Statistics of the projected angles between the black-hole spin and the host-galaxy rotation axes from NewHorizon

TL;DR

The paper investigates whether AGN jets tend to align with the rotation axes of their host galaxies by analyzing projected two-dimensional angles in the NewHorizon simulation, where BH mass and spin evolve self-consistently. It constructs 100 BH–galaxy systems and 5,000 mock observations (DESI-LS and Euclid imaging plus MaNGA-like velocity fields) to measure optical and kinematic position angles and their misalignments with the host spins. The results show a statistically significant jet–galaxy alignment that persists despite modeled scatter in jet orientation and projection effects, with kinematic PA providing a slightly tighter correlation than photometric PA; low-mass galaxies tend to obliterate the signal while high-mass systems maintain stronger alignment. These findings, aligning with recent VLBI and IFU surveys, highlight the value of integrating morphological, kinematic, and polarimetric data to understand BH spin evolution and jet feedback in galaxy co-evolution, and point to future surveys and simulations to further constrain the physics driving jet–galaxy alignment.

Abstract

Understanding the alignment between AGN jets and their host galaxies is crucial for interpreting AGN unification models, jet feedback processes, and the co-evolution of galaxies and their central black holes (BH). In this study, we use the high-resolution cosmological zoom-in simulation NewHorizon, which self-consistently evolves BH mass and spin, to statistically examine the relationship between AGN jet orientation and host galaxy structure. Building upon our previous work, we extend the analysis of projected (2-d) alignment angles to facilitate more direct comparisons with recent observational studies. In our methodology, galaxy orientations are estimated using optical position angles derived from synthetic DESI-LS and Euclid images, while BH spin vectors serve as proxies for AGN jet directions. From a carefully selected sample of 100 BH-galaxy systems at low redshift, we generate a catalog of 5,000 mock optical images using a Monte Carlo approach that samples random viewing angles and redshifts. Our results reveal a statistically significant tendency for AGN jets to align with the orientation of their host galaxies, consistent with recent observations combining Very Long Baseline Interferometry (VLBI) and optical imaging of nearby AGNs. Furthermore, we find a slightly stronger alignment when using kinematic position angles derived from synthetic MaNGA-like stellar velocity fields. These findings underscore the importance of combining morphological, kinematic, and polarimetric information to disentangle the complex interplay between black hole spin evolution, accretion mode, and the galactic environment in shaping the direction of relativistic jets.

Figures (7)

• Figure 1: Variations of the primary BH mass with respect to their host (stellar) galaxy mass at $z\sim0.18$. Our final catalog consists of 100 BHs all in radio mode, including BHs extracted from the two Galactica zooms. Overall, across all stellar mass bins, BHs in NewHorizon are under-massive in comparison to observations, lying up to one orders of magnitude below the bulk of the observed BH masses nhz.
• Figure 2: Variations of the effective radius estimated from synthetic DESI-LS $r$-band images with respect to the galaxy mass (upper panel). For each galaxy, we generated 50 synthetic images by assigning random spatial orientations and redshifts in the range $0.02 < z < 0.18$. For qualitative comparison with observational data, the black line represents the average half-light radius measured from the DESI Early Data Release (EDR), based on a sample of 151,530 galaxies within the same redshift range. The grey shaded area denotes the dispersion around the mean value. Overall, the trend observed in our synthetic sample agrees reasonably well with observational expectations, suggesting that the theoretical estimates of $R_{\mathrm{e}}$ are reliable. In the lower panel, we compare the corresponding $R_{\mathrm{e}}$ values derived from the synthetic Euclid images. On average, the effective radii estimated from the DESI-LS synthetic images are slightly larger than those obtained from Euclid, primarily due to the lower angular resolution of DESI-LS.
• Figure 3: Examples of synthetic optical and kinematic images derived in our analysis. Each row corresponds to a specific galaxy. The first column shows high resolution images using $u$-$g$-$r$ bands, along with the value of the inclination angle ($i$) at which the galactic plan is viewed by the observer peirani+24. The second column presents synthetic DESI-LS-like $r$-band images with a resolution of 0.262 arcsec/pixel, convolved with a PSF having a FWHM of 1.2 arcsec. The black and red dashed lines indicate the orientations of the major and minor photometric axes, respectively. The semi-major axis of each ellipse corresponds to the effective radius ($R_{\mathrm{e}}$), while the orientation of the semi-minor axis serves as a proxy for the direction of the projected stellar angular momentum. For comparison, these panels also include the projected BH spin vector (in black) and the projected stellar angular momentum vector (in cyan), the latter estimated from all stars within the half-mass radius. In the second galaxy, the dotted line and the dotted ellipse indicate measurements at 2$R_{\mathrm{e}}$. The axis ratio $q$ denotes the ratio of the semi-major to semi-minor axes. In the third column, we show the corresponding high-resolution projected velocity fields, while the fourth column displays the synthetic MaNGA-like velocity fields. These latter are computed on a 0.5 $\times$ 0.5 arcsec/pixel grid, based on the interpolation of three dithering fiber configuration illustrated in the top-right corner of the third-column panels. Kinematic position angles (PA$^{\mathrm{kin}}$), derived using the PaFit package krajnovic+06, are indicated with red dashed lines. In the first example, both the optical and kinematic PAs closely trace the direction of the projected stellar spin. In the second example, the optical PA is less reliable due to the near-circular shape of the galaxy at one effective radius ($q \approx 1$), while the kinematic PA more accurately captures the spin direction. The third example features a spheroidal galaxy with a kinematically decoupled core (a central stellar component rotating counter to the outer stellar population). This complex structure is not evident from the optical morphology alone, resulting in a more noticeable misalignment of $\Delta \mathrm{PA^{\mathrm{gal-opt}}}$=15.2$^\circ$. The kinematic estimate performs even worse, with $\Delta \mathrm{PA^{\mathrm{gal-kin}}}$=22.4$^\circ$ and a substantial 3$\sigma_{\mathrm{PA^k}}$ uncertainty of $28.2^\circ$, likely due to the presence of the decoupled core.
• Figure 4: Distributions of $\Delta \mathrm{PA^{\mathrm{gal-opt}}}$ and $\Delta \mathrm{PA^{\mathrm{gal-kin}}}$. This figure presents the distributions of $\Delta \mathrm{PA^{\mathrm{gal-opt}}}$, the 2-d misalignment angle between the projected stellar angular momentum vector of the galaxy and the optical position angle, defined as the orientation of the minor axis of the projected light ellipse. The first and second columns correspond to synthetic DESI-LS and Euclid images, respectively, with optical PAs estimated at one effective radius (top row) and at two effective radii (middle row). The third column shows the distributions of $\Delta \mathrm{PA^{\mathrm{gal-kin}}}$, defined as the angle between the projected spin vector of the galaxy and the kinematic PA derived from synthetic MaNGA-like stellar velocity fields. In all panels, red hatched histograms indicate results when only "good cases" (i.e., more reliable PA measurements) are selected. We also specify the number of "sources" or images used to derive the different histograms.
• Figure 5: Histogram of $\Delta \mathrm{PA^{\mathrm{opt-kin}}}$, the angles between optical and kinematic PAs derived from our sample of 5,000 synthetic DESI-LS r-band and MaNGA velocity fields. 72.8% and 82.4% of the sample have optical-kinematic misalignment lower than 10$^\circ$ and 20$^\circ$, respectively. Optical PAs are estimated at two effective radii.