What determines the $γ$-ray luminosities of classical novae?

TL;DR

This study tests the internal-shock model for GeV gamma-ray emission in classical novae by assembling a distance-independent, multiwavelength dataset (gamma-ray, optical, spectroscopic) for 2008–2021 novae. It shows a strong, physically motivated correlation between gamma-ray luminosity and the relative velocity between the slow and fast ejecta, supporting a shock-powered mechanism with $L_{ m\gamma} \propto (\Delta v)^3$, and finds that longer gamma-ray emission tends to occur in slower systems. Optical brightness correlates with gamma-ray strength, and dust formation events commonly follow the end of the gamma-ray phase, hinting at a connection between shock cooling, dust-condensation zones, and observational signatures. The results, consistent with internal shocks, emphasize the importance of ejecta kinematics and density in setting nova high-energy outputs and motivate further studies with larger samples to refine the dust-shock connection and emission efficiencies.

Abstract

Classical novae in the Milky Way have now been well-established as high-energy GeV $γ$-ray sources. In novae with main-sequence companions, this emission is believed to result from shocks internal to the nova ejecta, as a later fast wind collides with an earlier slow outflow. To test this model and constrain the $γ$-ray production mechanism, we present a systematic study of a sample of recent Galactic novae, comparing their $γ$-ray properties ($γ$-ray luminosity and duration) with their outflow velocities, peak $V$-band magnitudes, and the decline times of their optical light curves ($t_2$). We uniformly estimate distances in a luminosity-independent manner, using spectroscopic reddening estimates combined with three-dimensional Galactic dust maps. Across our sample, $γ$-ray luminosities ($>$100 MeV) vary by three orders of magnitude, spanning $10^{34}-10^{37}$ erg s$^{-1}$. Novae with larger velocity of the fast outflow (or larger differential between the fast and slow outflow) have larger $γ$-ray luminosities, but are detectable for a shorter duration. The optical and $γ$-ray fluxes are correlated, consistent with substantial thermal emission in the optical from shock-heated gas. Across six novae with $γ$-ray and infrared light curves, evidence for dust formation appears soon after the end of the detected $γ$-ray emission. Dusty and non-dusty novae appear to have similar $γ$-ray luminosities, though novae that have more material processed by the shocks may be more likely to form dust. We find that the properties of the $γ$-ray emission in novae depend heavily on the ejecta properties, and are consistent with expectations for internal shocks.

Figures (76)

• Figure 1: Left: the H$\alpha$ line profiles before (top) and after (bottom) optical peak for nova V906 Car. The red dashed lines represent rest velocity ($v_{\mathrm{rad}}$ = 0 km s$^{-1}$). The blue and green dashed lines represent the velocities of the slow flow ($v_1 = 250$ km s$^{-1}$) and fast flow ($v_2 = 2500$ km s$^{-1}$), respectively. The numbers in brackets are the day of observation relative to the optical peak. The spectrum on day +2 shows absorption from the slow component superimposed on emission from the fast component, suggesting that the slow flow is external to and distinct from the fast flow. We note that the fast outflow velocity shown here is faster than the one shown in Aydi_etal_2020b; this is an effect of the fast outflow velocity varying in this nova, and we adopt a velocity measured during the measured $\gamma$-ray emission for this nova. Right: Schematic illustrations showing the suggested evolution of the nova ejecta near visible peak. Top: Before visible peak, the accreted envelope puffs up due to the thermonuclear runaway, engulfing the system in a common envelope phase and becoming concentrated in the orbital plane (e.g. Livio_etal_1990Chomiuk_etal_2014Sokoloski_etal_2017). This slow flow manifests as a P Cygni profile in the spectrum, characterized by slow (a few 100s km s$^{-1}$) velocities. Bottom: Near peak, a continuous fast wind starts, driven by radiation from the ongoing nuclear burning on the surface of the white dwarf (e.g. Bath_Shaviv_1976Shaviv_2001). The fast flow can propagate more freely in the polar directions due to the concentration of the pre-existing slower ejecta in the orbital plane, and it manifests as broad emission in the spectral lines. When the two outflows collide, they produce shocks -- the likely origin of the $\gamma$-ray emission. Figure adapted from Aydi_etal_2020b.
• Figure 2: The $V$-band magnitude of optical peak ($V_{\mathrm{peak}}$), plotted against the time to decline from peak by 2 magnitudes ($t_2$). The blue triangular data points represent the novae that have not been detected by Fermi. The orange stars represent the systems that have been detected in $\gamma$-rays, and the magenta triangles show novae that are $\gamma$-ray detected, but only have a limit on the peak $V$-band magnitude. Green diamonds represent points that are $\gamma$-ray non-detections and only have limits on $V_{peak}$. Histograms in the top and right panels show the distribution of $t_2$ and $V_{\mathrm{peak}}$, respectively. In each histogram the orange bars display the $\gamma$-ray detected sample, while the blue bars represent the $\gamma$-ray limit sample. Novae with limits on $V_{\mathrm{peak}}$ are excluded from both histograms. Typically, optically bright novae are $\gamma$-ray detected, while fainter novae are not. There are several clear exceptions to this, however, indicating that there are other factors at play here beyond the optical flux.
• Figure 3: Same as Figure \ref{['Fig:vmax_vs_t2']}, except now showing the extinction corrected peak $V$ band magnitudes ($V_{\mathrm{peak},0}$). Sources with $V_{\mathrm{peak},0} < 5$ are frequently detected, and on occasion fainter optical novae are also Fermi detected.
• Figure 4: $\gamma$-ray luminosity for the novae in our sample plotted against their distances. The $\gamma$-ray luminosities in our sample span three orders of magnitude, ranging from $10^{34}$ to 10$^{37}$ erg s$^{-1}$. At larger distances, such as around 8 kpc for novae close to the Galactic center, only the brighter end of this luminosity range is detectable.
• Figure 5: The duration of the $\gamma$-ray emission plotted against the $\gamma$-ray luminosity. On the left panel, we have the average $\gamma$ ray luminosity, while the right panel displays the maximum $\gamma$-ray luminosity. In both cases our data indicates a significant negative correlation between these two parameters, with longer duration $\gamma$-ray emission corresponding to lower average and maximum luminosities. These correlations are driven by novae with $\gamma$-ray emission durations greater than 14 days, where the correlation is clear. This correlation does not translate down to the lower duration systems, where there are four novae at lower luminosities than would be expected from the observed correlation. The translucent black x marks V5668 Sgr at a distance of 1.2 kpc, while the green round point shows V5668 Sgr assuming a distance of 3.9 kpc. The orange stars represent all of the other novae in our sample.