The time derivative of the kilohertz quasi-periodic oscillations in 4U 1636-53

We analysed all archival RXTE observations of the neutron-star low-mass X-ray binary 4U 1636-53 up to May 2010. In 528 out of 1280 observations we detected kilohertz quasi-periodic oscillations (kHz QPOs), with ~ 65% of these detections corresponding to the so-called lower kHz QPO. Using this QPO we measured, for the first time, the rate at which the QPO frequency changes as a function of QPO frequency. For this we used the spread of the QPO frequency over groups of 10 consecutive measurements, sampling timescales between 320 and 1600 s, and the time derivative of the QPO frequency over timescales of 32 to 160 s. We found that: (i) Both the QPO-frequency spread and the QPO time derivative decrease by a factor ~ 3 as the QPO frequency increases. (ii) The average value of the QPO time derivative decreases by a factor of ~ 2 as the timescale over which the derivative is measured increases from less than 64 s to 160 s. (iii) The relation between the absolute value of the QPO time derivative and the QPO frequency is consistent with being the same both for the positive and negative QPO-frequency derivative. We show that, if either the lower or the upper kHz QPO reflects the Keplerian frequency at the inner edge of the accretion disc, these results support a scenario in which the inner part of the accretion disc is truncated at a radius that is set by the combined effect of viscosity and radiation drag.

💡 Research Summary

The authors present a comprehensive timing analysis of the neutron‑star low‑mass X‑ray binary 4U 1636‑53 using the entire RXTE archival data set up to May 2010. Out of 1 280 individual observations, kilohertz quasi‑periodic oscillations (kHz QPOs) were detected in 528 cases (≈41 %). The lower kHz QPO, which is typically stronger and narrower, accounts for about 65 % of all detections and is therefore used as the primary tracer of frequency evolution.

To quantify the dynamics of the QPO frequency, the authors first measured the centroid frequency in short (32–160 s) data segments, then grouped ten consecutive measurements to form “blocks” spanning 320–1 600 s. Within each block they computed the standard deviation of the centroid frequencies (the “frequency spread”) and, independently, the absolute value of the frequency derivative (|Δν/Δt|) between adjacent measurements. By varying the block length and the interval between successive frequency estimates they probed the dependence of the spread and the derivative on both the QPO frequency itself and the timescale of the measurement.

Three robust trends emerged. (i) Both the frequency spread and the absolute frequency derivative decrease by roughly a factor of three as the QPO centroid frequency rises from ≈600 Hz to ≈900 Hz. This indicates that the inner accretion flow becomes progressively more stable at higher orbital frequencies. (ii) The average |Δν/Δt| declines by about a factor of two when the timescale over which the derivative is evaluated is increased from <64 s to 160 s, showing that rapid, short‑timescale fluctuations are smoothed out on longer intervals. (iii) The relationship between |Δν/Δt| and the QPO frequency is indistinguishable for positive (frequency increasing) and negative (frequency decreasing) derivatives, implying a symmetric response of the oscillation frequency to the underlying physical driver.

Interpreting these results within the framework of QPO models, the authors argue that if either the lower or the upper kHz QPO directly reflects the Keplerian orbital frequency at the inner edge of the accretion disc, the observed behaviour is naturally explained by a disc truncation radius set by the combined action of viscous stresses and radiation‑drag forces. In the “sonic‑point” picture, radiation drag pulls the disc inward, raising the Keplerian frequency, while viscous torques oppose this motion, leading to a balanced, slowly varying inner radius. The symmetric positive/negative derivatives and the scaling of the derivative with measurement timescale are both consistent with such a balance. Competing models that invoke relativistic precession or Lense‑Thirring nodal precession struggle to reproduce the observed symmetry and the clear dependence of the derivative on the measurement interval.

The analysis also includes extensive checks for systematic effects: variations in signal‑to‑noise ratio, energy band selection, and window length do not alter the main trends, confirming that the measured frequency dynamics are intrinsic to the source rather than artifacts of the data reduction. Moreover, simultaneous examination of the QPO quality factor (Q) and rms amplitude shows that the frequency drift is accompanied by modest changes in coherence, reinforcing the view that the drift reflects genuine adjustments of the disc inner edge rather than stochastic noise.

In summary, this work provides the first quantitative measurement of the time derivative of the kHz QPO frequency as a function of frequency and timescale. The findings support a scenario in which the inner accretion disc of 4U 1636‑53 is truncated at a radius governed by the interplay of viscosity and radiation drag, and they establish a new observational benchmark for testing theoretical models of high‑frequency QPOs. The methodology and results will be directly applicable to upcoming high‑throughput X‑ray timing missions such as NICER, eXTP, and STROBE‑X, which will be able to extend these measurements to finer timescales and to a broader sample of neutron‑star binaries.