5 Thermonuclear X-ray Bursts
247
The implementation of large nuclear networks confirmed that approximated networks were responsible for irregular bursts in past models, especially with respect
to approximations of the rp-process [181]: full reaction networks are required
to simulate behaviour consistent with observations [201]. Moreover, interactions
between different zones influence the nuclear burning. Ashes of the previous burst
are mixed with fresh fuel, and typically shorten the waiting time for the next burst to
ignite. This effect is referred to as “compositional inertia” [90, 181, 201]. Bursts with
a smaller ignition column depth are less powerful, reach lower peak temperatures,
and the nuclear flow does not extend as far along the rp-process path. Therefore,
compositional inertia changes the composition of the ashes that fuel superbursts and
forms the outer crust. Although it is typically not taken into account self-consistently
in burst simulations, nuclear heating and cooling processes in the crust depend very
sensitively on the ash composition [39, 161]. Furthermore, as many of the required
nuclear reaction rates are highly uncertain, extensive studies have been undertaken
to find the most influential reactions and to quantify their effect on burst observables
[34, 74, 146]. The CNO breakout reaction 15 O(α, γ ) 19 Ne is found to be particularly
important for the stability of the burning processes ([34, 35, 44, 45, 103]; see also
Sects. 5.1.1.4 and 5.6.2).
The burning processes powering pure helium bursts were thought to consist of
3α and a straightforward series of α-captures. Simulations with full networks find,
however, that (α, p) reactions can produce a small number of protons. The protons
act as a catalyst and substantially speed up the burning process by by-passing
12 C(α, γ ) 16 O through the faster reactions 12 C(p, γ ) 13 N(α, p) 16 O.
In recent years multi-zone models of deep carbon burning have been created
to model superbursts [100, 194, 195]. These models directly accrete carbon-rich
material to avoid the computational expense of simulating ∼10 3 hydrogen/helium
flashes, as well as to avoid the problem that models of hydrogen/helium typically
do not produce sufficient carbon. In some studies a hydrogen/helium atmosphere is
added to the model close to the moment of ignition, such that the effect of the deep
carbon flash on the outer atmosphere can be investigated. Hydrogen/helium burning
at the superburst onset contributes to the precursor burst [100, 194]. Furthermore,
the hot ashes of the superburst cause freshly accreted hydrogen and helium to burn
stably, quenching short X-ray bursts for days to weeks [100, 107]. An unsolved issue
is that the carbon burning propagates radially as a convective flame, but convection
is not properly modeled by implicit codes. The one-dimensional simulations find
the flame to spread supersonically as a detonation [195], whereas multi-dimensional
models of comparable carbon flames in Type Ia supernovae produce deflagrations
[202].
Multi-zone models have been very successful at simulating a wide range of
observed bursts and other burning behaviour (e.g. [112]). It has proven a challenge,
however, to reproduce the observed conditions where the different burning regimes
occur. In part this is due to the large parameter space that needs to be investigated
because of uncertainties in, for example, the accretion composition and key nuclear
reaction rates. Furthermore, heating and cooling processes in the crust have a strong
effect on the burning behaviour [94, 98, 206], but their influence has been explored
247
The implementation of large nuclear networks confirmed that approximated networks were responsible for irregular bursts in past models, especially with respect
to approximations of the rp-process [181]: full reaction networks are required
to simulate behaviour consistent with observations [201]. Moreover, interactions
between different zones influence the nuclear burning. Ashes of the previous burst
are mixed with fresh fuel, and typically shorten the waiting time for the next burst to
ignite. This effect is referred to as “compositional inertia” [90, 181, 201]. Bursts with
a smaller ignition column depth are less powerful, reach lower peak temperatures,
and the nuclear flow does not extend as far along the rp-process path. Therefore,
compositional inertia changes the composition of the ashes that fuel superbursts and
forms the outer crust. Although it is typically not taken into account self-consistently
in burst simulations, nuclear heating and cooling processes in the crust depend very
sensitively on the ash composition [39, 161]. Furthermore, as many of the required
nuclear reaction rates are highly uncertain, extensive studies have been undertaken
to find the most influential reactions and to quantify their effect on burst observables
[34, 74, 146]. The CNO breakout reaction 15 O(α, γ ) 19 Ne is found to be particularly
important for the stability of the burning processes ([34, 35, 44, 45, 103]; see also
Sects. 5.1.1.4 and 5.6.2).
The burning processes powering pure helium bursts were thought to consist of
3α and a straightforward series of α-captures. Simulations with full networks find,
however, that (α, p) reactions can produce a small number of protons. The protons
act as a catalyst and substantially speed up the burning process by by-passing
12 C(α, γ ) 16 O through the faster reactions 12 C(p, γ ) 13 N(α, p) 16 O.
In recent years multi-zone models of deep carbon burning have been created
to model superbursts [100, 194, 195]. These models directly accrete carbon-rich
material to avoid the computational expense of simulating ∼10 3 hydrogen/helium
flashes, as well as to avoid the problem that models of hydrogen/helium typically
do not produce sufficient carbon. In some studies a hydrogen/helium atmosphere is
added to the model close to the moment of ignition, such that the effect of the deep
carbon flash on the outer atmosphere can be investigated. Hydrogen/helium burning
at the superburst onset contributes to the precursor burst [100, 194]. Furthermore,
the hot ashes of the superburst cause freshly accreted hydrogen and helium to burn
stably, quenching short X-ray bursts for days to weeks [100, 107]. An unsolved issue
is that the carbon burning propagates radially as a convective flame, but convection
is not properly modeled by implicit codes. The one-dimensional simulations find
the flame to spread supersonically as a detonation [195], whereas multi-dimensional
models of comparable carbon flames in Type Ia supernovae produce deflagrations
[202].
Multi-zone models have been very successful at simulating a wide range of
observed bursts and other burning behaviour (e.g. [112]). It has proven a challenge,
however, to reproduce the observed conditions where the different burning regimes
occur. In part this is due to the large parameter space that needs to be investigated
because of uncertainties in, for example, the accretion composition and key nuclear
reaction rates. Furthermore, heating and cooling processes in the crust have a strong
effect on the burning behaviour [94, 98, 206], but their influence has been explored