IGR J17091-3624: Microquasar Candidate

• IGR J17091-3624 is a Galactic X-ray transient and black hole candidate known for its microquasar characteristics and GRS 1915+105-like heartbeat oscillations.
• Its observations reveal a transition from standard hard-to-soft states to exotic timing behaviors, including multiple variability classes and high-frequency quasi-periodic oscillations.
• Multiwavelength studies, precise localization, and correlated radio/optical data make it a vital laboratory for probing accretion dynamics and jet phenomena in black hole systems.

Searching arXiv for recent and foundational papers on IGR J17091-3624. IGR J17091-3624 is a Galactic X-ray transient and black hole candidate that emerged as one of the most intensively studied analogues of GRS 1915+105 after its 2011 outburst revealed highly structured, quasi-periodic X-ray variability, including “heartbeat” oscillations, multiple variability classes, and high-frequency quasi-periodic oscillations (Altamirano et al., 2011). Its observational importance derives from the coexistence of canonical black hole transient behavior—hard-state evolution, state transitions, compact-jet phenomenology, and quiescent emission—with exotic timing and spectral states that are rare among black hole X-ray binaries (Rodriguez et al., 2011). Across subsequent work, the source has been localized with sub-arcsecond precision, associated securely with optical/IR and radio counterparts, monitored in quiescence and outburst, used as a laboratory for nonlinear time-series analysis, and revisited in later outbursts that further expanded its phenomenology (Bodaghee et al., 2012).

1. Discovery, classification, and source identity

IGR J17091-3624 is treated in the literature as a black hole candidate, a transient X-ray binary, and, after precise multiwavelength localization, a microquasar candidate (Bodaghee et al., 2012). It was discovered by INTEGRAL in 2003, while archival activity was later identified in 1994, 1996, and 2001, and subsequent outbursts were recorded in 2007, 2011, 2016, 2022, 2025, and again in 2025 in later summaries of the source’s history (Capitanio et al., 2013). A recurrent theme in the literature is that pre-2011 outbursts were comparatively standard for a transient black hole candidate, whereas the 2011 event revealed behavior previously associated mainly with GRS 1915+105 (Capitanio et al., 2012).

A decisive step in establishing the source’s identity came from a short \textit{Chandra}/HRC-I observation obtained during the 2011 outburst. That observation localized the X-ray source at

$\mathrm{R.A.\ (J2000)} = 17^{\mathrm h}\ 09^{\mathrm m}\ 07\fs59,$

$\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$

with an adopted positional uncertainty of

$0\farcs6 \quad \text{at 90\% confidence},$

described as the \textit{Chandra} boresight uncertainty for an on-axis source (Bodaghee et al., 2012). This HRC position is compatible with both the previously proposed optical/IR candidate and the compact radio counterpart, each lying within $0\farcs4$ of the X-ray position (Bodaghee et al., 2012). The paper states that this localization “solidif[ies]” or “cement[s]” the source’s status as a microquasar by tying together the radio, IR/optical, and X-ray detections (Bodaghee et al., 2012).

The same work identified the HRC counterpart as CXOU J170907.6-362425, with a background-subtracted HRC-I count rate of 37.050 counts s$^{-1}$ in 0.3–10 keV and detection significance 776, confirming that the localization was obtained during a bright active phase rather than in quiescence (Bodaghee et al., 2012). This secure astrometric association is central because much of the later interpretation of IGR J17091-3624—as a black hole candidate with jet activity, extreme variability, and optical accessibility relative to GRS 1915+105—depends on confident cross-band source identification (Bodaghee et al., 2012).

2. Outburst history and long-term evolution

The source’s documented outburst history includes events in 1994, 1996, 2001, 2003, 2007, 2011, 2016, 2022, and 2025, with the 2011 outburst serving as the canonical reference for the source’s exotic phenomenology (Capitanio et al., 2013). Before 2011, the source was described as only moderately bright, at roughly 10–20 mCrab in 20–100 keV, and its pre-2011 evolution was characterized as consistent with the standard spectral and temporal evolution expected for a transient black hole candidate (Capitanio et al., 2013). The 2011 event was different: it was the brightest ever observed from the source in that study, reaching about 100 mCrab in 20–40 keV and later developing pseudo-periodic flare-like events resembling those of GRS 1915+105 (Capitanio et al., 2013).

Long-term RXTE monitoring of the 2011–2013 outburst showed that the source followed a canonical early path—hard state, transition, softer states—before entering a prolonged interval of exotic variability (Court et al., 2017). In that atlas, \textit{Swift}/XRT first detected the 2011 outburst on MJD 55595; \textit{Swift}/BAT recorded a rise from $\sim 9$ mCrab to $\sim 110$ mCrab in 15–50 keV between MJD 55584 and 55608; and heartbeat-like flaring emerged after a 10 mHz QPO appeared on MJD 55634 (Court et al., 2017). The same study found that the 2011–2013 outburst lasted $\lesssim 952$ days and later showed three re-flares centered approximately at MJDs 56100, 56220, and 56375 (Court et al., 2017).

The 2022 outburst demonstrated that the source can again connect canonical and exotic behavior within a single event. NICER and NuSTAR data showed that the outburst began in the hard state, then passed through HIMS and SIMS, before entering an “exotic soft state” in which the source alternated among a relatively quiet soft state, Class V heartbeat-like variability, and a new Class X (Wang et al., 2024). This suggests that IGR J17091-3624 is not merely a static analogue of GRS 1915+105, but a system capable of moving between conventional and exotic accretion regimes within one outburst cycle (Wang et al., 2024).

A later 2025 hard-state-only outburst added a distinct new phenomenon: quasi-periodic X-ray dipping with a recurrence period of $2.83\pm0.07$ days, interpreted as recurrent obscuration by ionized material in the outer accretion disk (Lin et al., 24 Aug 2025). This further broadened the source’s phenomenology beyond the 2011 heartbeat framework.

3. X-ray variability classes and heartbeat phenomenology

The 2011 outburst established IGR J17091-3624 as the second black hole system after GRS 1915+105 to show a rich zoo of structured variability classes (Altamirano et al., 2011). RXTE observations over the first 180 days of the outburst showed at least the $\nu$, $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$0, $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$1, $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$2, $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$3, and $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$4 classes, as well as quiet intervals resembling $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$5, all occurring at 2–60 keV count rate levels that can be 10–50 times lower than in GRS 1915+105 (Altamirano et al., 2011). The source therefore reproduced the phenomenology of GRS 1915+105 at much lower observed count rates, often on faster timescales, and with important differences such as reversed loop direction in some hardness–intensity trajectories (Altamirano et al., 2011).

The heartbeat or $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$6-class behavior was especially prominent. These quasi-periodic flares can recur as fast as every few seconds and as slowly as $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$7 s, with fractional rms amplitudes ranging from $\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$8 to 50–60% (Altamirano et al., 2011). In hardness–intensity space, folded heartbeat cycles trace loops resembling those in GRS 1915+105, but IGR J17091-3624 traverses them anti-clockwise in the representative case emphasized by Altamirano et al., whereas GRS 1915+105 traverses them clockwise (Altamirano et al., 2011). This suggests that the detailed spectral evolution through the cycle differs even when the gross phenomenology is similar.

Subsequent work refined the phenomenology into a nine-class atlas for the 2011–2013 outburst, with Classes I–IX, of which seven had closest counterparts in GRS 1915+105 and two did not (Court et al., 2017). Class IV was the clearest heartbeat-like class, with regular oscillations every 25–39 s and flare peak count rates of 159–241 cts s$\mathrm{Dec.\ (J2000)} = -36^\circ\ 24^\prime\ 25\farcs4,$9 PCU$0\farcs6 \quad \text{at 90\% confidence},$0, whereas Class VIII combined very rapid flaring with infrequent deep dips and showed a coherent QPO with $0\farcs6 \quad \text{at 90\% confidence},$1 at a separation of $0\farcs6 \quad \text{at 90\% confidence},$2 s (Court et al., 2017). The same atlas emphasized that Classes VII and VIII have no clear GRS 1915+105 counterpart, showing that the source is not simply a faint reproduction of that system (Court et al., 2017).

Two additional variability classes, C1 and C2, were reported from the 2011 RXTE data (Pahari et al., 2012). C1 superficially resembles the $0\farcs6 \quad \text{at 90\% confidence},$3 class but evolves substantially within less than 200 s, including a lag reversal from a $0\farcs6 \quad \text{at 90\% confidence},$4 s soft lag to a $0\farcs6 \quad \text{at 90\% confidence},$5 s hard lag, while C2 alternates between variable and non-variable substates at similar mean colors, implying that a soft state is not required for strong variability to occur (Pahari et al., 2012). These classes expanded the phenomenological basis for interpreting accretion-instability behavior in IGR J17091-3624 beyond the original GRS 1915+105 class taxonomy.

The 2022 outburst further extended the classification scheme by identifying a new Class X, a nearly sinusoidal high-rms heartbeat-like mode with a narrow peak at

$0\farcs6 \quad \text{at 90\% confidence},$6

corresponding to a recurrence time of about 65 s, and often accompanied by a QPO between 2 and 3 Hz (Wang et al., 2024). This newly established class reinforces the source’s role as a bridge between ordinary transient black hole binaries and the more persistent exotic behavior associated with GRS 1915+105 (Wang et al., 2024).

4. Fast timing, QPOs, and nonlinear interpretations

IGR J17091-3624 exhibits timing phenomena across a wide frequency range. During the early hard-state-like phase of the 2011 outburst, RXTE data showed broad-band noise and a QPO drifting from 0.1 Hz up to $0\farcs6 \quad \text{at 90\% confidence},$7 Hz, behavior described as resembling that commonly seen in black hole transients in the hard state (Altamirano et al., 2011). After the onset of structured variability, power spectra during $0\farcs6 \quad \text{at 90\% confidence},$8-class intervals showed low-frequency QPOs with strong harmonic content, as well as a 6–10 Hz QPO and sometimes a 1–5 Hz bump (Altamirano et al., 2011).

A major timing result was the discovery of high-frequency quasi-periodic oscillations. RXTE/PCA data revealed a strong HFQPO at $0\farcs6 \quad \text{at 90\% confidence},$9 Hz with significance $0\farcs4$0, quality factor $0\farcs4$1, and fractional rms $0\farcs4$2 in the 2–25 keV band (Altamirano et al., 2012). In a subset of data after MJD 55841, the feature remained near $0\farcs4$3 Hz with $0\farcs4$4 and rms $0\farcs4$5, while a second, marginal feature appeared at $0\farcs4$6 Hz with significance $0\farcs4$7, $0\farcs4$8, and rms $0\farcs4$9 (Altamirano et al., 2012). The 66 Hz QPO has a hard spectrum, with rms increasing from about 5% at 3 keV to about 14% at 13 keV (Altamirano et al., 2012).

The near-$^{-1}$0 ratio of the two HFQPOs motivated an interpretive Letter proposing that the pair may reflect period doubling and nonlinear resonance, by analogy with pulsating stars (Rebusco et al., 2012). In that reading, the 66 Hz feature would be the fundamental $^{-1}$1, and the 164 Hz feature would be approximately $^{-1}$2, since $^{-1}$3 for $^{-1}$4 Hz is 165 Hz (Rebusco et al., 2012). The relevant resonance condition quoted there is

$^{-1}$5

and the authors suggested that, if the same underlying mode is compared with GRS 1915+105 and spin is similar, the inverse-frequency mass scaling

$^{-1}$6

would imply a black hole mass of about $^{-1}$7 for IGR J17091-3624 (Rebusco et al., 2012). The same paper explicitly stressed that this interpretation was provisional rather than definitive (Rebusco et al., 2012).

Nonlinear time-series analyses have produced a more nuanced picture. A 2020 correlation-integral study concluded that the source is mostly consistent with stochastic behavior and that high Poisson noise may suppress any intrinsic nonlinear character (Adegoke et al., 2020). In that work, the ratio $^{-1}$8 exceeded $^{-1}$9 in all nine RXTE classes, reaching values such as 0.79, 0.75, and 0.73 in Classes I–III, and even the classes with the lowest values—VII and VIII—were at 0.32 and 0.33 at 0.125 s binning (Adegoke et al., 2020). By contrast, denoising-based studies in 2024 argued that several classes show signs of determinism once Poisson contamination is mitigated, with classes V, VII, and VIII emerging as predominantly non-stochastic in a majority-consensus framework (Guria et al., 2024). This suggests that the earlier stochastic classification may have been influenced by low count rates rather than reflecting purely stochastic underlying dynamics (Guria et al., 2024).

5. Spectral states, luminosity, and the faintness problem

The spectral evolution of IGR J17091-3624 combines standard black hole transient states with later departures from canonical behavior. In the early 2011 outburst, simultaneous \textit{Swift}/XRT, INTEGRAL, and RXTE observations showed a hard state with $\sim 9$0 and $\sim 9$1 keV in one representative epoch, along with type-C QPOs at $\sim 9$2 Hz and $\sim 9$3 Hz in the first two RXTE observations (Rodriguez et al., 2011). By the third epoch, the spectrum required a disk component with $\sim 9$4 keV and $\sim 9$5, and the source was classified as being in the soft intermediate state (Rodriguez et al., 2011).

Longer monitoring of the 2011 outburst found that the source initially followed the standard hard-to-soft transition but then ceased to trace the canonical q-track. Instead, after the hard-to-soft transition around MJD 55614, it remained trapped in the top-left corner of the hardness–intensity diagram while developing heartbeat-like flaring (Capitanio et al., 2012). In the bright soft phase, the spectra were typically described by

$\sim 9$6

with $\sim 9$7–1.3 keV and $\sim 9$8–2.6 when the high-energy tail was constrained (Capitanio et al., 2012).

One of the source’s central interpretive problems is its low observed flux relative to the GRS 1915+105-like variability it displays. During the 2011 outburst, the HRC-I count rate of 37 counts s$\sim 9$9 was converted, using an absorbed power law with $\sim 110$0 and $\sim 110$1, to an absorbed 2–10 keV flux of

$\sim 110$2

which the authors compared to a \textit{Swift}/BAT peak-equivalent 2–10 keV intensity of $\sim 110$3 (Bodaghee et al., 2012). From this they derived an observed luminosity

$\sim 110$4

which directly fed into debates over whether the source might be unusually distant or host an unusually low-mass black hole if its behavior is associated with near-Eddington accretion (Bodaghee et al., 2012).

A 2012 study using phase-resolved spectroscopy of the heartbeat oscillations combined this faintness problem with disk modeling and argued for a high-inclination system (Rao et al., 2012). Using the DISKPN normalization

$\sim 110$5

with $\sim 110$6, the best-fit normalization implied $\sim 110$7, while the 90% upper limit implied a more conservative lower bound

$\sim 110$8

under the assumptions adopted there (Rao et al., 2012). Combining these fits with the assumption that heartbeat peaks correspond to roughly $\sim 110$9–$\lesssim 952$0, that paper inferred $\lesssim 952$1, $\lesssim 952$2 kpc, and very low or retrograde spin as a possible explanation for the source’s low observed luminosity (Rao et al., 2012). By contrast, later atlas-based work argued from empirical state-transition luminosities that the source likely accreted at $\lesssim 952$3 of Eddington during the 2011–2013 outburst, implying that Eddington-limited accretion is neither necessary nor sufficient for GRS 1915+105-like variability (Court et al., 2017). The coexistence of these positions remains one of the source’s major interpretive tensions.

The quiescent luminosity also became part of this debate. XMM-Newton detections in 2006 and 2007 measured 0.5–10 keV luminosities of $\lesssim 952$4–$\lesssim 952$5 for an assumed distance of 10 kpc, rising to $\lesssim 952$6–$\lesssim 952$7 at 20 kpc and $\lesssim 952$8–$\lesssim 952$9 at 35 kpc (Wijnands et al., 2012). Those values are unusually high for quiescent black hole transients and were interpreted either as evidence for a very long orbital period or as signs that the source had been observed in a sub-luminous low-level accretion state rather than true quiescence (Wijnands et al., 2012).

6. Multiwavelength environment: jet, wind, absorber, optical counterpart, and later developments

The source’s jet phenomenology was established through coordinated radio and X-ray campaigns. During the 2011 outburst, ATCA detected the radio counterpart in all four observations; in the hard state the radio spectrum was flat, with $2.83\pm0.07$0 and $2.83\pm0.07$1, consistent with optically thick synchrotron emission from a self-absorbed compact jet, while after the state transition the radio spectrum steepened to $2.83\pm0.07$2, interpreted as optically thin synchrotron from a discrete ejection event (Rodriguez et al., 2011). By the fourth epoch, the radio emission had faded strongly, indicating radio quenching after the state transition (Rodriguez et al., 2011). The same work argued that the source’s position on the radio versus X-ray luminosity diagram is compatible with that of other black hole sources for distances greater than 11 kpc, estimating a distance between $2.83\pm0.07$3 and $2.83\pm0.07$4 kpc for a typical mass of $2.83\pm0.07$5 (Rodriguez et al., 2011).

Wind and absorber behavior has been more complex. During the 2016 outburst, simultaneous XMM-Newton and ATCA data showed the source in a transition from a hard state to a hard-intermediate state, with compact jet emission present in all three radio observations and an intrinsic ionized static absorber securely identified in Obs3 (Gatuzz et al., 2019). That absorber had

$2.83\pm0.07$6

and was traced mainly by Ne X, Mg XII, Si XIII, and Fe XVIII (Gatuzz et al., 2019). Its simultaneous presence with a compact jet was highlighted because it complicates the often-invoked wind–jet anti-correlation in black hole X-ray binaries (Gatuzz et al., 2019).

The 2022 outburst provided a state-dependent view of iron-line phenomenology. When exotic variability was absent—in the Hard State, HIMS, SIMS, Soft State, and IMS Return—the spectrum showed a broad iron emission line attributed to relativistic reflection (Wang et al., 2024). By contrast, in the Transition to Class V, Class V, and Class X, the iron-K band showed absorption features from highly ionized iron, with Fe XXV near $2.83\pm0.07$7–$2.83\pm0.07$8 keV and Fe XXVI near $2.83\pm0.07$9–$\nu$0 keV (Wang et al., 2024). The paper’s central conclusion was that the appearance of highly ionized absorption lines is not determined by spectral state alone, but rather by the presence of exotic variability: in a soft spectral state, absorption lines are only detected along with exotic variability (Wang et al., 2024).

Optical and infrared work has become increasingly important because IGR J17091-3624 is less obscured than GRS 1915+105. Long-term optical monitoring over the 2011, 2016, and 2022 outbursts found that the optical and X-ray fluxes are significantly correlated, following

$\nu$1

which was interpreted as suggesting that the optical emission is dominated by an X-ray-irradiated accretion disk (Saikia et al., 20 Jan 2026). That study estimated the extinction toward the source as

$\nu$2

and found that the global optical/X-ray correlation suggests a distance estimate of 8–17 kpc (Saikia et al., 20 Jan 2026). It also reported tentative evidence of optical oscillations that may arise from reprocessed X-ray modulations, though the authors stressed that confirming this will require higher time-resolution optical data (Saikia et al., 20 Jan 2026).

The 2025 outburst introduced a distinct outer-disk absorption phenomenon. IXPE, NICER, EP, NuSTAR, and Swift observations revealed intermittent X-ray dips recurring with period

$\nu$3

accompanied by spectral hardening and explainable by obscuration from an ionized absorber with

$\nu$4

and

$\nu$5

(Lin et al., 24 Aug 2025). The periodic reappearance of the absorber was interpreted as likely caused by material in the outer accretion disk, modulated by the orbital period, with the additional implication that the donor star may be a partially stripped giant if the dip period corresponds to the binary orbit (Lin et al., 24 Aug 2025).

Taken together, these observations establish IGR J17091-3624 as a system in which compact jets, ionized absorbers, relativistic reflection, optical irradiation signatures, heartbeat-like timing states, and newly discovered periodic dipping can all be studied within one source. This suggests that its scientific value lies not only in its similarity to GRS 1915+105, but in its ability to connect that extreme phenomenology to more standard black hole X-ray binary behavior across multiple accretion states (Wang et al., 2024).