_{H}r_{g}/c)^{2}-9.2(\Omega_{H}r_{g}/c)^{4} , \Omega_{H}=ac/(2r_{h}) is the angular velocity of the horizon, \kappa is a magnetic geometry factor, and \Phi_{\rm BH} is the dimensionless magnetic flux threading the black hole (see A. Tchekhovskoy et al. 2010 ; A. Tchekhovskoy et al. 2011 for details). Adopting \kappa=0.053 and \Phi_{\rm BH}=50 , corresponding to a split-monopole configuration and the MAD state ( A. Tchekhovskoy et al. 2011 ; R. Narayan et al. 2012 ) , we obtain {L_{\rm BZ}}/{L_{\pm}}\approx 10^{5}\gg 1 for M87. This result indicates that the electromagnetic energy strongly dominates over the particle energy; therefore, pair injection driven by radiative reconnection is unlikely to significantly perturb or disrupt the current structure of the magnetosphere. This is also related with the fact that the resultant magnetization parameter \sigma\gg 1 . IV.2 Influence of general relativistic effects We next investigate how the spacetime of a black hole influences the spatial distribution of injected pairs. Even though a non-spinning black hole cannot drive a BZ jet ( E. Figueiredo et al. 2026 ) , we consider not only Kerr metric with spin parameters a=0.5 and 0.9375 , but also a Minkowski and Schwarzschild metric. Figure 5 shows the spatial distribution of injected pairs on the meridional plane at \varphi=0 for each spacetime metric. The magnetization parameter in the reconnection region, self-regulated by pair creation, is summarized in Table 2 , and we confirm that all cases lie in the weak cooling regime. From the spatial distributions in Figure 5 , we find two key results. First, the maximum multiplicity in the Minkowski and Schwarzschild cases ( \tilde{n}_{\pm}/n_{\rm GJ}\sim 10^{8} ) is smaller than in the Kerr cases ( \tilde{n}_{\pm}/n_{\rm GJ}\sim 10^{9} – 10^{10} ). This difference arises because, in the absence of frame dragging, no toroidal magnetic field is generated, and the larger horizon radius ( r_{h}=2\,r_{g} ) further reduces the field strength. Metric {\sigma} \gamma_{\rm syn} \dot{N}_{r>r_{h}}\;\left(\,\rm s^{-1}\right) Minkowski 1.7\times 10^{5} 6.7\times 10^{6} 3.6\times 10^{42} Schwarzschild 1.7\times 10^{5} 6.7\times 10^{6} 6.2\times 10^{42} Kerr ( a=0.5 ) 1.4\times 10^{5} 6.4\times 10^{6} 1.2\times 10^{43} Kerr ( a=0.9375 ) 5.6\times 10^{4} 5.1\times 10^{6} 6.0\times 10^{43} Table 2 : Comparison of the self-regulated upstream magnetization parameter and the synchrotron burn-off limit of the reconnection region, and the pair production rate outside the horizon for different spacetime metrics. In all cases, the self-regulated magnetization satisfies \sigma<\gamma_{\rm syn} . Second, the spatial distribution strongly depends on the spacetime geometry. In the Minkowski metric, where spacetime is flat, photon trajectories are not bent and do not wrap around the central object; as a result, pair production occurs only in the vicinity of the reconnection region. In contrast, in the Schwarzschild case, photon trajectories can bend around the black hole, allowing pairs to be produced on the opposite side of the reconnection region (see also the equatorial-plane distribution in Fig. 11 ). A particularly interesting result is obtained for spinning black holes: On the meridional plane, pairs are also injected near the rotation axis, i.e., the spine region. This feature is clearly seen in the polar-angle distribution of injected pairs at r=3\,r_{g} (Figure 6 ), which shows that the injected pairs into the spine region in the Kerr black hole is enhanced by \sim 10^{3} – 10^{5} compared to the non-spinning black hole. This behavior originates from the fact that photon orbit in the Kerr metric is not confined to a fixed orbital plane, as the fundamental frequencies in the radial, polar ( \theta ), and azimuthal ( \varphi ) directions are generally incommensurate ( R. Fujita & W. Hikida 2009 ; M. van de Meent 2020 ) . As a result, the fraction of head-on collisions near the rotation axis increases compared to the Schwarzschild case, enhancing pair injection into the spine region. The same effect leads to a broader spatial distribution of pairs in the jet region as the spin parameter increases, and the pair production rate correspondingly becomes higher for larger spins (Table 2 ). These results suggest that black hole spin plays an important role not only in energy injection but also in the spatial distribution of the injected plasma into the jet. Figure 6 : Polar-angle ( \theta ) distribution of the injected pair plasma at r=3\,r_{g} shown in Figure 5 . IV.3 Influence of photon anisotropic effects Figure 7 : Comparison of the spatial distribution of injected pairs on the meridional plane at \varphi=0 for different degrees of anisotropy, \xi=0.1,0.25,1.0, and 1000 (see Fig. 2 ), adopting a spin parameter a=0.9375 . The corresponding equatorial-plane distributions are shown in Section D (Fig. 12 ). The results presented so far have primarily assumed isotropic photons ( \xi=1000 ); however, radiation from relativistic electrons accelerated by magnetic reconnection is intrinsically anisotropic due to the electromagnetic field structure and relativistic beaming effect ( A. Chernoglazov et al. 2023 ) . In this section, we examine the spatial distribution of injected pairs for four representative cases, \xi=0.1,0.25,1.0, and 1000 (see Section II.3 ), adopting a spin parameter a=0.9375 , in order to assess the impact of photon anisotropy. Within the parameter range adopted in this work, all cases fall within the weak cooling regime (see Table 3 ), in which the photon distribution is preferentially concentrated along the direction of the reconnection electric field (see Section II.3 ). Figure 7 shows the spatial distribution of injected pairs on the meridional plane at \varphi=0 for different degrees of photon anisotropy. We find that the injected pairs are similarly concentrated near the reconnection region, while the anisotropy suppresses pair injection into the vicinity of the rotation axis even in the Kerr metric. This behavior arises because the reconnection electric field is oriented predominantly within the equatorial plane; consequently, increasing anisotropy leads to a progressively stronger concentration of photons along directions parallel to the equatorial plane. For such a photon distribution, a large fraction of photons satisfy the conserved quantity of geodesic motion, the Carter constant Q\simeq 0 , which suppresses oscillatory motion in the \theta direction. Consequently, photon trajectories are confined near the equatorial plane, leading to a suppression of pair production in the vicinity of the black hole rotation axis. Therefore, in contrast to the isotropic case, stronger anisotropy confines pair production to a limited region just above the magnetic reconnection site, rather than producing a spatial distribution that extends throughout the entire jet region. This trend is also evident in the polar angle distribution of injected pairs at r=3\,r_{g} (Figure 8 ). We note that Figure 8 indicates that the injected pair density still exceeds the Goldreich–Julian density in the jet region, although the plasma density near the rotation axis is reduced by up to \sim 3 orders of magnitude compared to the isotropic case. In addition, the anisotropy reduces the total pair production rate within the jet (see Table 3 ) because the injected pairs no longer populate the entire jet region. Nevertheless, our scenario still supplies a sufficient amount of plasma to account for the observed radio emission from the jet, even in strongly anisotropic cases. \xi {\sigma} \dot{N}_{r>r_{h}}\;\left(\;\rm s^{-1}\right) L_{\rm 43GHz}\;\left(\rm erg\;s^{-1}\right) 0.1 1.3\times 10^{3} 2.1\times 10^{42} 1.3\times 10^{37} 0.25 1.1\times 10^{4} 1.4\times 10^{43} 8.3\times 10^{37} 1.0 3.5\times 10^{4} 6.8\times 10^{43} 4.1\times 10^{38} 1000 5.6\times 10^{4} 6.0\times 10^{43} 3.5\times 10^{38} Table 3 : Comparison of the self-regulated upstream magnetization parameter of the reconnection region, the pair production rate, and the radio synchrotron emission at r=1000\,r_{g} at 43 GHz for different degrees of anisotropy. Within the parameter range considered here, the self-regulated magnetization satisfies \sigma<\gamma_{\rm syn} in all the cases. In addition, the radio luminosity reaches L_{43\,\mathrm{GHz}}\gtrsim 10^{37}\,\mathrm{erg\ s^{-1}} , indicating that a sufficient amount of plasma is supplied to account for the observed radio emission from the M87 jet. Figure 8 : Polar-angle ( \theta ) distribution of the injected pair plasma at r=3\,r_{g} (white dashed line) shown in Figure 7 . V Discussion The spatial distribution of injected plasma in our scenario has significant implications for the jet acceleration and very-high-energy flares from the jet base. In the following, we discuss these implications in detail focusing on the high-spin case with a=0.9375 and \xi=1.0 . 3 3 3 For the M87 parameters, we obtain the upstream magnetization parameter of \sigma/\gamma_{\rm syn}\sim 0.01 . This regime suggests that photon anisotropy may not play a dominant role, according to recent global PIC simulations (see Figure 14 of A. Chernoglazov et al. 2023 ). Although a certain level of anisotropy is present, we assume it to be moderate and approximate it by adopting \xi=1.0 in this section. V.1 Jet Acceleration by pressure gradient In Section IV.2 , we demonstrated that a spinning black hole plays a key role in facilitating pair injection across the entire jet region, as seen in the spatial distribution of Figure 5 . The injected plasma emits synchrotron radiation in the magnetosphere; however, the plasma may not cool efficiently and could instead become thermalized due to the high synchrotron self-absorption (SSA) opacity ( S. S. Kimura et al. 2022 ) . Injected pairs inside the stagnation surface—defined as the surface where the Lorentz force balances gravity—can be accelerated via adiabatic expansion by the thermal pressure gradient. Here, based on the results of our numerical calculations, we examine the thermalization of the injected pairs and its implication on the bulk acceleration. Figure 9 : Temperature distribution of SSA-thermalized injected plasma in M87 ( a=0.9375 ). The distribution is shown only in regions satisfying t_{\rm SSA}m_{e}c^{2} , indicating that the mean lorentz factor of the thermalized pairs is \langle{\hat{\gamma}}_{\pm}\rangle\approx 3kT_{e}/(m_{e}c^{2})\sim 10 . This value is larger than the estimate under the assumption of head-on collisions of photons in S. S. Kimura et al. 2022 by a factor of a few. This difference arises because the collisions of higher energy photons with smaller angles contribute to the pair creation. Such high-temperature plasma can be accelerated to relativistic speeds by the thermal pressure gradient even when located inside the stagnation surface. This result is also different from the conventional picture of MHD jet acceleration. MHD simulation suggests that there is little current crossing magnetic field lines near the jet spine, making Lorentz-force acceleration inefficient and leading to a non-relativistic flow and magnetic self-collimation near the axis. The whole jet is collimated by the external pressure from the disk wind, leading to a magnetic field configuration in the sheath that expands more widely than the spherical profile toward larger radii, with efficient acceleration occurring (see panel (a) of Figure 10 ) ( S. S. Komissarov et al. 2009 ; Y. Lyubarsky 2009 ; M. Nakamura et al. 2018 ) . In contrast, our results show that e^{\pm} pairs can be injected into the vicinity of the black hole rotation axis in Kerr spacetime, then, these pairs are efficiently thermalized via SSA and can be accelerated to relativistic speeds by the thermal pressure gradient. If relativistic plasma flows are formed even near the jet spine, the contribution of the outward electric field stress can no longer be neglected. As a result, magnetic self-collimation becomes less efficient, and it becomes difficult to form a magnetic field configuration that expands significantly beyond the spherical profile at the sheath (see Figure 10 , panel (b)). This implies that the efficiency of converting electromagnetic energy (or thermal energy) into plasma kinetic energy via the magnetic pressure gradient of the toroidal field (or thermal gradiant pressure) may be ineffective in the jet sheath. This is likely consistent with observations of active galactic nucleus jets, which suggest that jet acceleration is slower than that predicted by ideal MHD ( M. Nakamura et al. 2018 ; J. Park et al. 2019 ) . Our results indicate that the discrepancy between theoretical models of jet acceleration and observations could be explained by the thermal pressure gradient of plasma near the spine, which has not been fully considered in previous studies. The validity of this picture should be confirmed by numerical simulations of acceleration of injected matter. V.2 Very-High-Energy Emission from the Jet Base Very-high-energy ( 300\,{\rm GeV} – 10\,{\rm TeV} ) flares have been repeatedly observed in some active galactic nuclei, including M87 ( J. C. Algaba et al. 2024 ) . These flares exhibit rapid variability on timescales of days, suggesting a compact emission region located in the vicinity of the central black hole, i.e., at the jet base ( K. Hada et al. 2024 ) . The origin of the energetic particles responsible for TeV emission remains an open question. Two main scenarios have been proposed to explain the variable TeV emission: Large-scale magnetic reconnection in the MAD ( B. Ripperda et al. 2022 ; H. Hakobyan et al. 2023 ; S. Solanki et al. 2025 ) , and particle acceleration in spark gaps in BZ jets ( A. Levinson & F. Rieger 2011 ; K. Hirotani et al. 2016 ; B. Crinquand et al. 2020 ; S. Kisaka et al. 2022 ) . These scenarios can be significantly affected by the spatial distribution of pairs predicted in our matter-loading scenario. The physical properties of large-scale magnetic reconnection are essentially the same as those considered in our model, except for the size of the reconnection region (i.e., the current sheet). In the large-scale magnetic reconnection scenario, a current sheet of size \sim 10\,r_{g} is formed ( B. Ripperda et al. 2022 ) , and pairs accelerated by reconnection upscatter soft photons from the accretion disk via inverse Compton scattering to produce TeV photons. However, our results suggest that the self-regulated magnetization parameter is \sigma\sim 10^{5} . In this case, inverse Compton scattering of \varepsilon_{\rm soft}\approx 10^{-3}\,{\rm eV} photons by pairs with \sigma\sim 10^{5} occurs in the Thomson regime, yielding photon energies of only \sigma^{2}\varepsilon_{\rm soft}\sim 10\,{\rm MeV} , which is insufficient to explain the observed TeV emission. Even if the scattering enters the Klein–Nishina regime, the maximum photon energy is limited to \sigma m_{e}c^{2}\sim 100\,{\rm GeV} , which is still insufficient to account for TeV emission. Although this value of \sigma\sim 10^{5} is obtained near the black hole, in the case of large-scale magnetic reconnection at r\sim 10\,r_{g} , the magnetic field strength is expected to be lower than that near the horizon. This would reduce the pair production rate, potentially leading to a higher \sigma and possibly allowing inverse Compton scattering to produce TeV photons. However, at larger distances, the power released by reconnection decreases, which may make it difficult to account for the energetics required for TeV flares ( A. Y. Chen et al. 2023 ) . As discussed in S. Solanki et al. 2025 , the anisotropy of inverse Compton emission and the formation of current sheets away from the equatorial plane are likely to play a crucial role in resolving this issue. The spark-gap scenario operates in a low-density region within a BZ jet, where the plasma density falls below the Goldreich–Julian density and the electric field along the magnetic field cannot be fully screened. Under such conditions, particles are efficiently accelerated by the unscreened electric field, and the resulting pairs produce TeV gamma rays via inverse Compton scattering of disk photons. However, this scenario appears to be in tension with our matter-loading scenario. In our scenario, the plasma density exceeds the Goldreich–Julian density throughout the jet even in the presence of photon anisotropy (see Section IV ), such that the current required to sustain the BZ process is fully carried, leading to efficient screening of the electric field. Nevertheless, the results of E. Figueiredo et al. 2026 suggest a possible coexistence of our matter-loading scenario and the spark-gap scenario. Although their study is based on a 2D global GRPIC simulation of Bondi accretion onto a Schwarzschild black hole, they identify three main stages which repeat cyclically in the MAD; (1) an ideal advection phase, (2) a reconnection-regulated phase, and (3) a large-scale reconnection phase (similar to B. Ripperda et al. 2022 ). Our matter-loading scenario primarily operates during phases (2) and (3), supplying plasma to the BZ jet. In these phases, the electric field is expected to be screened throughout the jet region as discussed above. In contrast, during the advection phase (1), magnetic reconnection is not efficiently driven, and our matter-loading mechanism becomes ineffective. As a result, the magnetosphere near the black hole may transition to a low-density state because of jet launching and infall to the black hole. In this regime, the plasma density can fall below the Goldreich–Julian density, allowing spark gaps to form and potentially power TeV flares. It should be noted, however, that the results of E. Figueiredo et al. 2026 are based on 2D simulations, and it remains unclear whether these three phases persist in 3D systems. Future studies using global 3D GRPIC simulations will be necessary to verify the existence of these phases and to assess their role in jet dynamics and high-energy emission. VI Summary In this work, we have derived the spatial distribution of injected plasma by calculating the e^{\pm} pair production from high-energy photons originating from non-axisymmetric magnetic reconnection, as suggested by high-resolution 3D GRMHD simulations. Our calculation employs general relativistic ray tracing, taking into account curved spacetime, photon-field anisotropy, and collision angles. We find that pair injection driven by non-axisymmetric magnetic reconnection can supply a sufficient number of particles to account for the observed jet emission as claimed by S. S. Kimura et al. 2022 . We also find that the self-regulated magnetization achieved through pair creation is consistent with the values reported in A. Y. Chen et al. 2023 ; R. Kuze et al. 2024 . A key new result of this study is that this conclusion remains valid even when the anisotropy of the photon field is taken into account, thereby establishing a viable plasma source for the jet. We also demonstrate a new finding of this study: A spinning black hole plays a crucial role not only in powering the jet via the BZ process but also in injecting plasma into it. Owing to frame-dragging effects, photon trajectories are no longer confined to a single plane, which enhances head-on collisions in the spine region and leads to efficient pair injection. As a result, relativistic outflows can be formed in the spine via the thermalization by SSA and the adiabatic expansion while energy conversion in the sheath will become less efficient. This may lead to slower sheath flows than those typically found in MHD simulations, potentially bringing theoretical predictions into better agreement with observations. In addition, we find that the spatial distribution of the injected plasma can significantly influence very-high-energy emission scenarios, such as large-scale magnetic reconnection and spark gaps. In this study, we have evaluated the pair production rate assuming the mass and accretion rate of M87. It is important to investigate whether this matter-loading scenario can be applied to other black hole jet systems. A systematic parameter survey over black hole mass and accretion rate will therefore be essential in separate work. Furthermore, during thermal acceleration in the spine, inverse Compton cooling and adiabatic cooling may become important, and it remains unclear to what extent the spine–sheath structure can be maintained within the jet. Since the formation of a spine–sheath structure is crucial for understanding relativistic outflows in blazars, this issue warrants further investigation. Incorporating thermal pressure acceleration due to the SSA process into MHD simulations will be an important direction for future work. We would like to thank Tomohisa Kawashima, Koki Kin, Shota Kisaka, Riku Kuze, and Kengo Tomida for their useful comments. The numerical computations in this study are performed on the XD-B system at the National Astronomical Observatory of Japan and the Genkai supercomputer at Kyushu University. This work was supported by Graduate Program on Physics for the Universe (GP-PU). Appendix A The Kerr spacetime and the 3+1 electrodynamics We use Boyer–Lindquist coordinates (t,\varphi,r,\theta) and adopt units in which GM=c=1 (so that r_{g}=GM/c^{2}=1 ). The Kerr spacetime ( R. P. Kerr 1963 ) can be written as ds^{2}=g_{\mu\nu}dx^{\mu}dx^{\nu}=-\alpha^{2}dt^{2}+\gamma_{\varphi\varphi}(d\varphi-\Omega dt)^{2}+\gamma_{rr}dr^{2}+\gamma_{\theta\theta}d\theta^{2}, (A1) where \alpha=\sqrt{\frac{\rho^{2}\Delta}{\Sigma}},\quad\Omega=\frac{2ar}{\Sigma},\quad\gamma_{\varphi\varphi}=\frac{\Sigma\sin^{2}\theta}{\rho^{2}},\quad\gamma_{rr}=\frac{\rho^{2}}{\Delta},\quad\gamma_{\theta\theta}=\rho^{2}. (A2) Here, we have defined \Sigma=(r^{2}+a^{2})^{2}-a^{2}\Delta\sin^{2}{\theta} , \Delta=r^{2}+a^{2}-2r , \rho^{2}=r^{2}+a^{2}\cos^{2}{\theta} , where a denotes the dimensionless spin parameter. The Kerr spacetime admits two Killing vectors associated with stationarity and axisymmetry, \xi^{\mu}_{(t)}=(1,0,0,0),\quad\xi^{\mu}_{(\varphi)}=(0,1,0,0). (A3) In Boyer–Lindquist coordinates, the four-velocity of an observer whose worldline is normal to hypersurfaces of constant t is given by n^{\mu}=\left(\frac{1}{\alpha},\frac{\Omega}{\alpha},0,0\right),\qquad n_{\mu}=(-\alpha,0,0,0). (A4) This observer is referred to as a zero-angular-momentum observer (ZAMO), since n_{\mu}\xi^{\mu}_{(\varphi)}=0 . The ZAMO has a local orthonoramal basis ( K. S. Thorne et al. 1986 ; S. S. Komissarov 2004 ; K. Toma & F. Takahara 2014 ) , e_{(\hat{t})}=\frac{\partial_{t}+\Omega\partial_{\varphi}}{\alpha},\quad e_{(\hat{\varphi})}=\frac{\partial_{\varphi}}{\sqrt{\gamma_{\varphi\varphi}}},\quad e_{(\hat{r})}=\frac{\partial_{r}}{\sqrt{\gamma_{rr}}},\quad e_{(\hat{\theta})}=\frac{\partial_{\theta}}{\sqrt{\gamma_{\theta\theta}}}, (A5) w^{(\hat{t})}=\alpha dt,\quad w^{(\hat{\varphi})}=\sqrt{\gamma_{\varphi\varphi}}(d\varphi-\Omega dt),\quad w^{(\hat{r})}=\sqrt{\gamma_{rr}}dr,\quad w^{(\hat{\theta})}={\sqrt{\gamma_{\theta\theta}}}d\theta. (A6) Physical quantities expressed in the ZAMO orthonormal basis are denoted by a hat, e.g., A^{\hat{\mu}} or \hat{A} . The covariant Maxwell equations \nabla_{\nu}F^{\mu\nu}=4\pi I^{\mu} , \nabla_{\nu}{}^{*}F^{\mu\nu}=0 can be reduced by introducing the four vectors D^{\mu}=n_{\nu}F^{\mu\nu} , B^{\mu}=-n_{\nu}{}^{*}F^{\mu\nu} , E_{\mu}=\xi_{(t)}^{\nu}F_{\mu\nu} , and H_{\mu}=-\xi_{(t)}^{\nu}{}^{*}F_{\mu\nu} to L. D. Landau & E. M. Lifshitz 1975 ; S. S. Komissarov 2004 \nabla\cdot\bm{B}=0,\quad\frac{\partial\bm{B}}{\partial t}+\nabla\times\bm{E}=0, (A7) \nabla\cdot\bm{D}=4\pi\rho_{e},\quad-\frac{\partial\bm{D}}{\partial t}+\nabla\times\bm{H}=4\pi\bm{J}, (A8) where, \rho_{e}=-I^{\mu}n_{\mu} and J^{\mu}=(\xi_{(t)}^{\mu}I^{\nu}-\xi_{(t)}^{\nu}I^{\mu})n_{\nu} . \nabla\cdot\bm{A} and \nabla\times\bm{A} denote (1/\sqrt{\gamma})\partial_{i}(\sqrt{\gamma}A^{i}) and (1/\sqrt{\gamma})\varepsilon^{ijk}\partial_{j}A_{k} , and \varepsilon^{ijk} is the Levi-Civita pseudo-tensor. \bm{D} and \bm{B} are the electric and magnetic fields measured by ZAMO, while \bm{E} and \bm{H} are those in the coordinate basis, and they have constitutive relations, \bm{E}=\alpha\bm{D}-\Omega\bm{m}\times\bm{B},\quad\bm{H}=\alpha\bm{B}+\Omega\bm{m}\times\bm{D}, (A9) where \bm{m}=\partial_{\varphi} . Appendix B Electromagnetic field model around black hole High resolution GRMHD simulation by B. Ripperda et al. 2022 suggests that the accretion disk in the MAD becomes geometrically thin and the magnetic field configuration approaches a split-monopole geometry within r\lesssim 5r_{g} . The force-free condition provides a good approximation between the inner and light surfaces. In this paper, we take the assumptions of stationarity, axisymmetry, a force-free condition ( F_{\mu\nu}I^{\nu}=0 ), and a split-monopole magnetic field configuration ( B^{\theta}=0 ), and we model electomagnetic field around the black hole in the same manner as S. S. Kimura et al. 2022 . Under the assumptions of force-free conditions, stationarity, and axisymmetry, the electric fields \bm{E} and \bm{D} can be expressed, using Equations ( A7 ) and ( A9 ), as \bm{E}=-\Omega_{F}\bm{m}\times\bm{B},\qquad\bm{D}=\frac{1}{\alpha}(\Omega-\Omega_{F})\bm{m}\times\bm{B}, (B1) where \Omega_{F} denotes the angular velocity of magnetic field lines induced by the spinning black hole ( a\neq 0 ); notably, \Omega_{F} is conserved along each field line. In this work, we adopt \Omega_{F}=a/4r_{h} ( R. D. Blandford & R. L. Znajek 1977 ) , where r_{h}=r_{g}\left(1+\sqrt{1-a^{2}}\right) is the horizon radius. For the non-zero components of the magnetic field, assuming a split-monopole configuration as well as stationarity and axisymmetry, Eqs. ( A7 ) and ( A8 ) lead to the existence of conserved quantities along magnetic field lines: \sqrt{\gamma}B^{r}=\mathrm{const},\qquad H_{\varphi}=\alpha B_{\varphi}=\mathrm{const}. (B2) Using the ZAMO orthonormal basis ( B_{\hat{r}}=\sqrt{\gamma_{rr}}B^{r},\;B_{\hat{\varphi}}=B_{\varphi}/\sqrt{\gamma_{\varphi\varphi}} ), we obtain \frac{B_{\hat{\varphi}}}{B_{\hat{r}}}\propto\frac{\sqrt{\gamma_{\varphi\varphi}}}{\alpha} (B3) for a fixed \theta . Analytical studies of special relativistic MHD flows in magnetically dominated regions ( V. S. Beskin 2010 ; K. Toma & F. Takahara 2013 ) have shown that the magnetic field components satisfy B_{\hat{\varphi}}\approx-B_{\hat{r}} (B4) at the outer light surface. Under this condition we obtain the radial dependences of the field components as shown in Figure 5 of S. S. Kimura et al. 2022 . It is roughly B_{\hat{r}}\sim B_{\hat{\varphi}}\propto r^{-2} . When the accretion flow is in the MAD state, the magnetic field strength on the horizon in the ZAMO basis is approximately given by \hat{B}_{\rm mad}\approx\sqrt{\dot{M}c\Phi_{\rm mad}^{2}/16\pi^{2}r_{h}^{2}} ( F. Yuan & R. Narayan 2014 ; S. S. Kimura et al. 2022 ) , where \Phi_{\rm mad}\approx 50 is the saturated magnetic flux ( A. Tchekhovskoy et al. 2011 ; R. Narayan et al. 2012 ; J. C. McKinney et al. 2012 ) , and \dot{M} is a mass accretion rate. Therefore, by normalizing the field such that the radial magnetic field strength on the horizon becomes \hat{B}_{\rm mad} , we obtain B_{\hat{r}}=\frac{2r_{h}r_{g}}{\sqrt{\Sigma}}\hat{B}_{\rm mad}. (B5) We adopt this solution as the electromagnetic field configuration prior to the onset of magnetic reconnection. Appendix C The orthonormal basis vectors of the drift frame The four-velocity of the drift frame is given by u^{\mu}=u^{t}\left(1,\,\Omega-\frac{\alpha}{\sqrt{\gamma_{\varphi\varphi}}}\frac{D_{\hat{\theta}}B_{\hat{r}}}{\hat{B}^{2}},\,\frac{\alpha}{\sqrt{\gamma_{rr}}}\frac{D_{\hat{\theta}}B_{\hat{\varphi}}}{\hat{B}^{2}},\,0\right). (C1) The orthonormal basis vectors of the drift frame are constructed as ( J. H. Krolik et al. 2005 ; K. Beckwith et al. 2008 ; R. V. Shcherbakov & L. Huang 2011 ; J. Dexter 2016 ) \displaystyle e_{(\tilde{t})}^{\mu} \displaystyle= \displaystyle u^{\mu}, (C2) \displaystyle e_{(\tilde{\varphi})}^{\mu} \displaystyle= \displaystyle(u_{\varphi},0,0,-u_{t})/N_{\varphi}, (C3) \displaystyle e_{(\tilde{r})}^{\mu} \displaystyle= \displaystyle(u_{r}u^{t},-(u_{t}u^{t}+u_{\varphi}u^{\varphi}),0,u_{r}u^{\varphi})/N_{r}, (C4) \displaystyle e_{(\tilde{\theta})}^{\mu} \displaystyle= \displaystyle(u_{\theta}u^{t},u_{\theta}u^{r},1+u_{\theta}u^{\theta},u_{\theta}u^{\varphi})/N_{\theta}, (C5) where \displaystyle N_{r}^{2} \displaystyle= \displaystyle-\gamma_{rr}(u_{t}u^{t}+u_{\varphi}u^{\varphi})(1+u_{\theta}u^{\theta}), (C6) \displaystyle N_{\theta}^{2} \displaystyle= \displaystyle\gamma_{\theta\theta}(1+u_{\theta}u^{\theta}), (C7) \displaystyle N_{\varphi}^{2} \displaystyle= \displaystyle-(u_{t}u^{t}+u_{\varphi}u^{\varphi})\Delta\sin^{2}{\theta}. (C8) Using Eqs. ( C2 )–( C5 ), the electromagnetic fields in the drift frame basis are written as \displaystyle B_{\tilde{\varphi}} \displaystyle= \displaystyle e_{(\tilde{r})}^{\alpha}e_{(\tilde{\theta})}^{{\beta}}F_{\alpha{\beta}}\ =B_{\hat{\varphi}}-\hat{\gamma}v_{d}^{\hat{r}}D_{\hat{\theta}}+\frac{\hat{\gamma}^{2}}{\hat{\gamma}+1}v_{d}^{\hat{r}}\left(v_{d}^{\hat{r}}B_{\hat{\varphi}}-v_{d}^{\hat{\varphi}}B_{\hat{r}}\right), (C9) \displaystyle B_{\tilde{r}} \displaystyle= \displaystyle e_{(\tilde{\theta})}^{\alpha}e_{(\tilde{\varphi})}^{{\beta}}F_{\alpha{\beta}}=B_{\hat{r}}+\hat{\gamma}v_{d}^{\hat{\varphi}}D_{\hat{\theta}}+\frac{\hat{\gamma}^{2}}{\hat{\gamma}+1}v_{d}^{\hat{\varphi}}\left(v_{d}^{\hat{\varphi}}B_{\hat{r}}-v_{d}^{\hat{r}}B_{\hat{\varphi}}\right), (C10) \displaystyle B_{\tilde{\theta}} \displaystyle= \displaystyle e_{(\tilde{\varphi})}^{\alpha}e_{(\tilde{r})}^{{\beta}}F_{\alpha{\beta}}=0, (C11) \displaystyle D_{\tilde{i}} \displaystyle= \displaystyle e_{(\tilde{t})}^{\alpha}e_{(\tilde{i})}^{{\beta}}F_{\alpha{\beta}}=0, (C12) where \hat{\gamma}=\left(1-\hat{\bm{v}_{d}}^{2}\right)^{-1/2} is the Lorentz factor of the drift frame measured by the ZAMO. These expressions explicitly demonstrate that the electric field vanishes locally in the drift frame (Eq. ( C12 )). Moreover, Eqs. ( C9 ) and ( C10 ) show that the transformation between the ZAMO basis and the drift frame basis reduces exactly to the Lorentz transformation. Appendix D Spatial Distribution of Pair-Injected Plasma in the Equatorial Plane Here, we show the spatial distribution of pair-injected plasma on the equatorial plane corresponding to the results discussed in Sections IV.2 and IV.3 . Figure 11 : Comparison of the spatial distribution of pair injection into the magnetosphere on the equatorial plane for different spacetime geometries, corresponding to the equatorial-plane of Figure 5 . Figure 11 shows the spatial distribution of injected plasma on the equatorial plane for the Minkowski, Schwarzschild, and Kerr metric ( a=0.5 and 0.9375 ). Figure 12 shows the spatial distribution on the equatorial plane for different degrees of anisotropy ( \xi=0.1 , 0.25 , 1.0 , and 1000 ) in the Kerr metric with spin parameter a=0.9375 . Figure 12 : Comparison of the spatial distribution of injected pairs on the equatorial plane for different degrees of anisotropy, corresponding to the equatorial-plane of Figure 7 .