Catalog of Wide X-ray Energy Band Monitoring Sources in the 6-yr Galactic Plane Scanning Survey of Insight-HXMT

LIAO Jin-yuan¹†, GUAN Ju¹, WANG Chen¹,²
(1 Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049)
(2 China Center of Advanced Science and Technology, Beijing 100190)

Abstract

We review the six-year Galactic plane scanning survey of the Insight-HXMT (Insight Hard X-ray Modulation Telescope), focusing on the monitoring and analysis of known X-ray sources on the Galactic plane. During its first six years of orbital operation, Insight-HXMT spent approximately one-quarter to one-third of its total observation time conducting over 3000 scanning observations of the Galactic plane in the broad energy range of 1–100 keV. Long-term flux monitoring was performed for more than 1300 different types of X-ray sources (detecting X-ray signals from approximately 200 celestial bodies), and the monitoring results were compiled and analyzed, including the activity and spectral characteristics of different source types. This paper first introduces the data characteristics and analysis methods of the Insight-HXMT scanning observations (direct demodulation imaging and light curve fitting), then provides an overall description of the monitoring results from the Insight-HXMT scanning survey, and finally presents a statistical analysis of the properties of the monitored sources, such as spatial distribution characteristics, variability activity analysis, and hardness ratio analysis.

Keywords: space vehicles: instruments, methods: data analysis, X-rays: survey

1. Introduction

The Insight Hard X-ray Modulation Telescope (Insight-HXMT) is China's first general-purpose space astronomy satellite, launched on June 15, 2017 [1]. The Galactic plane scanning survey represents one of Insight-HXMT's primary scientific missions, accounting for approximately one-quarter to one-third of the total observation time, with the goal of searching for new transient phenomena and monitoring known variable sources across a broad X-ray energy range. Based on previous surveys by international X-ray telescopes such as ROSAT (Roentgen Satellite), INTEGRAL (International Gamma-Ray Astrophysics Laboratory), Swift, and MAXI (Monitor of All-sky X-ray Image), most hard X-ray sources in the Galactic plane are variable, primarily consisting of various types of X-ray binaries [2–3]. Insight-HXMT's narrow field-of-view design and large effective area across its detection energy bands provide advantages for surveying weak signals and variable sources.

Insight-HXMT carries three main payloads: the Low Energy X-ray Telescope (LE, 0.7–13 keV), the Medium Energy X-ray Telescope (ME, 5–40 keV), and the High Energy X-ray Telescope (HE, 20–250 keV). LE consists of three detector boxes, each containing eight detector modules with a total of 96 SCD (Swept Charge Device) detectors [4]. ME comprises three detector boxes with Si-PIN detector arrays [5]. HE consists of 18 NaI(Tl)/CsI(Na) scintillator detectors [6]. All are collimated telescopes composed of three groups of small field-of-view detectors with 60° angles between them [1]. Each small field-of-view contains one LE box, one ME box, and five HE detectors. Additionally, the large field-of-view detectors serve as supplements for scientific data analysis and background estimation [7]. Due to their relatively narrow field of view, the small field-of-view detectors can obtain more accurate source fluxes and positions during scanning observations compared to the large field-of-view detectors, so only the small field-of-view detectors are used in the Galactic plane scanning survey [8].

During its six-year Galactic plane scanning survey, Insight-HXMT achieved comprehensive and deep coverage of the Galactic plane. [FIGURE:1] shows the six-year cumulative exposure sky map and sensitivity sky map for the low-energy telescope. The optimal cumulative sensitivities over six years across the entire Galactic plane for the three telescopes are (LE, 2–6 keV), (ME, 7–40 keV), and (HE, 25–100 keV). Insight-HXMT performed long-term flux monitoring for more than 1300 different types of X-ray sources, detecting X-ray signals from approximately 200 celestial bodies and obtaining their long-term light curves across multiple energy bands between 1–100 keV. We conducted statistical analyses of the activity and spectral characteristics of different types of celestial objects based on this long-term monitoring data [9–10]. This paper reviews the Insight-HXMT Galactic plane scanning survey, focusing on the broad X-ray energy band monitoring source catalog. Section 2 describes the data and analysis methods. Section 3 presents the broad X-ray energy band monitoring source catalog. Source catalog analysis is introduced in Section 4. Finally, Section 5 provides a summary.

2. Scanning Data and Analysis Methods

In the Insight-HXMT Galactic plane scanning survey, the entire Galactic plane (Galactic longitude range: , Galactic latitude range: ) is divided into dozens of regions with identical radii. To achieve complete coverage of the Galactic plane without gaps, overlap exists between adjacent regions. Due to Sun angle constraints ( ), each region is observable for approximately half a year. For each scanning region, observations are conducted using a row-by-row scanning pattern as shown in [FIGURE:2].

Three scanning speeds ( , , ) and ten scanning intervals ( ) are available for scanning observations. Two scanning region radii ( ) are used. The duration of a single scanning observation is typically 2–3 hours, depending on the scanning parameters. For scanning observations with , the duration is approximately 3.3 hours.

[FIGURE:3] illustrates the data analysis processing pipeline for Insight-HXMT satellite data. Following the steps shown in [FIGURE:3], we use the Insight-HXMT data analysis software HXMTsoft to process the raw data, including basic processing, generating grade files, EHK files, GTI files, screening files, light curve generation, background subtraction, and further GTI selection to obtain background-subtracted net light curve data. Details of the data processing can be found in the works of Sai et al. [8] and Wang et al. [9–10].

[FIGURE:4] shows a net light curve from a single LE scan, where complex modulation signals can be observed. When the telescope's field of view sweeps across an X-ray source, the source leaves a modulation signal because the point spread function (PSF) of a collimated telescope is a function of position. Various methods can process these modulation signals, each with different advantages, but the accuracy of all results depends on the precision of PSF calibration. Due to collimator deformation caused by the satellite platform and detector design, as well as various factors during satellite launch and in-orbit operation, the PSF of each Insight-HXMT telescope cannot be simply determined by the collimator's geometric parameters.

Therefore, we calibrate the average PSF of each Insight-HXMT payload collimator annually. The basic approach involves adding rotational and two-dimensional effective area corrections to the geometric model of the collimator PSF. The geometric model of the collimator PSF, i.e., the detection efficiency at position ( ) in the field of view, can be expressed as:

$$
P(\theta;\phi) = C \cdot \left[1 - \frac{|\tan\theta|}{\tan\theta_0}\right] \cdot \left[1 - \frac{|\tan\phi|}{\tan\phi_0}\right] \cdot \sqrt{\tan^2\theta + \tan^2\phi + 1}; \quad (1)
$$

where and are the ranges of the long and short sides of the field of view, respectively, and is the detection efficiency at the center of the field of view ( ). Details of the PSF for each Insight-HXMT detector and the detailed correction process can be found in the work of Nang et al. [11].

In our work, we primarily employ two methods to analyze Insight-HXMT scanning data: direct modulation demodulation imaging and light curve PSF fitting. The direct demodulation method can image the scanned sky area with very intuitive results, while light curve fitting provides accurate flux and error information. The combination of these two methods ensures the accuracy and reliability of Insight-HXMT scanning data analysis results. Both methods are described below.

2.1 Direct Modulation Demodulation Imaging

The direct demodulation method is a high-resolution, high-precision imaging method proposed by Chinese scholars Li Tibei and Wu Mei in the 1990s [12–13]. It iteratively solves the observation equation system describing the observation process directly and introduces physical constraints during iteration to avoid pathological oscillations caused by incomplete observation data and low signal-to-noise ratios. Consequently, the iteration results achieve both high spatial resolution from direct demodulation and good convergence with genuine physical meaning. This method has been successfully applied to image reconstruction for various X-ray astronomy satellites, demonstrating strong imaging and spectral decomposition capabilities.

The direct demodulation method is one of the primary reconstruction algorithms used for Insight-HXMT Galactic plane scanning data. Applying it to process Galactic plane scanning data yields position and intensity information for numerous celestial sources, enabling flux monitoring and transient source discovery. [FIGURE:5] shows the reconstruction results for the Galactic Center region using the direct demodulation method, with positioning accuracy better than 8′ [1]. Since the direct demodulation method applies to image reconstruction of stable sources, this condition is generally satisfied—for instance, many variable sources in the Galactic plane have outburst timescales of weeks to months, which can be considered stable during a single Insight-HXMT Galactic plane scan lasting approximately 3 hours. Therefore, the direct demodulation method is suitable for processing most Galactic plane scanning data.

However, the Galactic plane contains variable sources with shorter timescales, such as Type II bursts on timescales of seconds, which cannot be treated as stable sources. Consequently, we developed a time-dependent direct demodulation method [14]. By modifying the imaging equation to incorporate short-timescale structures and their responses, we obtained a time-dependent imaging equation and derived its iterative solution formula through mathematical derivation. The improved direct demodulation method can provide accurate light curves for short-timescale variable sources within the scanning exposure time and improve positioning accuracy for variable sources.

Using the time-dependent direct demodulation method to reconstruct the Type II bursts discovered by Insight-HXMT in the Rapid Burster MXB 1730–335 (burst duration s, burst interval s), the method successfully reproduced the bursts from MXB 1730–335. [FIGURE:6] (left panel) shows its detailed flux monitoring curve. The burst frequency is twice per minute, consistent with the relaxation oscillator phenomenon mentioned in the literature, where [FIGURE:6] (right panel) demonstrates the positive correlation between the fluence of each burst peak and the time to the next peak.

The time-dependent direct demodulation method has significant application prospects in detecting short-timescale variable signals, particularly providing strong technical support for Insight-HXMT's detection of Supergiant Fast X-ray Transients (SFXTs). These transients have typical outburst timescales of only a few hours, with light curves characterized by fast rise and slow decay, and flux variations reaching times the quiescent level. Based on Monte Carlo simulations, we simulated an Insight-HXMT observation of an SFXT and then used the time-dependent direct demodulation method to reconstruct the simulated data. The reconstruction results shown in [FIGURE:7] clearly reproduce the fast-rise, slow-decay outburst characteristics of SFXTs, with reconstructed intensity parameters matching the input values and improved positioning accuracy—in this simulation example, the positioning accuracy is 0.6′. Currently, only about a dozen SFXTs have been discovered, and their outburst mechanisms remain unclear. Approximately three models exist to explain these rare flares: clumpy stellar winds, eccentric orbital motion of companion stars, and transitions in accretion mechanisms [15–16]. If Insight-HXMT's Galactic plane scanning can discover more SFXTs, it can provide observational clues and test evidence for theoretical models.

2.2 Light Curve PSF Fitting

In Insight-HXMT scanning data analysis, we employ PSF models to directly fit light curves from different fields of view to obtain source positions and flux information, with the overall process shown in [FIGURE:8]. During PSF fitting, we first identify sources covered by the field of view based on effective area, then fit their flux information. To improve fitting efficiency and accuracy, we first classify these sources by flux strength, defining bright sources as those with fluxes above the satellite sensitivity threshold (producing noticeable contributions to the light curve) and faint sources otherwise. We then perform joint fitting of all bright sources to obtain flux information for each bright source. After completing bright source fitting, we use residuals to search for new source candidates—this step aims to identify potentially existing but unconfirmed sources. Finally, we adopt a single PSF model fitting strategy for all faint sources entering the field of view to avoid coupling and excessive computational resource consumption.

Following this procedure, after performing PSF fitting on each scanning dataset, we conduct comprehensive analysis of the results, including checking the fitting quality, determining whether outbursts of known sources occurred, and identifying any new source candidates. [FIGURE:9] shows an example of Insight-HXMT scanning data analysis. The three blue light curves in FIGURE:9 represent the background-subtracted light curves from three boxes during good time intervals. The red curve represents the final fitting result. To further assess fitting quality, gray lines below each blue curve show the corresponding residuals, illustrating differences between the fitting results and observational data. FIGURE:9 shows the corresponding satellite scanning trajectory, where purple lines represent the satellite trajectory during good time intervals, blue scatter points indicate faint sources passed by the scanning field of view during this observation, and red five-pointed stars represent bright sources in this observation, each marked with a serial number.

3. Overview of Survey Results

During its six-year Galactic plane scanning survey, Insight-HXMT conducted over 3000 scanning observations, obtaining a large amount of scanning data. Following the processing pipeline described in Section 2, each scanning dataset undergoes direct demodulation imaging and light curve fitting to obtain source positions and flux information. Insight-HXMT monitored 1336, 957, and 935 known sources in 1–6 keV, 7–40 keV, and 25–100 keV, respectively. We compiled the monitoring results into a monitoring source catalog, which mainly includes the following information for each energy band: (1) source name, (2) source coordinates, (3) source type, (4) source flux and error, (5) variability amplitude and error, and (6) hardness ratio and error. Among all monitored sources, 223 sources were detected with signal-to-noise ratio S/N > 5 in 2–100 keV, and 33 sources showed significant signals in all three payloads. [FIGURE:10] shows the positions of Insight-HXMT monitoring sources in Galactic coordinates. By statistically analyzing the relationship between signal-to-noise ratio and flux for all monitoring sources in each scanning observation, we can estimate the limiting sensitivity of a single Insight-HXMT scan; systematic errors of the scans can be estimated through long-term flux variations of the Crab [8, 11]. Insight-HXMT's detection results in different energy bands and detector properties are listed in [TABLE:1].

4. Basic Analysis of the Monitoring Source Catalog

This section primarily presents multi-dimensional statistical displays of the monitoring results for 223 bright sources in the Insight-HXMT monitoring source catalog, based on the statistical analysis work of Wang et al. on the first four years of Insight-HXMT scanning results [9–10]. The analysis focuses on the activity levels, active characteristics, spectral features, and spatial distributions of different types of celestial objects.

4.1 Variability Amplitude of Bright Sources in the Monitoring Catalog

Long-term monitoring light curves of various X-ray sources in different energy bands represent one of Insight-HXMT's important scientific products. Different types of sources exhibit different activity levels in long-term light curves, which we quantify using to measure their variability amplitude, with the formula:

$$
F_{\rm rms} = \sqrt{S^2 - \langle\sigma^2\rangle} / \langle f\rangle; \quad (2)
$$

where is the mean of the long-term flux, represents the mean of the squared errors of the long-term flux, denotes the best-fit flux of the source in the ith scan, is the number of scanning monitoring data points, and is the variability amplitude. A larger value indicates greater flux variation during the monitoring period. The error of can be calculated as:

$$
dF_{\rm rms} = \sqrt{\left(\sqrt{2/N} \cdot \langle\sigma^2\rangle / \langle f\rangle^2\right)^2 + \left(\sqrt{\langle\sigma^2\rangle/N} \cdot 2F_{\rm rms} / \langle f\rangle\right)^2}. \quad (4)
$$

When a source's is less than or equal to the Crab's in the corresponding energy band, the source is defined as flux-stable in that energy band; otherwise, it is a variable source. [FIGURE:11] shows the distribution histogram of for 223 sources, with the horizontal and vertical axes representing values and source counts, respectively, and each subplot representing a different energy band. The median of each source type (marked on each subplot) is used to indicate the overall trend. The figure shows that High-Mass X-ray Binaries (HMXBs) are more active in flux than Low-Mass X-ray Binaries (LMXBs) in any energy band, and Black Hole Binaries (BHBs) show greater variability than Neutron Star Binaries (NSBs) in 2–6 keV and 7–40 keV. Additionally, LMXBs, NSBs, and BHBs show a decreasing trend in with increasing energy band. Supernova Remnants (SNRs), isolated pulsars, and Seyfert 1 galaxies have more stable fluxes than X-ray binaries, although some sources in these three categories also have relatively large , possibly due to their low average flux.

4.2 Characteristics of Different Types of Binary Systems in the Monitoring Catalog

We analyzed the flux variations of 32 X-ray binaries detected in all three detectors. [FIGURE:12] shows the distribution of for these sources across three energy bands. As shown in FIGURE:12, the values of 15 Neutron Star Low-Mass X-ray Binaries (NS-LMXBs) in 2–6 keV are lower than those in 7–40 keV or 25–100 keV. FIGURE:12 shows that the median of LMXBs gradually increases with energy band, consistent with the study by Mitsuda et al. (1984) [18]: the spectrum of NS-LMXBs consists of multi-temperature blackbody radiation from optically thick accretion disks and blackbody radiation from the neutron star surface, with the former being stable and the latter showing active variations.

However, HMXBs exhibit the following variability characteristics: (1) Compared with LMXBs across three energy bands, HMXBs show more active flux variations in all three bands; (2) NS-LMXBs have lower than BH-HMXB in each energy band, and BH LMXBs have lower than NS-HMXB; (3) NS-HXMBs show a trend of first increasing then decreasing with increasing energy band. These variations may be related to the selected energy bands but could also be associated with the accretion processes in HMXBs. Based on companion star differences, HMXBs can be divided into three subclasses: (a) Be/X-ray binaries; (b) supergiant X-ray binaries; (c) supergiant fast X-ray transients. Since different subclasses have different companion star accretion mechanisms, this may lead to more diverse outburst patterns.

4.3 Hardness Ratio and Spatial Distribution of Bright Sources with Different Activity Levels

We calculated and analyzed the hardness ratios (HR) of bright sources in the catalog under different combinations of high and low energy bands. HR is defined as the ratio of flux in high and low energy bands and is often used to reflect the spectral characteristics of X-ray sources since different radiation mechanisms produce photons in different energy bands. In this work, we first divided bright sources into flux-stable ( ) and flux-variable ( ) groups based on flux stability, then further divided them into spectrum-stable ( ) and spectrum-variable ( ) groups based on spectral stability. Thus, all bright sources are ultimately classified into three activity levels: , , and sources. [TABLE:2] shows the number of sources with different activity levels under various high-low energy band combinations. When judging flux stability, if a source's 1- lower limit is greater than the Crab's , the flux is considered active; otherwise, it is stable. For spectral stability, we compare with the long-term HR stability of the Crab using a standard chi-square test. A source is considered spectrally active in the target energy band if its chi-square value exceeds the limit corresponding to the relevant degrees of freedom; otherwise, it is spectrally stable. This limit is obtained from the chi-square of the Crab's long-term HR in the corresponding high-low energy bands.

[FIGURE:13] shows the long-term light curves and HR of Crab, Cen X–3, and Cyg X–1. The Crab's long-term HR has 78 degrees of freedom and a chi-square of 77.3, corresponding to a significance level of 0.5, which is therefore used to determine the flux activity level of other bright sources in these two energy bands. Cen X-3 and Cyg X-1 have 64 and 146 degrees of freedom, respectively. The chi-square values corresponding to a 0.5 significance level for these degrees of freedom are 63.3 and 145.3, while the actual long-term HR chi-square values are 38.3 and 17808.6. Clearly, Cen X-3 is a source, while Cyg X-1 is a source. Using this method, we calculated the spectral and flux activity levels of bright sources in various energy band combinations, with the numbers of sources for each activity level shown in [TABLE:3].

Bright sources with different activity levels show different spatial distributions. [FIGURE:14] shows the distribution of sources along Galactic longitude and latitude. Each subplot has upper and lower parts: the upper part shows the overall distribution of all sources, while the lower shaded area highlights the concentration range containing 68% of sources for each activity level (marked by upper and lower ticks). The width represents the span in Galactic longitude or latitude, and the height corresponds to the number of sources, with specific values marked in the figure. Different activity levels show distinct clustering trends: sources mainly concentrate within the range of , sources concentrate in the region of , while sources have a more dispersed distribution.

4.4 Overall Hardness Ratio Distribution of X-ray Binary Systems

During the Insight-HXMT Galactic plane scanning survey, the bright sources monitored are predominantly X-ray binary (XRB) systems. We therefore first analyzed the overall characteristics of XRB HR. In XRBs, more detailed classification into many subclasses is possible based on the mass and evolutionary stage of the primary or companion star. In this section, we first study the overall HR characteristics of LMXBs and HMXBs. We fix the low-energy band at 2–4 keV and sequentially select 4–6, 5–7, 7–40, and 25–100 keV as high-energy bands, calculating HR for these high-low energy band combinations and using the median HR to characterize overall trends. As shown in [TABLE:3], most binary systems have active flux.

[FIGURE:15] shows the relationship between HR and source counts for these sources, with subplots (a)–(d) representing all sources, , , and sources, respectively. The low-energy band is fixed at 2–4 keV, with each color corresponding to different high-energy bands (as shown in the legend). Vertical dashed lines indicate median positions. By comparing the positions of these median lines, we see that in each subplot, the HR distribution of HMXBs is always shifted rightward compared to LMXBs, indicating that HMXBs have harder spectra. Comprehensive analysis of [FIGURE:15] reveals:

1. In all high-low energy band combinations, the overall HR of HMXBs is consistently higher than that of LMXBs. This trend is evident in all sources, , and sources. Since the number of flux-stable XRBs is relatively small, with some energy bands having none or only one source, they are difficult to include in the comparison.
2. In LMXBs, sources have overall harder spectra than sources, while in HMXBs this trend is opposite below 7 keV. The overall higher HR of HMXBs compared to LMXBs means HMXBs have harder spectra, consistent with the conclusion of Fabbiano [19]. We attribute this to the modulation effect of different magnetic field strengths on accreted material: (1) In accreting neutron star systems, HMXBs are generally younger than LMXBs, and their relatively short accretion times allow them to retain more primordial magnetic fields, so NS-HMXBs generally have higher surface magnetic field strengths than LMXBs; (2) In strong magnetic field environments, charged particles accreted from the companion star experience strong modulation and are dragged to the magnetic poles to form accretion columns. These columns contain extreme environments with high temperature, high pressure, and strong magnetic fields, where various thermal and non-thermal radiation processes occur; (3) In weak magnetic field accreting binary systems, magnetic field modulation decreases, making it difficult to support the formation of accretion columns at magnetic poles. In this case, accreted material is pulled into a circular orbit, forming a Keplerian disk, and the resulting X-ray radiation is mainly composed of thermal radiation from the accretion disk and neutron star surface [20].

In summary, in accreting neutron star systems, the spectra of HMXBs are generally harder than those of LMXBs. It should be noted that some XRBs in both LMXBs and HMXBs are BHB systems, but since BHBs constitute a small proportion of binary systems [21], they have little impact on the overall distribution trend and are therefore ignored here.

4.5 Hardness Ratio Distribution of Different Types of Bright Sources in 2–7 keV

Among the bright sources detected in 2–7 keV, in addition to many known types, there are many currently unclassified sources. We analyzed the distribution of hardness ratios HR1 (5–7 keV/3–5 keV) and HR2 (4–6 keV/2–4 keV) for bright sources. In addition to studying the HR characteristics of known source types, we also attempted to explore possible types for unclassified sources. There are 142 bright sources in the target energy band, but seven source types have too small sample sizes for statistical analysis and were therefore excluded. These seven types are: cataclysmic variables (IGR J18308–1232, SS Cyg), Seyfert 2 galaxies (IGR J16024–6107), radio sources (AX J1841.3–0455), binary or multiple star systems (AX J1847.6–0156), galaxy clusters (Oph Cluster), and Be stars (gam Cas). The remaining 135 sources include the following types: HMXB, LMXB, SNR, Pulsar, and Seyfert 1 galaxies. [FIGURE:16]–17 show the distribution of HR1 and HR2 for these sources, with different colors representing different source types.

In [FIGURE:17], each subplot represents one source type, with dashed lines indicating the median HR position and specific values shown in each subplot. Using medians to characterize overall trends, we find that HMXBs have the largest HR1 and HR2 among these types. We calculated the correlation between HR1 and HR2 for different source types, obtaining correlation coefficients of 0.62 (LMXB), 0.93 (HMXB), 0.66 (Pulsar), 0.69 (SNR), 0.48 (Seyfert 1), and 0.59 (unclassified). This suggests that the correlation between HR1 and HR2 may differ among source types. Notably, the correlation coefficient does not have much physical meaning here but can be used to roughly estimate which type unclassified sources resemble. Based on correlation coefficients and histograms, we find:

1. The HR1 of LMXBs is relatively uniformly distributed in the range of 0.2 to 0.6, while HR2 shows a trend of concentrating near 0.4.
2. HMXBs have the widest HR distribution among these types and the hardest overall spectra.
3. The HR distribution range of unclassified sources is relatively close to that of LMXBs, with a correlation coefficient closest to LMXBs. The overall distribution trend also shows a shift from relatively uniform to clustering around a certain value. Therefore, based purely on distribution, we can speculate that LMXBs may constitute a larger proportion among unclassified sources.

[FIGURE:18]–19 show the HR1 and HR2 distributions for binary systems. In these binary systems, most primary stars have clear classifications, but some have uncertain classifications (marked as '?' in the figures). From the HR distribution of binary systems with known primary star types, we can see that NS-HMXBs have significantly higher HR than NS-LMXBs, which in turn have higher HR than all BHBs. In BHBs, the primary star is believed to lack a solid surface [22], so in the quiescent state, radiation is dominated by thermal components from the disk, making their spectra softer than NSBs. When entering the outburst state, they exhibit rich spectral evolution, and when non-thermal radiation from jets, coronae, and other regions dominates, the spectrum becomes very hard. BHBs spend a relatively small fraction of time in outburst, thus showing relatively soft spectral characteristics overall. Additionally, we can use [FIGURE:18]–19 to speculate on possible types for several XRBs with uncertain primary star classifications. First, the HR of 'LMXB/?' is located at the peak of the NS-LMXB distribution, larger than the median HR of BH-LMXB. Second, the HR of 'HMXB/?' is significantly higher than BH-HMXB, while in FIGURE:19 several 'HMXB/?' sources have HR near BH-HMXB. Therefore, we can speculate that the source with uncertain primary star type in LMXBs is more likely to be NS-LMXB, while in HMXBs, although some sources may be BH systems cannot be ruled out, NS-HMXB may constitute a larger proportion.

4.6 Hardness Ratio of Transient XRBs Between Low-Flux and Outburst States

We analyzed the HR of seven transient sources that showed dramatic outbursts during the first four years of monitoring, comparing their HR in outburst and low-flux states. Here, a source is considered to have entered an outburst state when the 3 lower limit of the flux in a single scan exceeds the detector sensitivity; otherwise, it is considered to be in a low-flux state. [TABLE:4] lists basic information for these seven sources (serial number, name, position, type, average HR in outburst and low-flux states). [FIGURE:20] shows the distribution of average HR in these two states, with the middle dashed line representing equal average hardness ratios between low-flux and outburst states. Error bars for each data point represent the errors of the mean HR in both states; HR errors in the low-flux state are larger due to lower flux.

From [TABLE:4] and [FIGURE:20], we can observe:
1. The three black hole binary systems GX 339–4, MAXI J1820+070, and MAXI J1348–630 have larger lower limits of average HR in the low-flux state than in the outburst state. This is consistent with the q-shaped track in the HID diagram commonly used to describe state transitions in BHBs.
2. In both low-flux and outburst states, the HR of the two NS-HMXBs SWIFT J0243.6+6124 and 1A 0535+262 is higher than the remaining five binary systems, further confirming the conclusion in the previous subsection.

5. Summary

As China's first general-purpose space X-ray telescope, Insight-HXMT has been operating smoothly in orbit for over eight years. The Galactic plane scanning survey, as one of its three core missions, occupies one-quarter to one-third of the total observation time, achieving cumulative sensitivities of , , and in 2–6 keV (LE), 7–40 keV (ME), and 25–100 keV (HE), respectively. Insight-HXMT performed long-term flux monitoring for more than 1300 known X-ray sources on the Galactic plane, detecting X-ray signals from over 200 celestial bodies and constructing a broad-band monitoring source catalog.

Based on analysis of these bright sources' flux variations, we studied the relationship between and source type, finding:
1. Most supernova remnants have stable flux. Most isolated pulsars and Seyfert 1 galaxies show small or negligible flux variations.
2. HMXBs are more active in flux than LMXBs in any energy band, and BHBs show greater variability than NSBs in 2–6 keV and 7–40 keV.
3. Most NS-LMXBs show smaller flux variation amplitudes in 2–6 keV than in higher energy bands, possibly related to the spectral components of NS-LMXBs.
4. The variation of HMXBs across different energy bands is more complex, possibly related to more diverse accretion processes in different types of HMXBs.

By classifying all bright sources into three categories ( , , and sources) based on flux and spectral activity, we analyzed the HR distribution characteristics of these three groups. Comparing HR distributions of different source types below 7 keV, we found that LMXBs may constitute a larger proportion among unclassified sources, while NSXBs may be more prevalent among binary systems with uncertain primary star classifications. Studying the spatial distribution characteristics of the three activity groups revealed clear clustering trends: sources mainly concentrate within , sources concentrate in the region of , and sources are more dispersed—this is a noteworthy phenomenon.

With more domestic and international space X-ray telescopes coming online, Insight-HXMT's scanning observations will be adaptively adjusted, with plans for joint Galactic plane scanning observations with the EP (Einstein Probe) satellite to fully leverage both advantages. With its broad detection energy band, high energy resolution, and high time resolution, Insight-HXMT can conduct in-depth spectral and timing studies of sources. The EP satellite has the largest instantaneous field of view (3600 ) in the soft X-ray band and sub-arcminute positioning accuracy for weak sources, providing significant advantages in detecting faint variable sources. Through joint scanning observations by Insight-HXMT and EP, complementary advantages are expected to yield new results in the field of unknown short-timescale variable sources.

Acknowledgments

This work utilizes data from the Insight-HXMT mission, funded by the China National Space Administration (CNSA) and the Chinese Academy of Sciences (CAS).

References

Cao X, Jiang W, Meng B, et al. SCPMA, 2020, 63: 249502
Liu C, Zhang Y, Li X, et al. SCPMA, 2020, 63: 249503
Liao J Y, Zhang S, Chen Y, et al. JHEAp, 2020, 27: 24
Sai N, Liao J Y, Li C K, et al. JHEAp, 2020, 26: 1
Wang C, Liao J Y, Guan J, et al. ApJS, 2023, 265: 523
Wang C, Liao J Y, Guan J, et al. RAA, 2024, 24: 015
Nang Y, Liao J Y, Sai N, et al. JHEAp, 2020, 25: 39
Li T P, Wu M. Ap&SS, 1993, 206: 91
Li T P, Wu M. Ap&SS, 1994, 215: 213
Guan J, Lu F J, Zhang S, et al. JHEAp, 2020, 26: 11
Chen Y, Li T P, Wu M. A&AS, 1998, 128: 363
Lu F J, Aschenbach B, Song L M. A&A, 2001, 370: 570
Vaughan S, Edelson R, Warwick R S, et al. MNRAS, 2003, 345: 1271
Mitsuda K, Inoue H, Koyama K, et al. PASJ, 1984, 36: 741
Zhang S N. SCPMA, 2020, 63: 249502
Baumgartner W H, Tueller J, Markwardt C B, et al. ApJS, 2013, 207: 19
Fabbiano G. ARA&A, 2006, 44: 323
Alpar M A, Cheng A F, Ruderman M A, et al. Nature, 1982, 300: 728
Bird A J, Bazzano A, Malizia A, et al. ApJS, 2016, 223: 15
Belczynski K, Ziolkowski J. ApJ, 2009, 707: 870
Chen Y, Cui W, Li W, et al. SCPMA, 2020, 63: 249505
Zhang S N. FrPhy, 2013, 8: 630