Preamble

Vol. 66 No. 4

July 2025

Acta Astronomica Sinica Vol. 66 No. 4 Jul., 2025 doi: 10.15940/j.cnki.0001-5245.2025.04.003

A Review of the First Six-Year In-orbit Background of Insight-HXMT*

(Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049)

Abstract

This paper reviews the in-orbit background of the Insight-Hard X-ray Modulation Telescope (Insight-HXMT) during its first six years of operation, encompassing the geographical distributions, spectral and temporal characteristics, and long-term evolution of the background for each payload. Additionally, we review the background estimation methodologies for each Insight-HXMT payload and provide a comprehensive overview of the background models and their estimation accuracy. Overall, the in-orbit background of Insight-HXMT meets expectations, and the background models for each payload can reliably estimate both the spectral and temporal variations of the in-orbit background.

Keywords spacecraft: instruments, methods: data analysis, X-rays: background

CLC number: P111; Document code: A

1 Introduction

Insight-HXMT (Insight-Hard X-ray Modulation Telescope) is China's first general-purpose space X-ray telescope, which has achieved fruitful results in X-ray astronomy since its launch on June 15, 2017 [1]. The satellite carries three main payloads: the Low Energy telescope (LE) [2], Medium Energy telescope (ME) [3], and High Energy telescope (HE) [4]. The coordinated operation of these three payloads endows Insight-HXMT with broad energy coverage, large effective area, and high time resolution. The main performance parameters of Insight-HXMT are listed in Table 1 [TABLE:1].

Most of Insight-HXMT's scientific results derive from pointed observations of Galactic X-ray sources [5–6]. In these observations, the acquired data contain both physical signals from X-ray sources and varying proportions of background signals. Accurate background estimation is therefore crucial for obtaining reliable physical results. In addition to pointed observations of known Galactic X-ray sources, Insight-HXMT has two other primary missions: the Galactic plane scanning survey and gamma-ray burst all-sky monitoring. These tasks employ different observation strategies and consequently exhibit distinct background characteristics and estimation methods. For gamma-ray burst monitoring, background estimation is performed by averaging pre- and post-burst observations [7]; for the Galactic plane scanning survey, background signals are estimated by smoothing the data after removing the source modulation [8]. Compared with these two modes, background estimation for Insight-HXMT pointed observations is considerably more complex. As Insight-HXMT is a collimated telescope, it lacks the ability to directly obtain accurate backgrounds from images like focusing imaging telescopes such as XMM-Newton (X-ray Multi-Mirror Mission) [9–10] and Chandra X-ray Observatory [11], nor does it employ the traditional on-off observation mode used by other collimated telescopes such as BeppoSAX/PDS (Satellite per Astronomia X/Phoswich Detection System) [12–13] and RXTE/HEXTE (Rossi X-ray Timing Explorer/High Energy X-ray Timing Experiment) [14–16]. Therefore, Insight-HXMT requires background estimation methods tailored to its specific design. To enable more accurate background estimation, Insight-HXMT has developed background models based on its own characteristics [17–19].

Insight-HXMT operates in a near-Earth circular orbit at an altitude of 550 km with an inclination of 43°. Previous studies have shown [17–22] that the space environment in Insight-HXMT's orbital region is complex, with various particles interacting with the satellite platform and instruments to produce multiple background components [23–24]. Among these, cosmic-ray protons contribute most significantly to the background, while electrons, neutrons, the Cosmic X-ray Background (CXB), and Earth albedo gamma rays also make contributions. Insight-HXMT is equipped with a Particle Monitor (PM) on its top panel to detect environmental protons (energy E > 8 MeV) and electrons (energy E > 0.15 MeV) [25]. During the first six years of operation, the geographical distribution of PM count rates showed minimal variation across low-latitude regions, high-latitude regions, and the South Atlantic Anomaly (SAA). However, since 2022, enhanced solar activity has increased atmospheric density, leading to an overall decrease in PM count rates. We have observed some long-term changes in the Insight-HXMT background, which are attributable not only to the solar activity cycle but also, to a large extent, to radiation damage in the LE and ME detectors and activation effects in HE from charged particle bombardment. Therefore, when reviewing Insight-HXMT's background, we must examine both the evolution of its observational characteristics and the validity of the background models.

This paper provides a comprehensive review of Insight-HXMT's background during its first six years in orbit. We investigate the in-orbit background of the three main payloads using high Galactic latitude blank sky observations, covering the observational characteristics of each payload's background and systematic error analysis of the background models. Table 2 [TABLE:2] presents detailed information about these blank sky observations. It is noteworthy that the three telescopes have different field-of-view (FoV) sizes and orientations. Figure 1 [FIGURE:1] and Table 1 illustrate the orientations and parameters of these fields of view. For LE and ME pointed observations, the use of small FoV detectors is recommended for scientific analysis. Consequently, this paper focuses on the observational characteristics and background models for these detectors. The paper is organized as follows: Sections 2–4 describe the backgrounds of LE, ME, and HE, respectively, while summary and conclusions are presented in Section 5.

2 Low Energy X-ray Telescope

LE consists of a series of Swept Charge Device (SCD) detectors with a geometrical area of 384 cm², covering an energy range of 0.7–13 keV. LE has three telescope boxes with 60° differences in their FoV orientations. Each box contains 20 small FoV detectors (some of which have failed sequentially), 6 large FoV detectors, and 2 fully blocked detectors (with collimators sealed by aluminum caps). The blocked small FoV detectors are designed to measure particle background, while the blocked large FoV detectors are implanted with ⁵⁵Fe radioactive isotopes to monitor energy response. During Insight-HXMT's first six years of operation, some LE detectors have failed and been turned off. Detailed information about LE's failed detectors can be obtained from the "bad detector FITS files" included in the Insight-HXMT data analysis software. Figure 2 [FIGURE:2] shows a light curve from a blank sky observation, displaying the typical profile of LE background and its characteristic features. The entire time range can be divided into anomalous and normal instrument phases. During anomalous phases, LE is typically disturbed by large numbers of low-energy charged particles and visible light due to its relatively large field of view, which enter through the collimator and are difficult to estimate accurately. In severe cases, LE detectors saturate due to on-board storage overflow. The normal instrument phase can be categorized into three types. First are Earth occultation intervals, during which light curves from detectors with different fields of view coincide and no CXB photons are recorded. Second are "count rate burst" periods, which can be detected by both small and large FoV detectors, with burst flux being essentially proportional to FoV size. Finally, intervals with neither Earth occultation nor count rate bursts are considered Good Time Intervals (GTIs), which are typically used for scientific analysis. To accurately estimate background, the background analysis software must perform not only routine GTI selection but also compare count rates between small and large FoV detectors [17]. Below we present the observational characteristics of LE background and the validity of its background model during the first six years.

2.1 Observational Characteristics and Long-term Evolution of LE Background

Figure 3 [FIGURE:3] compares the geographical distributions of LE background before and after June 30, 2020. The background distribution across geographic longitude and latitude remains unchanged, but its intensity has increased significantly. Figure 4 [FIGURE:4] (left panel) shows background spectra from small FoV detectors at the same geographic location. The spectral shape shows minimal variation at low energies but exhibits more pronounced evolution at high energies. During the first four years, the high-energy background level gradually increased due to instrument radiation damage (details below). From the fifth year onward, the high-energy background level decreased due to reduced charged particle levels in the satellite's orbit. Various spectral lines in the background spectrum become increasingly less prominent and eventually difficult to visually identify because of their low equivalent widths and continuous broadening due to degraded LE energy resolution. Details about background spectral lines (e.g., energies, broadening) and detector characteristics can be found in works on Insight-HXMT's in-orbit operation and calibration [26]. As demonstrated by ground simulations [22] and previous in-orbit observations, LE background can be simplified into diffuse X-ray background dominating at low energies and particle background dominating at high energies. Therefore, the differences between the two geographical distributions shown in Figure 3 are primarily due to high-energy variations. The right panel of Figure 4 shows background spectra from blocked detectors over the six-year period, which are consistent with results from small FoV detectors. As described by Zhang et al. [22], Insight-HXMT background can be produced by various incident particles. Background recorded immediately after incidence is called prompt background, while background recorded long after incidence (hours to months) is called delayed background. Notably, backgrounds caused by both CXB and cosmic-ray protons are prompt background. LE background light curves show minimal variation during the first six years, with the most obvious characteristic being stable count rates at low energies and significant modulation by Earth's magnetic field at high energies, as also shown in Figure 2.

The spectral shape of LE blocked detector background does not change with geographic location and can be used to characterize the particle background spectral shape of small FoV detectors [17]. The LE background model exploits this characteristic to provide simple and reliable background estimation. Although LE background evolved during the first six years, the changes were not extremely significant (Figure 3). Figure 4 shows that small and blocked FoV detectors exhibit similar evolutionary trends: as in-orbit time increases, the lower limit of the energy range becomes higher and count rates increase accordingly. For LE detectors, large signals can be recorded simultaneously in several pixels as split events. However, only events above a certain threshold are recorded and participate in subsequent split event reconstruction. For example, a large signal with energy E can be recorded as two signals with energies E₀ and E₁ (E₀ + E₁ = E). If E₁ is below the threshold, this large signal will be treated as a single event with energy E₀. As radiation damage to LE detectors increases, the noise signal distribution becomes broader. To eliminate the impact of noise signals on the operational energy range, the threshold has been adjusted higher. This raises the low-energy limit of LE detectors, as shown in the low-energy portion of the spectra in Figure 4. Moreover, small signals that could be recorded and participate in split event reconstruction before threshold adjustment will no longer exceed the threshold afterward; that is, after threshold adjustment, previously reconstructible double events will no longer be reconstructed. As the threshold increases, a larger proportion of double events will not be reconstructed and will instead be treated as single events with lower energy. As shown in Figure 4, the evolutionary trend of the background spectrum during the first four years is a leftward shift at the high-energy end each year. Therefore, the increasing trend of LE high-energy background during the first four years is essentially the result of radiation damage to LE detectors. Additionally, the blocked detector spectrum at high energies may be mixed with above-threshold signal peaks, so the effective energy range for blocked detectors in the background model must be adjusted accordingly.

2.2 LE Background Model

The validity of the background model must be examined as it is crucial for scientific analysis. Following previously developed methods [17], we perform background estimation for each blank sky observation. Figure 5 [FIGURE:5] shows an example of background spectrum estimation. For each year, background model parameters are updated to maintain estimation accuracy, and systematic errors of the background model are investigated. Figure 6 [FIGURE:6] shows the systematic errors in different energy bands between 2–10 keV for each year since Insight-HXMT's launch. The results indicate that systematic errors have not changed significantly compared with the first two years after launch, demonstrating that the background model is stable and can provide accurate background estimation. However, since data around 1.5 keV are frequently affected by electronic noise and the detection threshold has been raised, this paper presents systematic errors only above 2 keV.

3 Medium Energy X-ray Telescope

As shown in Table 1, ME is a collimated telescope covering the energy range 5–40 keV, with a total geometrical area of 952 cm². It consists of 54 detectors, each containing 32 Si-PIN pixels. Each box contains 18 detectors: 15 with small FoV collimators, two with large FoV collimators, and one with a fully blocked collimator for background estimation. ME background characteristics share some similarities with LE at high energies, particularly in light curve features and geographical distribution. However, the proportions of different background components differ significantly, with particle background dominating throughout the entire energy range [19, 22].

Figure 7 [FIGURE:7] compares the geographical distributions of ME background in the first and sixth years. The background in the sixth year is slightly higher than in the first year. In regions near the SAA (330° < lon < 360°, 0° < lat < 30°), the background is significantly higher than in most other regions at similar latitudes. This indicates that when the satellite passes through high particle flux regions such as the SAA, the ME background first increases and then decreases over time, demonstrating that ME background has delayed components. By comparing geographical distributions of background in ascending (south-to-north) and descending (north-to-south) orbital phases (Figure 7), we find that ME background has relatively strong short-timescale delayed components, which contribute to the long-term evolution of ME background. To improve background estimation accuracy, background model parameters should be updated annually. ME background shows a clear anti-correlation with geomagnetic cutoff rigidity [19].

The evolution of ME background, particularly the intensity of the silver line, must be carefully treated to ensure model accuracy. Figure 8 [FIGURE:8] shows background light curves from ME small FoV detectors in six energy bands, revealing clear orbital modulation. The light curves exhibit a prominent peak caused by particle events, typically occurring at high latitudes, with corresponding times excluded from GTIs. Figures 9 FIGURE:9–(b) show variations in background spectra from small FoV detectors within the geographic region (340° < lon < 350°, 5° < lat < 15°). First, the background level in ascending phase is much higher than in descending phase, caused by short-decay-timescale delayed background components since the satellite has just exited the SAA at this geographic location. During the first five years, ME background level slowly increased with time due to cumulative effects of weak delayed components. In the sixth year, the background level decreased due to reduced charged particle levels in the satellite's orbit. The center of the silver line also shifted over time, indicating changes in the energy-channel relationship. Figures 9(c)–(d) show spectral evolution of blocked detectors. The silver line position did not shift significantly, indicating that blocked detectors experienced less radiation damage than small FoV detectors.

3.1 Observational Characteristics and Long-term Evolution of ME Background

ME background spectral shape remains nearly constant across different geomagnetic cutoff rigidity ranges, but the background level varies substantially. The background shows a clear anti-correlation with geomagnetic cutoff rigidity [19].

3.2 ME Background Model

We have previously developed the ME background model and corresponding database [19]. Since ME background spectral shape varies little with geographic location but more than that of LE, the background model must account for ME background at each geographic location for each detector. In each background estimation, we first use the database to obtain preliminary predicted background spectra for ME small FoV and blocked detectors at each geographic location the satellite passes, then use observations from blocked detectors for further correction. The ME database produces time-averaged, normalized background spectra for each geographic location, while the particle intensity at that time can be determined using blocked detectors to calibrate the background model.

For each year, backgrounds of all blank sky observations are estimated using model parameters from the corresponding year. Figure 10 [FIGURE:10] shows an example of background estimation for a blank sky observation. Statistical analysis of all background estimation residuals yields systematic errors for each energy band [19]. Figure 11 [FIGURE:11] shows systematic errors in six energy bands for each year. The results indicate no significant increasing trend in systematic errors during the first five years. Systematic errors in the 10–15 keV band are relatively large, averaging about 2%, while errors in the 10–40 keV band are about 1.6%. In the sixth year, systematic errors above 15 keV increased, but remain below 2.5% in all bands, demonstrating that the ME background model remains reliable.

4 High Energy X-ray Telescope

HE comprises 18 NaI(Tl)/CsI(Na) composite crystal detectors (numbered DetID = 0; 1; 2; :::; 17), surrounded by 18 anti-coincidence detectors (ACDs) for active background shielding. Among these 18 detectors, 15 have small FoV, two have large FoV, and one has a fully blocked FoV for background estimation. Ground simulations indicate that NaI and CsI crystals can be activated by charged particles around Insight-HXMT's orbit, and radioactive decay of activated crystals is the main source of HE background. As the satellite operates continuously in orbit, crystals in HE detectors are constantly activated. The background level increased significantly during the first year of operation and then gradually slowed [20–21].

4.1 Observational Characteristics and Long-term Evolution of HE Background

Figure 12 [FIGURE:12] shows geographical distributions of HE background in the first and sixth years. Overall distribution differences are small, but the background count rate in the sixth year is significantly higher than in the first year. Unlike LE and ME backgrounds, HE background is dominated by delayed components due to crystal activation by charged particles. Therefore, even at the same geographic location, backgrounds differ substantially between ascending and descending orbital phases. Figure 13 [FIGURE:13] shows spectra at different orbital phases and geographic locations for each year. Spectra at geographic location (lon; lat) = (345°; 15°) are shown in subpanels (c) and (d). In ascending phase, when the satellite passes through the SAA, detector crystals become severely activated. Without sufficient time for decay, the background level is relatively high. In descending phase, however, the background is lower since a long time has elapsed since the satellite last passed through the strong charged particle region, making the background dominated by long-timescale decay components. HE background spectra at different geographic locations show long-term evolution. Compared with results shown in other subpanels, the evolution shown in subpanel (d) is less significant because the satellite has just passed the SAA, so a large fraction of the background is contributed by short-timescale components. Moreover, charged particle intensity in the SAA did not change significantly over the six years. As described in previous work [20–21], HE background spectra consist of various emission lines caused by interactions between detectors and high-energy particles. Figure 13 shows that spectral shape remained stable over the six years.

Figure 14 [FIGURE:14] shows light curves of blank sky observations in six different energy bands. In each band, background intensity rises to high levels when the satellite has just passed the SAA, then gradually decays and shows significant geomagnetic modulation. Differences exist between energy bands because the background comprises many components with different proportions, spectral shapes, and characteristic timescales.

4.2 HE Background Model

Based on HE background characteristics, we have developed the HE background model [18]. Its principle is similar to ME but more complex. To obtain background at any geographic location and time, we constructed empirical functions with time as the independent variable to describe the long-term evolution of HE background. Preliminary background estimates can be obtained using orbital parameters and observation time, with further correction using data from blocked detectors. Therefore, HE background estimation heavily depends on the mathematical description of long-term background evolution, making the accuracy of empirical functions crucial for background estimation. Figure 15 [FIGURE:15] shows long-term evolution of background count rates in the 46–74 keV band at six different geographic locations, demonstrating effects of different geomagnetic cutoff rigidities and different SAA delayed backgrounds. For each energy channel, the long-term evolution can be described by a broken line with several slopes. The fitted curve is formed by merging broken lines with different break times within this energy range, showing smooth transitions without obvious jumps. Notably, we chose broken-line functions to describe the long-term evolution of background count rates with time. While other functions might be acceptable, broken lines already describe the observational data well. As predicted by ground simulations [20–21], activated isotopes cause rapid background decay after each SAA passage, with long-term accumulation as days in orbit increase. This accumulation rises rapidly during the initial period after launch and becomes slower after hundreds of days because long-half-life isotopes are not dominant. This predicted long-term evolution is consistent with observational results shown in Figure 15. Notably, HE background shows a decreasing trend in the sixth year, resulting from increased orbital atmospheric density due to enhanced solar activity.

Using the background model, we can estimate backgrounds for all blank sky observations. Figure 16 [FIGURE:16] shows an example of background estimation for a blank sky observation. Following methods introduced in previous work [18], systematic errors in different energy bands can be obtained. Figure 17 [FIGURE:17] shows systematic errors in eight energy bands for each year. The results indicate that the average systematic error each year is less than 3%, showing no significant difference from results during the satellite's first two years of operation, demonstrating that the HE background model remains effective.

5 Summary and Conclusions

After six years of in-orbit operation, the backgrounds of Insight-HXMT's three telescopes show different evolutionary trends compared with the initial operational period. The main background characteristics (temporal variation, spectral shape, and long-term evolution) are all consistent with design expectations.

LE detectors have undergone a series of operations to address radiation damage issues, which have also caused changes in LE background. For example, detection thresholds were raised to avoid noise signals, whose distribution broadens as LE radiation damage increases. Additionally, the continuous broadening of emission lines in background spectra results from degraded LE energy resolution. With increasing in-orbit operation time, ME background level increased due to accumulation of weak delayed components. Furthermore, ME background at low energies may also be affected by low-energy noise in some pixels. HE detector crystals are continuously activated, which is why their background intensity increased significantly over time. During the first five years, the HE background growth trend gradually slowed and showed saturation behavior; in the sixth year, HE background showed a decreasing trend due to increased orbital atmospheric density from enhanced solar activity. The inconsistent evolution of background at different energies means that background spectral shape at a given geographic location also evolves with time.

Although the temporal evolution characteristics of LE and ME backgrounds are not significant, background model parameters must be updated annually to maintain estimation accuracy. For the HE background model, time evolution was considered from the beginning of model construction. Statistical analysis shows that systematic errors for all three telescopes changed little during Insight-HXMT's first six years of operation, so the background models remain effective and reliable.

As described in previous work [17], the LE background model constructed using blank sky observations can effectively estimate particle background and diffuse X-ray background caused by CXB. Therefore, it can be used for pointed observations at high Galactic latitudes (|b| > 10°). To accurately estimate diffuse background in low Galactic latitude regions, the diffuse X-ray background obtained from Galactic plane scanning should be used in LE background estimation [27].

Notably, the current background models for all three telescopes rely heavily on blocked detectors. Blocked detectors are therefore crucial, especially for HE which has only one blocked detector. This poses a potential risk for background estimation due to insufficient redundancy. Consequently, an alternative background estimation method that does not depend on blocked detectors must be planned in advance, such as using ACD and PM as prompt particle monitors for LE and ME background estimation. For HE, a parameterized background model independent of blocked detectors has already been established [28]. By considering various physical factors contributing to HE background, a mathematical model accounting for these physical processes has been successfully constructed.

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

Frontera F, Costa E, dal Fiume D, et al. AA&S, 1997, 122: 357
Frontera F, Costa E, dal Fiume D, et al. SPIE, 1997, 3114: 206
Rothschild R E, Blanco P R, Gruber D E, et al. ApJ, 1998, 496: 538
García J A, McClintock J E, Steiner J F, et al. ApJ, 2014, 794: 73
García J A, Grinberg V, Steiner J F, et al. ApJ, 2016, 819: 76
Chen Y, Cui W, Li W, et al. SCPMA, 2020, 63: 249505
Cao X, Jiang W, Meng B, et al. SCPMA, 2020, 63: 249504
Liu C, Zhang Y, Li X, et al. SCPMA, 2020, 63: 249503
Zhang Y, Ge M Y, Song L M, et al. ApJ, 2019, 879: 61
Li C K, Lin L, Xiong S L, et al. NatAs, 2021, 5: 378
Luo Q, Liao J Y, Li X F, et al. JHEAp, 2020, 27: 1
Liao J Y, Zhang S, Chen Y, et al. JHEAp, 2020, 27: 24
Liao J Y, Zhang S, Lu X F, et al. JHEAp, 2020, 27: 14
Guo C C, Liao J Y, Zhang S, et al. JHEAp, 2020, 27: 40
Li G, Wu M, Zhang S, et al. ChA&A, 2009, 33: 333
Xie F, Zhang J, Song L M, et al. Ap&SS, 2015, 360: 47
Zhang J, Li X B, Ge M Y, et al. Ap&SS, 2020, 365: 158
Alcaraz J, Alvisi D, Alpat B, et al. PhLB, 2000, 472: 215
Sai N, Liao J Y, Li C K, et al. JHEAp, 2020, 26: 1
Strüder L, Briel U, Dennerl K, et al. A&A, 2001, 365: L18
Alcaraz J, Alpat B, Ambrosi G, et al. PhLB, 2000, 484: 241
Turner M J L, Abbey A, Arnaud M, et al. A&A, 2001, 365: 27
Garmire G P, Bautz M W, Ford P G, et al. SPIE, 2003, 4851: 28
Lu X, Liu C, Li X, et al. JHEAp, 2020, 26: 77
Li X B, Li X F, Tan Y, et al. JHEAp, 2020, 27: 64
Jin J, Liao J Y, Wang C, et al. ApJS, 2022, 260: 42
You Y, Liao J Y, Zhang S N, et al. ApJS, 2021, 256: 47

The First 6-year In-orbit Background of Insight-HXMT

LIAO Jin-yuan (Institute of High Energy Physics, Chinese Academy of Sciences, Beijing 100049)

ABSTRACT This paper reviews the in-orbit background of the Hard X-ray Modulation Telescope (HXMT) over the first six years, including the geographical distributions, spectral and temporal characteristics, as well as the long-term evolution of the in-orbit background of each payload. In addition, we also review the estimation methods for the in-orbit background of each payload of Insight-HXMT, providing a comprehensive introduction to the strategies for background estimation and the accuracy of the estimation. Overall, the in-orbit background of Insight-HXMT is consistent with expectations, and the background models for each payload can reliably estimate the spectrum and light curve of the in-orbit background.

Key words space vehicles: instruments, methods: data analysis, X-rays: background