IGR J17511-3057: Accreting Millisecond Pulsar

• IGR J17511-3057 is a transient neutron-star low-mass X-ray binary exhibiting accreting millisecond pulsations, type-I bursts, and hard-state spectra.
• It features recurring outbursts with coherent pulsations, thermal Comptonization-dominated emission, and distinct QPOs that challenge standard models.
• Observations reveal precise orbital and spin parameters along with evolving accretion geometry, including evidence for a transition near the propeller regime at low luminosity.

IGR J17511−3057 is a neutron-star low-mass X-ray binary discovered by INTEGRAL on 2009 September 12 and subsequently established as an accreting millisecond X-ray pulsar with a spin frequency near $244.8$ Hz, corresponding to a spin period of about $4.1$ ms. It is also one of the few systems classified as both an accreting millisecond X-ray pulsar and a nuclear-powered X-ray pulsar, because pulsations are observed in the persistent emission and during type-I X-ray bursts. On timing and spectral grounds it is further classified as an atoll source, and during the observed outbursts it has predominantly occupied a hard accretion state characterized by thermal Comptonization, coherent pulsations, thermonuclear bursts without photospheric radius expansion, and atypical high-frequency timing behavior (Paizis et al., 2012, Kalamkar et al., 2011, Sanna et al., 19 Sep 2025).

1. Source class, position, and binary parameters

The source is a transient neutron-star LMXB in which disk accretion is magnetically channeled onto the stellar surface, producing coherent accretion-powered pulsations. The most accurate X-ray position was obtained from a 20 ks Chandra HETGS observation performed on 2009 September 22. Because the zeroth-order image was affected by pile-up, the position was derived from the readout streak and grating arms, yielding

$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$

with a $90\%$ uncertainty of $0.6''$ (Paizis et al., 2012).

XMM-Newton timing during the 2009 outburst provided a precise circular-orbit solution. The measured parameters were $a\sin i/c = 0.275196(4)$ lt-s, $P_{\rm orb}=12487.51(2)$ s, $T^*=55094.9695351(7)$ MJD, and $e<8\times10^{-5}$. The corresponding mass function,

$$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$

implies a main-sequence companion with a mass between $4.1$0 and $4.1$1 (Papitto et al., 2010).

The same system parameters were refined during later outbursts. In 2015, XMM-Newton timing yielded $4.1$2 Hz, $4.1$3 lt-s, $4.1$4 s, and $4.1$5 MJD (Papitto et al., 2016). In 2025, NICER and NuSTAR timing gave $4.1$6 s, $4.1$7 light-seconds, $4.1$8 MJD/TDB, and $4.1$9 Hz, again with a very small eccentricity, $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$0 (Sanna et al., 19 Sep 2025).

2. Outburst history and long-term phenomenology

The first observed outburst began in September 2009 and lasted about a month. RXTE, Swift, XMM-Newton, Chandra, and INTEGRAL showed that the source combined the standard AMXP observables—hard Comptonized persistent emission, coherent pulsations, and type-I bursts—with unusually rich timing phenomenology (Falanga et al., 2010, Ibragimov et al., 2011). The outburst profile was fitted with an exponential decay with an e-folding time of about $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$1 days, and application of the disk instability model yielded an outer disk radius of $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$2 cm (Falanga et al., 2010).

A second outburst occurred between 2015 March 23 and April 25. Swift/XRT measured a peak $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$3–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$4 keV unabsorbed flux of $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$5 erg cm$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$6 s$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$7 on March 24, followed by an approximately month-long decay to quiescence. The outburst flux and spectral properties were reported to be remarkably similar to those observed during 2009, and the source again showed coherent pulsations and type-I bursts without photospheric radius expansion (Papitto et al., 2016).

After nearly a decade in quiescence, a third outburst began on 2025 February 11. NICER observed the source from February 11 to February 14, when the mean $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$8–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$9 keV count rate was about $90\%$0 counts s$90\%$1, and NuSTAR observed it on February 19 during the decay, when the $90\%$2–$90\%$3 keV flux had already dropped by roughly $90\%$4–$90\%$5 relative to the NICER epoch (Sanna et al., 19 Sep 2025). XMM-Newton then observed the source on 2025 March 4, about $90\%$6 days after onset, catching it in a very late and faint phase with $90\%$7 erg s$90\%$8, still above the 2019 Chandra quiescent level of $90\%$9 erg s$0.6''$0 (Illiano et al., 17 Jul 2025).

Across all three outbursts, the recurring pattern is a hard-state AMXP with persistent pulsations near $0.6''$1 Hz and repeated type-I bursting. The principal novelty of the 2025 activity is that the decline was followed to much lower luminosity than before, revealing coherent pulsations in a regime usually associated with centrifugal inhibition of accretion (Illiano et al., 17 Jul 2025).

3. Persistent spectrum, reflection, and accretion geometry

Broad-band spectroscopy during the 2009 outburst established thermal Comptonization as the dominant persistent-emission process. Using INTEGRAL, RXTE, and Swift/XRT, the average spectrum was fit with a slab-geometry Comptonization model with $0.6''$2 keV, $0.6''$3 keV, $0.6''$4, and unabsorbed bolometric flux $0.6''$5 erg cm$0.6''$6 s$0.6''$7. Using 58 RXTE/PCA+HEXTE spectra, the same study found that $0.6''$8, $0.6''$9, and $a\sin i/c = 0.275196(4)$0 remained constant within errors through the outburst decay, indicating a remarkably stable hard state (Falanga et al., 2010).

A simultaneous XMM-Newton and RXTE study during the same outburst required at least three continuum components: a cool multicolour disc blackbody, a hotter blackbody attributed to the neutron-star surface, and a hard Comptonized component. In the preferred broad-band fit the Comptonizing plasma had $a\sin i/c = 0.275196(4)$1 keV and $a\sin i/c = 0.275196(4)$2. Reflection signatures were also detected, specifically a broadened Fe K$a\sin i/c = 0.275196(4)$3 emission line near $a\sin i/c = 0.275196(4)$4 keV and a Compton hump around $a\sin i/c = 0.275196(4)$5 keV. Reflection fits implied an inclination of approximately $a\sin i/c = 0.275196(4)$6–$a\sin i/c = 0.275196(4)$7 and an inner disc radius of order $a\sin i/c = 0.275196(4)$8–$a\sin i/c = 0.275196(4)$9 km for a $P_{\rm orb}=12487.51(2)$0 neutron star (Papitto et al., 2010).

The 2009 Chandra HETGS observation sampled the source later in the outburst and again found a Comptonized persistent spectrum. The absorbed $P_{\rm orb}=12487.51(2)$1–$P_{\rm orb}=12487.51(2)$2 keV luminosity was $P_{\rm orb}=12487.51(2)$3 erg s$P_{\rm orb}=12487.51(2)$4 for an adopted distance of $P_{\rm orb}=12487.51(2)$5 kpc, and the best-fit continuum was a thermal Comptonization model with soft seed photons of temperature $P_{\rm orb}=12487.51(2)$6 keV, a hot corona, fixed $P_{\rm orb}=12487.51(2)$7 keV, and asymptotic photon index $P_{\rm orb}=12487.51(2)$8 (Paizis et al., 2012).

The 2015 outburst reinforced this picture. XMM-Newton, Swift, and INTEGRAL found a spectrum dominated by thermal Comptonization with photon index $P_{\rm orb}=12487.51(2)$9–1.8 and $T^*=55094.9695351(7)$0 keV; XMM-Newton also detected a broad Fe line at $T^*=55094.9695351(7)$1 keV with $T^*=55094.9695351(7)$2 keV and equivalent width of about $T^*=55094.9695351(7)$3 eV, consistent with K$T^*=55094.9695351(7)$4 emission from ionized iron in the inner disk (Papitto et al., 2016). A later NuSTAR analysis of the same outburst used relxill to model the reflection spectrum self-consistently and obtained $T^*=55094.9695351(7)$5, $T^*=55094.9695351(7)$6, $T^*=55094.9695351(7)$7 solar, and $T^*=55094.9695351(7)$8, corresponding to $T^*=55094.9695351(7)$9 km for a $e<8\times10^{-5}$0 neutron star (Mondal et al., 2024).

The 2025 outburst again showed a hard-state continuum early on. NICER spectra were fit with Tbabs*(Thcomp*bbodyrad) and gave $e<8\times10^{-5}$1 cm$e<8\times10^{-5}$2, $e<8\times10^{-5}$3 keV, $e<8\times10^{-5}$4, and $e<8\times10^{-5}$5 keV, while NuSTAR later in the decay measured a lower flux and weak broad residuals around $e<8\times10^{-5}$6–$e<8\times10^{-5}$7 keV (Sanna et al., 19 Sep 2025). By the late-stage XMM-Newton observation on 2025 March 4, the spectrum had softened to $e<8\times10^{-5}$8 with $e<8\times10^{-5}$9 keV in the absorbed Comptonization model constant * TBabs * (thComp*bbodyrad) (Illiano et al., 17 Jul 2025). This suggests that the accretion flow evolves substantially near the end of the outburst, even though the source remains measurably above quiescence.

4. Coherent pulsations, pulse morphology, and secular timing evolution

The spin signal is a defining property of IGR J17511−3057. XMM-Newton measured coherent pulsations at $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$0 Hz with a total rms pulsed fraction of $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$1. The average EPIC-pn pulse profile required five harmonics, although the fundamental dominated (Papitto et al., 2010). Broad-band timing in 2009 further showed pulsations up to $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$2 keV. The pulsed fraction decreased from about $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$3 at $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$4 keV to $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$5–$$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$6 between $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$7 and $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$8 keV, and then to about $$f(M_1,M_2,i)=\frac{(M_2\sin i)^3}{(M_1+M_2)^2} =1.07086(5)\times10^{-3}\ M_\odot,$$9 at $4.1$00 keV, while the soft lag increased monotonically with energy and saturated at about $4.1$01 ms around $4.1$02–$4.1$03 keV (Falanga et al., 2010).

Swift and RXTE studies of the 2009 outburst found that the fundamental pulse amplitude declined smoothly from about $4.1$04 to about $4.1$05 as the outburst decayed, without dramatic shape changes during most of the slow-decay phase. The soft lag between spectral components increased from about $4.1$06–$4.1$07s to about $4.1$08s, and the sharp transition to the rapid-drop stage around MJD $4.1$09 was accompanied by a pulse-phase shift of about $4.1$10–$4.1$11 cycles (Ibragimov et al., 2011). The same work interpreted the decreasing hotspot area, decreasing iron-line equivalent width, and pulse evolution as signatures of a receding inner disk and changing accretion-column geometry.

The 2015 outburst showed that the spin ephemeris remained stable across outbursts. XMM-Newton measured an average spin frequency of $4.1$12 Hz, and comparison with the 2009 value gave $4.1$13 Hz after inclusion of position-systematic effects, consistent with zero at less than $4.1$14. After accounting conservatively for possible spin-up during the outbursts, the quiescent spin-down upper limit was reported as $4.1$15 Hz s$4.1$16. Interpreted as electromagnetic dipole braking with $4.1$17, this gave $4.1$18 G cm$4.1$19, corresponding to an equatorial magnetic field of order $4.1$20 G (Papitto et al., 2016).

The 2025 timing campaign extended the baseline to three outbursts. A linear fit to the measured spin frequencies yielded an approximate secular spin-down derivative $4.1$21 Hz s$4.1$22, which was interpreted as compatible with a quiescent rotation-powered phase (Sanna et al., 19 Sep 2025). Using the rotating-dipole estimate,

$4.1$23

that study inferred $4.1$24 G cm$4.1$25, corresponding to a polar field $4.1$26 G for a canonical $4.1$27 neutron star (Sanna et al., 19 Sep 2025).

The same 2025 study also argued for rapid orbital shrinkage. Direct fitting of the three measured orbital periods gave $4.1$28 s s$4.1$29, and a Bayesian fit to the ascending-node epochs gave $4.1$30 s s$4.1$31, implying a characteristic timescale of $4.1$32 Myr. Because the gravitational-wave-driven timescale for plausible system parameters is $4.1$33–$4.1$34 Gyr, the reported contraction is much faster than standard binary-evolution expectations (Sanna et al., 19 Sep 2025).

5. Thermonuclear bursts and burst oscillations

Thermonuclear bursting is central to the source classification. Altamirano et al. reported 10 RXTE type-I bursts during the 2009 outburst and detected burst oscillations in all 10, very close to the spin frequency $4.1$35 Hz. The oscillations showed frequency drifts of about $4.1$36 Hz, usually starting below spin during the rise, reaching a maximum near the burst peak, and stabilizing close to spin in the tail. Several bursts also showed evidence for harmonic content, with fundamental amplitudes roughly $4.1$37–$4.1$38 rms and harmonic amplitudes typically $4.1$39–$4.1$40 rms (Altamirano et al., 2010).

The same burst sample established several systematic trends with declining accretion rate during 2009. Bursts became brighter, their rise and decay time scales became shorter, and the oscillation amplitude fell below the detection threshold first in the burst peak and then also in the rise. None of the bursts showed photospheric radius expansion, allowing an upper limit $4.1$41 kpc from the brightest burst with $4.1$42 erg s$4.1$43 cm$4.1$44 under an empirical Eddington luminosity $4.1$45 erg s$4.1$46 (Altamirano et al., 2010).

A simultaneous XMM-Newton study of two 2009 bursts found rise times of about $4.1$47 s, decay times $4.1$48–$4.1$49 s, peak fluxes $4.1$50–$4.1$51 erg cm$4.1$52 s$4.1$53, and fluences $4.1$54–$4.1$55 erg cm$4.1$56. The inferred ignition conditions were stated to be consistent with nearly pure helium burning. Because no PRE was observed, that work derived an upper limit of about $4.1$57 kpc from the peak flux and a lower limit of about $4.1$58 kpc from the burst-radius plateau under the assumption that the decaying burst emission involved the whole neutron-star surface (Papitto et al., 2010).

Chandra observed another 2009 burst on September 22 at 12:54:56 UTC. In the $4.1$59 s interval, the average absorbed $4.1$60–$4.1$61 keV luminosity was $4.1$62 erg s$4.1$63. The burst spectrum was well described by a blackbody with $4.1$64 keV and $4.1$65 km, while time-resolved analysis showed clear cooling from $4.1$66 keV at peak to $4.1$67 keV in the tail, with the inferred radius remaining roughly constant at about $4.1$68 km (Paizis et al., 2012). The burst profile was fitted by a FRED function with rise time $4.1$69–$4.1$70 s and decay time $4.1$71 s, although the paper explicitly noted that Chandra timed-exposure mode broadens the apparent rise relative to RXTE (Paizis et al., 2012).

The 2015 outburst again produced type-I bursts. XMM-Newton detected three bursts at MJD 57108.01023, 57108.34130, and 57108.67847, with e-folding decay times $4.1$72, $4.1$73, and $4.1$74 s and recurrence intervals $4.1$75 ks and $4.1$76 ks. Their peak $4.1$77–$4.1$78 keV flux reached $4.1$79 erg cm$4.1$80 s$4.1$81, and coherent burst oscillations at the spin frequency were detected with $4.1$82 rms amplitude; the burst-oscillation phase was generally consistent with the persistent pulse phase, with deviations up to $4.1$83 cycles near burst peaks (Papitto et al., 2016). INTEGRAL/JEM-X detected two additional bursts, and none of the five showed PRE (Papitto et al., 2016).

NuSTAR detected one type-I burst during the 2025 outburst. It rose in about $4.1$84 s, decayed exponentially with an e-folding time of $4.1$85 s, lasted about $4.1$86 s in total, and reached a peak bolometric flux of $4.1$87 erg s$4.1$88 cm$4.1$89. Time-resolved spectroscopy showed blackbody cooling from about $4.1$90 keV near the rise and peak to $4.1$91 keV in the decay, with the apparent radius increasing from $4.1$92 km to $4.1$93 km. No PRE was detected, but marginally significant burst oscillations were found at $4.1$94 Hz with a global significance of about $4.1$95 and pulsed amplitude $4.1$96 in the best window (Sanna et al., 19 Sep 2025).

6. Aperiodic timing, possible twin kHz QPOs, and the relativistic-precession problem

Kalamkar et al. analyzed all 71 RXTE/PCA pointed observations from 2009 September 12 to October 6 and concluded that IGR J17511−3057 behaved as an atoll source in the hard or extreme island state. In groups 1, 2, and 7 they identified two simultaneous high-frequency peaks, around $4.1$97 and $4.1$98 Hz, $4.1$99 and $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$00 Hz, and $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$01 and $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$02 Hz, respectively. If these are twin kHz QPOs, their separation is approximately $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$03 Hz, close to $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$04 rather than $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$05, despite the source being a slow rotator with $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$06 Hz (Kalamkar et al., 2011).

That result is controversial because the component identification is not secure. The source does not fit cleanly into the existing frequency-correlation scheme for neutron-star LMXBs, and Kalamkar et al. showed that no single multiplicative shift factor can align all of its characteristic frequencies with standard atoll-source tracks; one context would require $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$07, whereas the empirical $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$08 Hz comparison would imply about $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$09 (Kalamkar et al., 2011). The low-frequency timing components are atoll-like, but the high-frequency pair and the hump component remain ambiguous.

The source therefore became a test case for relativistic QPO models. In the relativistic precession model as applied to the twin high-frequency pair,

$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$10

with $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$11, $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$12, and $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$13 constrained by the observed pair (Stefanov, 2015). Application of this model to groups 1, 2, and 7 found that the predictions of group 2 were “practically in conflict” with the other two, and the implied mass ranges were very high: $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$14 The same study inferred that all three QPO-producing radii lay outside the ISCO and followed the ordering $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$15, implying a broad kHz-QPO production region rather than a single narrow annulus (Stefanov, 2015).

A later statistical reformulation used the method of Bambi for multiple simultaneous triads, including the low-frequency hump $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$16, with the associations

$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$17

Because three observed triads yield nine data terms and five fitted parameters $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$18, the joint fit has $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$19 degrees of freedom and can be tested as a true $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$20 statistic (Stefanov, 2023). For IGR J17511−3057 the full three-triad fit gave $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$21 for $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$22 dof, far above the $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$23 critical value $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$24, so the fit was judged poor. Among two-triad combinations, only groups 1+2 were acceptable, with $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$25 for $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$26 dof, whereas any combination including group 7 was poor (Stefanov, 2023). Even in the acceptable 1+2 case, the mass estimate remained well above $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$27, leading to the conclusion that the RPM is not fully viable for this source (Stefanov, 2023).

7. Low-luminosity accretion, propeller tension, and broader significance

The late 2025 XMM-Newton observation introduced a new regime for IGR J17511−3057. The spectrum yielded an unabsorbed $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$28–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$29 keV flux $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$30 erg cm$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$31 s$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$32, corresponding to $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$33 erg s$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$34 at $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$35 kpc. Although this was more than an order of magnitude above the 2019 quiescent luminosity, it is low enough that standard estimates place the source close to or within the propeller regime (Illiano et al., 17 Jul 2025).

Despite that expectation, coherent X-ray pulsations were detected with unusually large amplitudes: $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$36 in $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$37–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$38 keV, $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$39 in $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$40–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$41 keV, $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$42 in $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$43–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$44 keV, and $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$45 in $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$46–$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$47 keV, with no significant detection above $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$48 keV at the highest energies. The pulsation frequency was $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$49 Hz, the false-alarm probability was $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$50, and the pulse profile was described by a single harmonic (Illiano et al., 17 Jul 2025). By contrast, a later NICER observation on 2025 March 15 at $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$51 erg s$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$52 found no significant pulsations, and ATCA on 2025 April 12 detected no radio counterpart, with a $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$53 upper limit of $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$54Jy beam$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$55 at $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$56 GHz (Illiano et al., 17 Jul 2025).

The propeller tension can be expressed through the co-rotation radius,

$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$57

or equivalently as quoted in the late-2025 study,

$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$58

which gives $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$59 km for a $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$60 neutron star spinning at about $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$61 ms (Illiano et al., 17 Jul 2025). Using the low luminosity, that study estimated $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$62 g s$$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$63 and obtained $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$64 G for $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$65 from the requirement $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$66, but noted that this is inconsistent with the independent spin-down-based estimate $$\alpha_{\rm J2000}=17^{\rm h}51^{\rm m}08^{\rm s}.66,\qquad \delta_{\rm J2000}=-30^\circ57'41.0'',$$67 derived from the long-term ephemerides (Illiano et al., 17 Jul 2025, Sanna et al., 19 Sep 2025).

The authors of the late-2025 study therefore argued that accretion is not simply switched off at the nominal propeller boundary. They discussed radiatively inefficient accretion, a trapped disk, and a propeller state with coexisting inflow and outflow as possible resolutions, and compared the source to transitional millisecond pulsars in the sub-luminous disk state (Illiano et al., 17 Jul 2025). That comparison is deliberately limited: there was no clear bimodal high/low mode switching in the XMM-Newton light curve and no radio detection, so the source was not identified as having entered a transitional-millisecond-pulsar state (Illiano et al., 17 Jul 2025).

Within the broader phenomenology of accreting millisecond pulsars, IGR J17511−3057 is therefore important for two distinct reasons. First, it is a rare source in which persistent pulsations, burst oscillations, and type-I bursts can all be studied across multiple outbursts. Second, it combines otherwise standard hard-state AMXP behavior with two persistent anomalies: QPO phenomenology that is difficult to reconcile with standard identification schemes or with the relativistic precession model, and late-stage coherent pulsations at luminosities where simple propeller arguments would predict centrifugal inhibition of accretion (Altamirano et al., 2010, Kalamkar et al., 2011, Stefanov, 2023, Illiano et al., 17 Jul 2025).