INSIGHT-HXMT Observations of the New Black Hole Candidate MAXI J1535−571: Timing Analysis

We plot the diagrams commonly used for the study of BHTs in Figures 1 and 2. For comparison, we also show the MAXI/GSC¹⁵ and Swift/BAT¹⁶ results taken from the web sites for each instrument.

Zoom In Zoom Out Reset image size

The background-subtracted and dead-time-corrected Insight-HXMT light curves and hardness of MAXI J1535−571 are shown in Figure 1 (left panels). The LE count rate (1–12 keV) slowly rose from its initial value, reached a peak of 3212 cts s⁻¹ on MJD 58015, and then stayed stable at that level. The ME count rate (6–38 keV) increased from 355 cts s⁻¹ on MJD 58002 to 820 cts s⁻¹ on MJD 58014, and decreased abruptly to 522 cts s⁻¹ on MJD 58015, followed by several rises and falls. The HE count rate (26–100 keV) showed a decrease in the early phase, then was similar to the ME. Hardness is defined as the count rate in the 3–12 keV energy band divided by that in the 1–3 keV energy band. We found that the hardness remained the same (∼2.1) in the first several exposures around MJD ∼58002, but suddenly decreased to ∼1.5 on MJD ∼58008, and then slowly decreased to a low level. The trends of the light curves and hardness observed by Insight-HXMT, MAXI/GSC and Swift/BAT are consistent with each other.

The hardness–intensity diagram (HID) and the hardness–rms diagram (HRD) are shown in the left panel of Figure 2. Because only the rising part of the outburst was observed by Insight-HXMT, the source exhibited part of the standard q-shaped pattern. A relatively complete pattern is described by MAXI data in the right panel, with Insight-HXMT observations marked with red points. The outburst starts at the lower right of the figure, corresponding to the LHS, where the fractional rms remains at ∼26%. When the intensity increases, the source on the HID starts moving to the upper left, and the fractional rms drops to ∼15% on MJD 58008. In the corresponding PDS, strong type-C QPOs are detected (see Section 3.2), indicating that the system is in the HIMS. It is not possible to decide the precise transition position from the Insight-HXMT observations, as the instruments were switched off during that period. After several days in the HIMS, the fractional rms suddenly decreases to 1.9% on MJD 58015, and type-B QPOs (see Section 3.2) are seen in the PDS, indicating the system is in the SIMS. Then, the source moves irregularly in the HID but remains in the upper left. The fractional rms increases to 7.7%, then decreases to ∼2%.

Zoom In Zoom Out Reset image size

Figures 3–6 show results of the PDS. In Table 2, we present a summary of the results on LFQPO parameters, i.e., the centroid frequency (ν), the coherence parameter Q(=ν/Δν), and the rms of the QPOs; Δν is the FWHM of the QPO.

Zoom In Zoom Out Reset image size

Figure 3 shows the QPO evolution with time. In Figure 4, we show six representative PDSs of Insight-HXMT/ME whose corresponding positions are indicated by red arrows in Figure 3. At the beginning of the outburst (the first four exposures), the PDS shown in Figure 4(a) is very similar to those observed in other black holes during their typical LHSs (Belloni 2010), and can be fitted with two broad Lorentzian components. Later (from MJD 58008 to MJD 58014, Figures 4(b) and (c)), the PDSs show a strong type-C QPOs, sometimes with its second harmonic, and the centroid frequency of QPO decreases from 2.5 to 1.7 Hz and then increases to 3.3 Hz. During the ME count rate decline on MJD 58015 (see Figure 1), we detect a 9.98 Hz QPO with a rather low rms amplitude (5.3%) and a weak red-noise component at very low frequency, indicating that it may be type-B (Figure 4(d)). The dynamical PDS for the first 2000 s of this observation shows a rapid transition (see Figure 5). During the first ∼800 s, the PDS reveals appearances of type-B (with a QPO frequency ∼10 Hz). During the decrease phase in the light curve, no significant QPO with ME is detected. The HE data show a similar behavior, while the LE is saturated during this time. From MJD 58017 to the end of our sample (MJD 58023), the behavior of PDS is rather complex. On MJD 58017, while the ME rate increases, we detect a ∼10 Hz QPO with a high rms amplitude (∼13%) compared to the previous one (Figure 4(e)). Even though the QPO centroid frequency is different from previous type-C QPOs, the rms suggests that this QPO is type-C. After this, the QPO (∼12 Hz) becomes weaker and broader with a low-amplitude red-noise component (Figure 4(f)), suggesting a transition to a type-A QPO.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

The PDSs of the HE and LE detectors are approximately the same in evolution. In Figure 6 we present the PDSs of the three detectors for two observations, in which the shape of the PDS significantly evolves with energy.

Zoom In Zoom Out Reset image size

In order to quantitatively study the energy-dependent behavior of the QPO properties, we extract power spectra in several energy bands. To improve the statistics, we only derive the energy dependence of the type-C QPOs with high amplitude. The fractional rms and the centroid frequency of the type-C QPOs as functions of photon energy are shown in Figures 7 and 8, with the QPO frequencies and obsID marked in each panel. We consider the background contribution to the fractional rms calculation. The formula is rms = $\sqrt{P}* (S+B)/S$ (Bu et al. 2015), where S and B stand for source and background count rates, respectively, and P is the power normalized according to Miyamoto et al. (1991). In the region where LE and ME or ME and HE overlap, there is good agreement between the two detectors. In all cases, the rms increases with photon energy up to ∼20 keV, from ∼5% in the lowest energy band up to ∼13% during the HIMS, and from ∼1% to ∼15% during the SIMS, and stays more or less constant afterwards, while no significant decrease is seen above 30 keV. The QPO centroid frequencies are also related to photon energy. Unlike the rms, it does not have a unified trend. In four panels of Figure 8 (the top two of each column), with increasing photon energy, the frequencies first increase and then decrease after ∼10 keV. In the bottom two panels of the first column, the frequency is almost constant and independent of photon energy. However, it shows a monotonically increasing trend with photon energy in the remaining panels.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

Phase lags between soft and hard variabilities are computed from the LE data. Due to the statistics limit, only the first period is selected for further study. Figure 9 shows the phase lags as a function of frequency in the HIMS. Due to poor statistics, lags became hard to measure at high frequencies; thus we plot them only below 16 Hz. During the HIMS, the lags of the fundamental QPO are negative, while the harmonic nearly always show positive phase lags. We also derive phase lags at the QPO centroid frequency and its harmonic (shown in Figure 10). The lags of the fundamental QPO are strongly correlated with centroid frequency, with a trend toward zero when the QPO frequency decreases.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

In this work, we have presented timing results of a new BHC MAXI J1535−571 during its outburst in 2017 using Insight-HXMT data. The outburst evolution is consistent with the scenario typically observed in BHCs (Belloni et al. 2005; Belloni 2010; Muñoz-Darias et al. 2011). Based on the combined timing and color properties, we have identified three main states according to the classification criteria given by Belloni (2010). The source experienced a state transition from the LHS to the HIMS in the early phase, and then to the SIMS.

Figure 2 shows typical hardness and timing properties of the canonical LHS, although the vertical branch of the HID is completely observed. The PDS is dominated by strong band-limited noise (see panel (a) in Figure 3), with typical rms values of ∼26%. From observation P011453500119 (MJD 58008), the hardness ratio shows a significant decline from ∼2.1 to ∼1.5 before the source enters the top left in the HID (see Figure 2). During this period, the PDS (Figure 3 panel (b) and (c)) shows band-limited noise and a strong typc-C QPO with a lower total rms than in the LHS. The results indicate that the source enters the HIMS. Around MJD 58015 the timing variabilities show a clear difference from those in the HIMS, while a type-B QPO appears in the PDS (Figure 3 panel (d) and Figure 5), indicating a transition to the SIMS.

The above state transitions are also consistent with the spectral fit of the Swift observations given by Tao et al. (2018). As shown in their paper, the power-law photon index Γ stayed around ∼1.5 until MJD 58007, and increased suddenly to ∼2.0 until MJD 58014. From MJD 58015, Γ increased from ∼2.0 to ∼2.5. The inner disk temperature and the disk flux ratio stabilized at low values before MJD 58015, then jumped to a high value thereafter.

LFQPOs, consisting of three types (type-A, -B, and -C), are observed in the range 1.78–13.88 Hz (see Table 2).

Type-C QPOs are observed in the HIMS, similar to XTE J1859+226 (Casella et al. 2004), and in the early stages of the SIMS, with a higher centroid frequency, by all three detectors. When type-C QPOs are observed in the SIMS, a hard flaring occurs, suggesting an association with the hard component. Their frequencies are correlated with count rates and hardness, similar to those observed in other BHTs (Tomsick & Kaaret 2001; Belloni et al. 2005). The QPO frequencies observed by Insight-HXMT are consistent with the NICER results, which gave a QPO frequency between 1.9 and 2.8 Hz during September 12, 10:53:39 and September 13, 22:40:40 (Gendreau et al. 2017). In our case, the QPO frequency is between 1.78 and 2.74 Hz during September 12, 10:38:59 and September 14, 08:06:59.

The second harmonics of type-C QPOs are constantly detected in LE and HE observations, but only in some of the ME observations, which might due be to the low signal-to-noise ratio (see Figures 4 and 6). For observation P011453500301, the second harmonic is clearly detected in the ME energy band, thus we can measure the fractional rms as a function of photon energy for both the QPO and its second harmonic (see Figure 11). The relation of the rms of the second harmonic QPOs with photon energy has been observed in XTE J1550−564 (Li et al. 2013a) and GRS 1915+105 (Yadav et al. 2016). However, while it displays an arch-like relation whose maximum amplitude appears at ∼7 keV in XTE J1550−564, the rms of the harmonic QPO increases up to ∼10 keV and then seems to decline until ∼30 keV, with a large uncertainty, in GRS 1915+105. Using frequency-resolved spectroscopy, Axelsson & Done (2016) found that the second harmonic spectrum is dramatically softer than the QPO spectrum and the time-averaged spectrum, and can be described by an additional soft Comptonization component. The reason for the lack of a second harmonic in the ME observations may be physical. However, beyond ∼30 keV the fractional rms of the second harmonic increases with photon energy, suggesting that the second harmonic may be related to an additional component (i.e., the reflection component).

Zoom In Zoom Out Reset image size

Type-B QPOs are usually detected when a source experiences a rapid transition to the SIMS. Fast transitions have been observed in GS 1124-68 (Takizawa et al. 1997), XTE J1859+226 (Casella et al. 2004), and GX 339-4 (Belloni et al. 2005). A very sharp threshold in count rate was observed, suggesting a transition. However, for MAXI J1535−571 the QPO has a frequency around ∼10 Hz, which is different from the typical frequency of ∼6 Hz. A correlation of type-B QPO frequencies with the power-law flux has been reported by Motta et al. (2011) and Gao et al. (2017). The higher frequencies of type-B QPOs could indicate that MAXI J1535−571 has a higher hard luminosity compared to other systems. Jet ejections are thought to be associated with type-B QPOs and the X-ray flux peak (Fender et al. 2009). In MAXI J1535−571, a type-B QPO is found in correspondence with the count rate peak (see Figure 1). Future multi-wavelength observation are needed to verify the existence of relativistic jet emissions during the X-ray flux peak.

Type-A QPOs are observed in the SIMS, with a clear QPO peak at around 10 Hz present only in the ME and HE observations. Similar behavior has been reported in GX 339-4 (Belloni et al. 2005).

For the first time we have studied the fractional rms and the centroid frequency of the QPO as a function of photon energy up to 100 keV (see Figures 7 and 8).

The QPO rms amplitude increases with photon energy up to ∼20 keV and stays more or less constant in all observations. The background estimation we applied is based on the blind FoV detectors. The background consists of the cosmic X-ray and particle background including cosmic rays, albedo radiation, and South Atlantic Anomaly (SAA) induced background for a low-Earth orbit satellite (Xie et al. 2015). Since there is no sign of other bright sources in the MAXI images,¹⁷ most of the LE detector background comes from the cosmic X-ray background (dozens counts s⁻¹), which can be neglected compared to the high count rate in MAXI J1535−571. However, for ME and HE, the background is dominated by the particle background, which is related to the position and attitude of the satellite. The HE and ME background typically accounts for ∼10% to ∼20% for sub-energy bands, except for the highest sub-energy band of the HE detector which can be around 50%. In order to investigate the accuracy of our background estimation, we applied several blank sky observations, and found that the count rate ratio between the small FoV and the blind FoV detectors is independent of time. Our background estimation method is thus reasonable for rms calculation.

In addition to MAXI J1535−571, similar energy dependence relations for type-C QPOs were found in GRS 1915+015 (Rodriguez et al. 2004; Yan et al. 2012, 2013; Yadav et al. 2016), H1743−322 (Li et al. 2013b), XTE J1859+226 (Casella et al. 2004), and XTE J1550−564 (Li et al. 2013a), in which a corona origin of type-C QPOs is considered. For GRS 1915+105, HEXTE results showed that the QPO rms decreases above 20 keV (Tomsick & Kaaret 2001). However, Rodriguez et al. (2004) found that this cut-off was not always present, but rather related to compact jets, which contribute to the hard X-ray component mostly through synchrotron emission. You et al. (2018) computed the fractional rms spectrum of the QPO in the context of the Lense–Thirring precession model (Ingram et al. 2009). They found that the rms at higher energy E > 10 keV becomes flat when the system is viewed with large inclination angle. Our result is consistent with the simulation.

The correlation between the centroid frequency of QPOs and the photon energy shows three different shapes: flat, positive, and "arch"-like. For energies <20 keV, this relation in GRS 1915+105 (Qu et al. 2010; Yan et al. 2012, 2018) and XTE J1550−564 (Li et al. 2013a) evolves from a negative correlation to a positive one when the QPO frequency increases, but with a different turnover QPO frequency. The pattern in H1743−322 shows no apparent turnover frequency, which might be due to the lack of observational data for the hard state (Li et al. 2013b). The energy dependence of the QPO frequency could be caused by differential precession of the inner accretion flow (van den Eijnden et al. 2016). The inner-part flow causes a higher QPO frequency than the outer-part flow, and the evolution of the spectral properties of the inner and outer parts can cause the frequency–energy relation to change from negative to positive. When the inner-part flow has a harder spectrum than the outer-part flow, this causes a positive correlation. In MAXI J1535−571, the turnover of the relation at high energy, E > 10 keV, would suggest that this is due to the reflection bump being prominent at those energies. The reflected spectrum is expected to be dominated by photons emitted by the outer-part flow, thus the reflected spectrum will show a relatively low precession frequency.

We have calculated the phase lag between the 1–3 and 3–7 keV energy bands, and found that the phase lags of the fundamental and the harmonic of type-C QPOs remain opposite: the lags of the fundamental peak are soft, while those of the harmonic are hard. As found in GRS 1915+105 (Lin et al. 2000; Reig et al. 2000; Qu et al. 2010), and XTE J1859+226 (Casella et al. 2004), the lag is strongly correlated with the centroid frequency of the QPO, and decreases with increasing frequency.

Recently, from the inclination dependence of phase lags in a sample of 15 black hole binaries, van den Eijnden et al. (2017) found that the phase lags of type-C QPOs strongly depend on inclination, in evolution with both QPO frequency and sign. All samples possess a slightly hard lag at low QPO frequencies. At high frequencies, high-inclination sources turn to soft lags while lags in low-inclination sources become harder. These results support the geometrical origin of type-C QPOs.

MAXI J1535−571 clearly follows the trend of high-inclination sources presented in van den Eijnden et al. (2017). Xu et al. (2018) performed a spectral analysis of NuSTAR observations in the hard state, and found that the energy spectra can be well fitted by two different models which both consist of a multi-temperature thermal component, but with different reflection models (one for relxilllpCp+xillverCp, the other for relxillCp+xillverCp). They found that the inclination angles are ${57}_{-2}^{+1}$° and ${75}_{-4}^{+2}$°, respectively. The spectral fitting result from NICER suggested a similar inclination of 67.4(8)° (Miller et al. 2018). Both are consistent with our phase lag result.