On the origin of M31's 0.6–0.8 keV disk emission: a variable-W1 annulus decomposition with XMM-Newton MOS关于 M31 0.6–0.8 keV 圆盘发射起源的研究：基于 XMM-Newton MOS、变 W1 环带分解

Authors:作者： <作者列表——由 PI 决定>

Abstract摘要

M31's diffuse X-ray emission is dominated by a cool (\(\sim\)0.2 keV) thermal component, but when its temperature is left free the fit converges to kT \(\approx\) 0.74 keV. Is this hot component real M31 emission, a Milky Way foreground, or a fitting degeneracy? We use 33 XMM-Newton EPIC MOS ObsIDs from the run6 campaign (490 complete source/QPB/RMF/ARF task bundles) in variable-W1 elliptical annuli, fitted with a validated XSherpa template (foreground local bubble + M31CGM halo + cosmic X-ray background plus a free SrcApec component, abundances angr, band 0.4–7.2 keV). The SrcApec surface brightness rises from 8.6e-16 to 2.5e-16 erg/s/cm^2/arcmin^2 across annuli a = 16' to 88', while the observed/Ponti+26 stellar prediction ratio rises 0.38 to 1.64 (crossing unity at 60–68'). The MWhalo residual is consistent with the M31CGM baseline everywhere (max |sigma| = 1.37; a = 20' reaches \(\approx\)2.26sigma after a conservative nH correction). The halo-field 0.7 keV term is detected (q = 34.84; kT \(\le 0.804\) keV at 90 %; norm interval [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)) but its level is estimator-dependent (9.66 \(\pm\) 1.16\(\times\)10\(^{-8}\) vs 5.0 [3.56, 7.43]\(\times\)10\(^{-8}\), 2.3–2.7\(\sigma\); registered levels span \(\times\)11.8), so "one common level" is Questioned. A single Galactic 0.7 keV foreground remains the provisional reading, while a cored M31 halo fits equally well. The inner-disk hard excess survives point-spread and background-systematics tests (3.6–4.8 \(\sigma\)). We conclude that a strong Sigma_*-tracking component exists but that no 2-parameter Sigma_*-only model is statistically acceptable; the physical origin remains open pending foreground-free errors and the pre-registered outer-annulus rule.

M31 的弥散 X 射线发射由一个凉（\(\sim\)0.2 keV）热成分主导，但当温度被自由设定时拟合收敛到 kT \(\approx\) 0.74 keV。这个热成分是真实的 M31 发射、银河前景，还是拟合退化？我们使用 run6 运动的 33 个 XMM-Newton EPIC MOS ObsID（490 个完整源/QPB/RMF/ARF 任务包），在变 W1 椭圆环带上，用经过验证的 XSherpa 模板拟合（前景：局部泡 + M31CGM 晕 + 宇宙 X 射线背景，加上一个自由 SrcApec 成分，丰度表 angr，带 0.4–7.2 keV）。SrcApec 表面亮度从环带 a = 16' 到 88' 上升为 8.6e-16 到 2.5e-16 erg/s/cm^2/arcmin^2，而观测/Ponti+26 恒星预测比值从 0.38 上升到 1.64（在 60–68' 处穿过 unity）。MWhalo 残余在各处都与 M31CGM 基准一致（最大 |sigma| = 1.37；a = 20' 在保守 nH 修正后达到 \(\approx\)2.26sigma）。晕场 0.7 keV 项被检出（q = 34.84；90% 下 kT \(\le 0.804\) keV；norm 区间 [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)），但其水平依赖于估计器（9.66 \(\pm\) 1.16\(\times\)10\(^{-8}\) vs 5.0 [3.56, 7.43]\(\times\)10\(^{-8}\)，2.3–2.7\(\sigma\)；已注册水平跨度为 \(\times\)11.8），因此“一个共同水平”受质疑。单一银河系 0.7 keV 前景仍是临时性读法，而带核 M31 晕拟合同样好。内圆盘硬超额在点扩散和背景系统误差检验后仍然存在（3.6–4.8 \(\sigma\)）。我们得出结论：存在一个强烈追踪 Sigma_* 的成分，但没有任何 2 参数仅 Sigma_* 模板在统计上可接受；物理起源仍待前景自由误差和预注册的外环带规则解决。

Keywords: X-ray galaxies: diffuse; X-ray galaxies: individual (M31); methods: data analysis; techniques: imaging spectroscopy.

关键词：X 射线星系：弥散；X 射线星系：个别（M31）；方法：数据分析；技术：成像光谱学。

1. Introduction1. 引言

Diffuse soft X-ray emission in disk galaxies probes the hot interstellar medium and its connection to the circumgalactic medium. In M31 the diffuse emission is dominated by a cool (\(\sim\)0.2 keV) vapec component (Zhang et al. 2019), but a free-temperature fit converges to kT \(\approx\)0.74 keV with O \(\approx\)1.9 and Fe \(\approx\)0.9 (angr) — a fundamentally different regime whose origin is disputed. Competing explanations are: unresolved stellar emission (active binaries, cataclysmic variables, the LMXB tail) tracing stellar mass M_*; star-formation feedback tracing Sigma_SFR; and diffuse hot halo/CGM gas tracing projected radius R. Because the particle and soft-proton backgrounds dominate the EPIC MOS soft band, and because a foreground and an M31 component can trade flux with almost unchanged chi-square, the question is hard.

圆盘星系中的弥散软 X 射线发射探测热星际介质及其与星周介质的联系。在 M31 中，弥散发射由一个凉（\(\sim\)0.2 keV）vapec 成分主导（Zhang et al. 2019），但自由温度拟合收敛到 kT \(\approx\)0.74 keV，带 O \(\approx\)1.9 和 Fe \(\approx\)0.9（angr）——一个根本不同的体系，其起源有争议。竞争解释包括：未解析恒星发射（活跃双星、猫变星、LMXB 尾部）追踪恒星质量 M_*；恒星形成反馈追踪 Sigma_SFR；以及弥散热晕/CGM 气体追踪投影半径 R。由于粒子和软质子背景主导 EPIC MOS 软带，且前景与 M31 成分可以在几乎不变的 chi-square 下互换通量，这个问题很难回答。

This paper reports the run6 campaign decomposition and its current state: what is established, what is open, and what each running job will decide. It is written for readers who do not know the project; every number carries a source marker resolved in the Source map appendix.

本文报告 run6 运动的分解及其现状：什么是确立的，什么是开放的，以及每个运行中的任务将决定什么。它面向不了解项目的读者撰写；每个数字都带一个来源标记，在附录的来源图中解析。

2. Observations and data reduction2. 观测与数据约减

We used 33 XMM-Newton ObsIDs observed toward M31, reduced with ESAS/SAS (esas_playground v1.6.1) on Virgo into MOS1/MOS2 source spectra, quiescent particle background, RMF and ARF products. Point sources are excluded with edetect DET_ML > 6 (user decision 2026-07-19). The MOS 1.4–2.0 keV interval (Al K-alpha + Si K-alpha + soft protons) is included in the current campaign. The radial coordinate is a variable-W1 isophotal ellipse (q(a), PA(a); outer q=0.280, PA=37.7 deg). Nine non-overlapping 8' annuli from a=16' to a=88' are retained (a=92' excluded as unreliable; a=60' included with an elevated-rstat caveat). Fits use the no-background SrcApec contract: LB kT=0.1 keV fixed; MWhalo kT free 0.10–0.35 keV, Z=0.3; SrcApec kT free 0.5–1.0 keV, Z=1; CXB photon index Gamma=1.4 fixed; band 0.4–7.2 keV; abundances angr throughout. Sky area is applied as a Python float multiplier via scale_skyarea (never Const1D, per the XSherpa contract). Vertical errors are conditional Delta-chi^2=1 with saved nuisance parameters fixed.

我们使用了指向 M31 的 33 个 XMM-Newton ObsID，用 ESAS/SAS（esas_playground v1.6.1）在 Virgo 上约减为 MOS1/MOS2 源谱、静止粒子背景、RMF 和 ARF 产品。点源用 edetect DET_ML > 6 排除（用户决定 2026-07-19）。MOS 1.4–2.0 keV 区间（Al K-alpha + Si K-alpha + 软质子）包含在当前运动中。半径坐标是变 W1 等光度椭圆（q(a), PA(a)；外部 q=0.280, PA=37.7 度）。保留了 9 个不重叠的 8' 环带，从 a=16' 到 a=88'（a=92' 作为不可靠被排除；a=60' 带升高的 rstat 说明被纳入）。拟合使用无背景 SrcApec 契约：LB kT=0.1 keV 固定；MWhalo kT 自由 0.10–0.35 keV，Z=0.3；SrcApec kT 自由 0.5–1.0 keV，Z=1；CXB 光子指数 Gamma=1.4 固定；带 0.4–7.2 keV；丰度表全程 angr。sky area 作为 Python 浮点乘数通过 scale_skyarea 应用（绝不使用 Const1D，按 XSherpa 契约）。垂直误差是条件性 Delta-chi^2=1，nuisance 参数固定于已保存值。

3. Results3. 结果

3.1 The SrcApec profile. SrcApec kT declines from 0.90 keV at a=20' to 0.71–0.78 keV across a=36–84', with a sharp drop between a=28' and a=36'. The 0.5–2 keV surface brightness falls from 8.6e-16 to 2.5e-16 erg/s/cm^2/arcmin^2. The observed/Ponti+26 stellar-prediction ratio rises monotonically from 0.38 at a=20' to 1.64 at a=84', crossing unity at 60–68'.

3.1 SrcApec 剖面。 SrcApec kT 从 a=20' 处的 0.90 keV 下降到 a=36–84' 间的 0.71–0.78 keV，在 a=28' 和 a=36' 之间有急剧下降。0.5–2 keV 表面亮度从 8.6e-16 降到 2.5e-16 erg/s/cm^2/arcmin^2。观测/Ponti+26 恒星预测比值从 a=20' 处的 0.38 单调上升到 a=84' 处的 1.64，在 60–68' 处穿过 unity。

3.2 The MWhalo residual. The disk MWhalo norm (1.95e-6–3.99e-6 arcmin^-2) lies within the M31CGM halo-field range; with the baseline's field-to-field scatter (3.875e-16) as the per-annulus uncertainty, every annulus is consistent with the baseline (max |sigma|=1.37). A conservative nH correction raises only a=20' to \(\approx\)2.26sigma.

3.2 MWhalo 残余。 圆盘 MWhalo norm（1.95e-6–3.99e-6 arcmin^-2）位于 M31CGM 晕场范围内；用基准的场间散射（3.875e-16）作为每环带不确定性，每个环带都与基准一致（最大 |sigma|=1.37）。保守的 nH 修正仅将 a=20' 提升到 \(\approx\)2.26sigma。

3.3 Template decomposition. The E034 old_only (chi^2=11771.817, dof 7550) and young_only (12541.776) fits completed; mixed never ran. The E038 identifiability forecast finds old+C and old+FUV+C identifiable (lower-bound sigma(A_young,FUV) = 3.3% of A_old) but W4dust+mixed not identifiable (corr(A_young,C)=-1.00) because both WISE templates carry the sky pedestal (W1 off-disk 6.87 DN = 53–97% of the template; W4dust ~99% pedestal). The old-only fit gives A_old = 1.7713e-7, kT = 0.736 keV.

3.3 模板分解。 E034 old_only（chi^2=11771.817，dof 7550）和 young_only（12541.776）拟合完成；mixed 从未运行。E038 可辨识性预测认为 old+C 和 old+FUV+C 可辨识（下界 sigma(A_young,FUV) = A_old 的 3.3%），但 W4dust+mixed 不可辨识（corr(A_young,C)=-1.00），因为两个 WISE 模板都带有天空底座（W1 场外 6.87 DN = 模板的 53–97%；W4dust ~99% 底座）。old-only 拟合给出 A_old = 1.7713e-7，kT = 0.736 keV。

4. Discussion4. 讨论

The profile tracks Sigma_* strongly in both functional forms tried (linear-plus-constant chi^2/dof = 56.4/7; power law chi^2/dof = 27.0/7), but neither is statistically acceptable and the errors must grow by factors of 2.84 (linear) and 1.96 (power law) for chi^2/dof = 1 — and those are conditional errors with the foreground frozen. SrcApec is strongly degenerate with the foreground, so profile-likelihood errors with the foreground free are required (V1b). The rising obs/pred ratio argues against a purely old-population explanation: a constant M/L_W1 would predict a flat ratio, not one crossing unity at 60–68'. Lowering the outer M/L steepens the rise; flattening it would require M/L to rise by \(\times\)4.33 outward (opposite sign to the gradient caveat). The halo-field 0.7 keV question has moved: the withdrawn E033 branches are replaced by a detected component (q = 34.84; kT \(\le 0.804\) keV; norm interval [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)) whose level is estimator-dependent — so "one common level" is now Questioned rather than Provisional. A cored M31 halo of 1.1–1.3e10 solar masses inside 150 kpc fits the 20 halo fields as well as a constant, and the intercept from SrcApec alone is \(-1.32 \pm 0.10\times10^{-16}\) (−1.32\(\sigma\), R115): the pre-registered outer-annulus rule v2 is now undecided and power-limited, because the halo-field scatter is 2–4\(\times\) the whole detected halo range. The halo-field column preference is likewise no longer established: with the band below 0.5 keV removed, the reduced refits prefer the frozen \(N_\mathrm{H}=0.03\) over 0.0585 (\(\Delta\chi^2\) 3.79 and 16.73), so the preference most likely reflects sub-0.5 keV calibration or soft-sky modelling rather than the column (R126). The soft QPB scale that broke the first age-resolved attempt is now identified as the 1.4–2.0 keV Al K\(\alpha\) + Si K\(\alpha\) line band missing from the run6 continuum template, not soft protons (R127). The regression is on its third generation and is still not established: D1c fixed the QPB per file from the detector corners but its support rule removed 605/622 regions, leaving 13 regions and 0/24 primary cases passing, so it is formal only, with the full-sample row giving old \(\epsilon\) 1.99 [1.72, 2.25]e27 per formed M☉, ratio 82, dev/dof 1.74 and QPB pull 8.4, and D1d was launched at pixel level with an absolute corner→FOV QPB (R129, R130). The disk-floor link is model-dependent: 0.65–0.71 of the halo level with a soft star-formation term, 1.16–1.67 if the floor sits behind M31's own absorbing layer. The pre-registered outer-annulus rule v2 and the halo null test decide the hypothesis.

剖面在两种尝试的函数形式中都强烈追踪 Sigma_*（线性+常数 chi^2/dof = 56.4/7；幂律 chi^2/dof = 27.0/7），但两者在统计上都不可接受，且误差必须分别放大 2.84（线性）和 1.96（幂率）倍才能使 chi^2/dof = 1——而这些是前景冻结下的条件性误差。SrcApec 与前景强烈退化，所以需要前景自由的 profile-likelihood 误差（V1b）。上升的 obs/pred 比值反对纯旧人群解释：恒定的 M/L_W1 会预测平坦的比值，而不是在 60–68' 处穿过 unity 的比值。降低外层 M/L 会使上升更陡；要使其变平，M/L 必须向外上升 \(\times\)4.33（与梯度警告符号相反）。晕场 0.7 keV 问题已有进展：被撤回的 E033 分支被一个检出的成分（q = 34.84；kT \(\le 0.804\) keV；norm 区间 [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)）取代，其水平依赖于估计器——“一个共同水平”现在受质疑而非临时性。150 kpc 内 1.1–1.3e10 太阳质量的带核 M31 晕对 20 个晕场视场的拟合与常数一样好，且仅用 SrcApec 的截距为 \(-1.32 \pm 0.10\times10^{-16}\)（−1.32\(\sigma\)，R115）：预注册的外环带规则 v2 现在为未决且受功效限制，因为晕场散度是整个已检出晕范围的 2–4 倍。晕场柱密度的偏好同样不再确立：移除 0.5 keV 以下波段后，约化重拟合偏好冻结的 \(N_\mathrm{H}=0.03\) 而非 0.0585（\(\Delta\chi^2\) 3.79 与 16.73），因此该偏好最可能反映 0.5 keV 以下的校准或软天空建模，而非柱密度本身（R126）。破坏首次按年龄分辨尝试的软 QPB 尺度现已定位为 run6 连续谱模板中缺失的 1.4–2.0 keV Al K\(\alpha\) + Si K\(\alpha\) 线带，而非软质子（R127）。该回归已到第三代且仍未确立：D1c 用探测器角区逐文件固定 QPB，但其支撑规则移除了 605/622 个区域，只剩 13 个区域、24 个主用例中 0 个通过，因此只是形式结果，全样本行给出旧年龄 \(\epsilon\) 1.99 [1.72, 2.25]e27 每已形成 M☉、比值 82、dev/dof 1.74、QPB pull 8.4，于是启动了像素级支撑、绝对角区→视场 QPB 的 D1d（R129、R130）。圆盘底板的联系依赖于模型：带软恒星形成项时为晕场水平的 0.65–0.71，若底板位于 M31 自身吸收层之后则为 1.16–1.67。

5. Summary5. 总结

1. SrcApec kT = 0.71–0.90 keV across 9 retained annuli; obs/Ponti+26 rises 0.38 to 1.64, crossing unity at 60–68'.
2. SrcApec kT 在 9 个保留环带上为 0.71–0.90 keV；obs/Ponti+26 从 0.38 上升到 1.64，在 60–68' 处穿过 unity。
3. The MWhalo residual is consistent with the M31CGM baseline everywhere (max |sigma| = 1.37; a = 20' \(\approx\) 2.26sigma after conservative nH correction).
4. MWhalo 残余在各处都与 M31CGM 基准一致（最大 |sigma| = 1.37；a = 20' 在保守 nH 修正后 \(\approx\) 2.26sigma）。
5. A strong Sigma_*-tracking component exists, but no 2-parameter Sigma_*-only model is statistically acceptable; foreground-free errors are required.
6. 存在一个强烈追踪 Sigma_* 的成分，但没有任何 2 参数仅 Sigma_* 模板在统计上可接受；需要前景自由误差。
7. The halo-field 0.7 keV term is detected (q = 34.84; kT \(\le 0.804\) keV at 90 %; norm interval [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)), but its level is estimator-dependent, so "one common level" is Questioned; one Galactic foreground remains Provisional and its origin (Milky Way versus a cored M31 halo) is open.
8. 晕场 0.7 keV 项被检出（q = 34.84；90% 下 kT \(\le 0.804\) keV；norm 区间 [3.15, 5.82]\(\times\)10\(^{-8}\) arcmin\(^{-2}\)），但其水平依赖于估计器，因此“一个共同水平”受质疑；单一银河系前景仍为临时性，其起源（银河系还是带核 M31 晕）仍然开放。

Acknowledgments致谢

Based on observations obtained with XMM-Newton, an ESA science mission with instruments and contributions directly funded by ESA Member States and NASA. Funding and facility placeholders are for the PI to complete. If AI agents assisted in preparing this manuscript, that assistance is disclosed here.

基于 XMM-Newton 获得的观测，XMM-Newton 是一项 ESA 科学任务，仪器和贡献由 ESA 成员国和 NASA 直接资助。资金和设施占位符留待 PI 完成。如果 AI 代理协助了本稿的准备，此处予以披露。

Facilities and Software设施与软件

Facilities: XMM. Software: SAS/ESAS (esas_playground v1.6.1), CIAO/Sherpa 4.18, XSPEC 12.14.0k (APEC v3.0.9, angr abundances, vern cross-sections), astropy, numpy, scipy, matplotlib, SciencePlots.

设施：XMM。软件：SAS/ESAS（esas_playground v1.6.1），CIAO/Sherpa 4.18，XSPEC 12.14.0k（APEC v3.0.9，angr 丰度，vern 交叉截面），astropy，numpy，scipy，matplotlib，SciencePlots。

Data availability数据可用性

Raw data: XMM-Newton Science Archive (ObsIDs listed in Table 1). Products and code: placeholders for the PI; run6 spectra live on lang at /u/Lang3/Data/M31HotISM/m31_run6_band_flux_inputs_20260901/campaign/.

原始数据：XMM-Newton 科学档案（ObsID 列于表 1）。产品和代码：留待 PI 填充的占位符；run6 谱位于 lang 的 /u/Lang3/Data/M31HotISM/m31_run6_band_flux_inputs_20260901/campaign/。

Source map (handbook-only)来源图（仅手册）

References参考文献

• Zhang et al. 2019 — M31 RGS diffuse emission, kT=0.20–0.30 keV.Zhang et al. 2019——M31 RGS 弥散发射，kT=0.20–0.30 keV。
• Ponti et al. 2026, arXiv:2602.03941 — L_X/M_* = 1.05e28 / 6.9e27 erg/s/M_sun (0.1–10 keV, kT=0.7 keV).Ponti et al. 2026，arXiv:2602.03941——L_X/M_* = 1.05e28 / 6.9e27 erg/s/M_sun（0.1–10 keV，kT=0.7 keV）。
• Mineo et al. 2012b, MNRAS 426, 1870 — hot-ISM L_X/SFR = 8.3e38 erg/s per M_sun/yr; two-temperature (0.24, 0.71 keV).Mineo et al. 2012b，MNRAS 426, 1870——热 ISM L_X/SFR = 8.3e38 erg/s 每 M_sun/yr；两温度（0.24, 0.71 keV）。
• McMillan 2017 — Milky Way mass model (H1 prediction).McMillan 2017——银河质量模型（H1 预测）。
• Telford et al. 2020 — M/L_W1 median \(\approx\)0.60, 0.4 dex spread.Telford et al. 2020——M/L_W1 中位数 \(\approx\)0.60，0.4 dex 离散。
• Ibata et al. 2014 — PAndAS M31 stellar halo (M/L_V 1.9).Ibata et al. 2014——PAndAS M31 恒星晕（M/L_V 1.9）。

Newest source read: cognition-os/08_RUNNING.md revisions 24–49; 07_REVISION_LOG.md R071–R131; 04_DECISION_LOG.md D26–D30.

最新读取的来源：cognition-os/08_RUNNING.md 修订版 24–49；07_REVISION_LOG.md R071–R131；04_DECISION_LOG.md D26–D30。