Arxiv:1707.04966V3 [Astro-Ph.HE] 17 Dec 2017

From hadrons to quarks in neutron stars: a review

Gordon Baym,1, 2, 3 Tetsuo Hatsuda,2, 4, 5 Toru Kojo,6, 1 Philip D. Powell,1, 7 Yifan Song,1 and Tatsuyuki Takatsuka5, 8 1Department of Physics, University of Illinois at Urbana-Champaign, 1110 W. Green Street, Urbana, Illinois 61801, USA 2iTHES Research Group, RIKEN, Wako, Saitama 351-0198, Japan 3The Niels Bohr International Academy, The Niels Bohr Institute, University of Copenhagen, Blegdamsvej 17, DK-2100 Copenhagen Ø, Denmark 4iTHEMS Program, RIKEN, Wako, Saitama 351-0198, Japan 5Theoretical Research Division, Nishina Center, RIKEN, Wako 351-0198, Japan 6Key Laboratory of Quark and Lepton Physics (MOE) and Institute of Particle Physics, Central China Normal University, Wuhan 430079, China 7Lawrence Livermore National Laboratory, 7000 East Ave., Livermore, CA 94550 8Iwate University, Morioka 020-8550, Japan (Dated: December 19, 2017) In recent years our understanding of neutron stars has advanced remarkably, thanks to research converging from many directions. The importance of understanding neutron star behavior and structure has been underlined by the recent direct detection of gravitational radiation from merging neutron stars. The clean identification of several heavy neutron stars, of order two solar masses, challenges our current understanding of how dense matter can be sufficiently stiff to support such a mass against gravitational collapse. Programs underway to determine simultaneously the mass and radius of neutron stars will continue to constrain and inform theories of neutron star interiors. At the same time, an emerging understanding in quantum chromodynamics (QCD) of how nuclear matter can evolve into deconfined quark matter at high baryon densities is leading to advances in understanding the equation of state of the matter under the extreme conditions in neutron star interiors. We review here the equation of state of matter in neutron stars from the solid crust through the liquid nuclear matter interior to the quark regime at higher densities. We focus in detail on the question of how quark matter appears in neutron stars, and how it affects the equation of state. After discussing the crust and liquid nuclear matter in the core we briefly review aspects of microscopic quark physics relevant to neutron stars, and quark models of dense matter based on the Nambu– Jona-Lasinio framework, in which gluonic processes are replaced by effective quark interactions. We turn then to describing equations of state useful for interpretation of both electromagnetic and gravitational observations, reviewing the emerging picture of hadron-quark continuity in which hadronic matter turns relatively smoothly, with at most only a weak first order transition, into quark matter with increasing density. We review construction of unified equations of state that interpolate between the reasonably well understood nuclear matter regime at low densities and the quark matter regime at higher densities. The utility of such interpolations is driven by the present inability to calculate the dense matter equation of state in QCD from first principles. As we review, the parameters of effective quark models – which have direct relevance to the more general structure of the QCD phase diagram of dense and hot matter – are constrained by neutron star mass and radii measurements, in particular favoring large repulsive density-density and attractive diquark pairing interactions. We describe the structure of neutron stars constructed from the unified equations of states with crossover. Lastly we present the current equations of state – called “QHC18” for quark-hadron crossover – in a parametrized form practical for neutron star modeling.

I. INTRODUCTION International Space Station will signiﬁcantly constrain the neutron star equation of state. Such constraints are

arXiv:1707.04966v3 [astro-ph.HE] 17 Dec 2017 crucial for understanding observations of dynamical neu- Neutron stars provide a cosmic laboratory in which the tron star phenomena, from neutron star seismology [21] phases of cold dense strongly interacting nuclear matter to binary neutron star inspirals, now detected gravita- are realized [1–4]. Indeed, while heavy ion collision ex- tionally [22] and subsequent mergers detected by multi- periments and lattice quantum chromodynamics (QCD) messenger electromagnetic signals [23]. Reliable equa- simulations provide insight into the properties of hot and tions of state, at zero and elevated temperatures, are dense QCD, neutron stars are the only known window crucial for predicting the gravitational wave signatures into the rich structure of cold dense QCD. Recent astro- of neutron star–black hole and neutron star–neutron star physical inferences of neutron star masses, M, and radii, mergers [24–30] to be detected at gravitational wave ob- R, in low mass x-ray binaries [5–14], and the wealth of servatories present and future, including LIGO [31–33], new data, on masses and radii of isolated neutron stars Virgo [34], GEO [35], KAGRA [36], LIGO-India [37], and as well, expected from the NICER (the Neutron Star In- LISA and other spaced-based observatories [38], as well terior Composition Explorer) experiment [15–20] on the 2 as via pulsar timing arrays [39]. The purpose of this review is to outline our current understanding of the microscopic physics of dense matter in the interior of neutron stars, and from this standpoint to construct families of equations of state useful for interpretation of both electromagnetic and gravitational observations. In addition, as the only source of “data” on cold high density matter in QCD, neutron stars provide a rich testing ground for microscopic theories of dense nuclear matter, providing an approach complementary to probing dense matter in ultrarelativistic heavy ion collision ex- periments at the Relativistic Heavy Ion Collider (RHIC) in Brookhaven and the Large Hadron Collider (LHC) at CERN. A major challenge is to understand the facets of microscopic interactions that allow the existence of massive neutron stars. Discoveries in recent years of neutron stars with M ∼ 2 solar masses (M ), includ- FIG. 1: . Schematic phase diagram of dense nuclear matter, in ing the binary millisecond pulsar J1614-2230, with mass the baryon chemical potential µB –temperature T plane. At zero 1.928±0.017M [40] (the original mass measurement was temperature, nucleons are present only above µB ∼ MN , the 1.97 ± 0.04M [41]), and the pulsar J0348+0432 with nucleon mass. At the low temperatures inside neutron stars, matter evolves from nuclear matter at low densities to a quark-gluon mass 2.01 ± 0.04M [42] present a direct challenge to 1 plasma at high density. BCS pairing of quarks in the plasma theoretical models of dense nuclear matter. regime leads to the matter being a color superconductor. (Low The existence of such massive stars has important im- temperature BCS pairing states of nucleons are not shown.) At plications for dense matter in QCD. For example, they higher temperatures, matter becomes a quark-gluon plasma, with require a stiff equation of state, i.e., with large pressure a possible line of first order transitions, the solid line, terminat- for a given energy (or mass) density, and thus rule out ing at high temperatures at the proposed Asakawa-Yazaki critical point [48]. In addition, the solid line may terminate in a low tem- a number of softer theoretical models, and at the same perature critical point [49]. time impose severe constraints on the possible phases of dense QCD matter. In particular, massive neutron stars are difficult (but not impossible) to explain in the context below. The temperatures in neutron stars, characteris- of hadronic models of neutron star matter in which the tically much smaller than 1 MeV (or 1010 K), are well emergence of strange hadrons around twice nuclear sat- below the temperature scale in Fig.1, of order 10-10 2 uration density softens the equation of state and limits MeV; matter in neutron stars lives essentially along the the maximum stable star mass. chemical potential axis in this figure. The exception is at neutron star births in supernovae where temperatures can be tens of MeV, and in final gravitational mergers A. Phases of dense matter where temperatures could reach ∼ 102 MeV. The main problem on which we focus in this review is the descrip- Figure1 summarizes the phases of dense nuclear mat- tion of such matter, and the resulting models of neutron ter in the baryon chemical potential µB – temperature stars, i.e., the profiles of baryons density, etc., as a func- T plane [47]. [The baryon chemical potential, increas- tions of radius from the center of the star. ing with increasing baryon density, here nucleons, is the derivative of the free energy density with respect to the density of baryons.] At low temperature and chemical B. Neutron star models – the TOV equation potential the degrees of freedom are hadronic, i.e., neutrons, protons, mesons, etc.; and at high temperature or chemical potential matter is in the form of a quark-gluon We briefly recall that to construct a model of a neu- plasma (QGP) in which the fundamental degrees of free- tron star – once one specifies the equation of state, which gives the pressure, P , as a function of the mass density, dom are quarks and gluons. The nature of the transitions 2 from hadronic to a QGP are sketched in Figs.2 and3 ρ = ε/c , where ε is the total energy density and c is the speed of light – one integrates the general relativistic Tolman-Oppenheimer-Volkov (TOV) equation of hydrostatic balance [50, 51]:

1 In addition the extreme black widow millisecond pulsars PSR ∂P (r) J1957+20 [43], PSR J2215+5135 [44], and PSR J1311-3430 [45, = (1) 46] possibly have masses as large as 2.5 M ; however the masses ∂r 2 remain uncertain owing to the need for more complete modeling ρ(r) + P (r)/c 3 2 of the heating of the companion stars by the neutron stars. −GN 2 [m(r) + 4πr P (r)/c ], r (r − 2GN m(r)/c ) 3 where GN is Newton’s gravitational constant and a hard core radius, ∼ 0.5 fm or the range of two pion −1 Z r exchange, ∼ (2mπ) = 0.7 fm. Then the measure of m(r) = 4πr02dr0ρ(r0) (2) importance of the next order forces becomes the param- 0 3 eter (4πr0/3)nB ∼ (0.1-0.2)nB/n0. An alternative way is the mass inside radius r. Therefore, the mass (M) of to estimate the importance of higher body forces would the neutron star is given by M = m(R) with R the stellar be via chiral perturbation theory. Comparison of four- radius. In practice, to calculate the pressure, P (ρ), one body interaction energies, based on a subset of possible first calculates the energy density, ε, as a function of the processes [58], with three-body interaction energies [59], baryon density, nB, and then uses the thermodynamic re- suggests prima facie that the relative importance of four 2 lation P = nB∂(ε/nB)/∂nB. Equivalently, one can also body compared with three is measured by the parameter 2 calculate P directly as a function of the baryon chemical (gA/fπ) (nB/∆) ∼ 0.9nB/n0, times numerical factors of potential, µB. Generally, we discuss the equation of state order unity, where gA ' 1.25 is the pion axial vector- in the form P (µB). As we show in AppendixA, when the renormalization constant, fπ ' 93 MeV is the pion de- basic scale of the energy entering the equation of state cay constant, and ∆ ' 293 MeV is the mass difference is the proton mass, mp, the typical scale of the neutron of the excited nucleonic state ∆(1232) and the nucleon. 3/2 Although a more accurate estimate from chiral pertur- star mass is given by M ∼ mp/α = 1.86M , and the G bation theory remains an open problem, these estimates scale of radii is given by ( /m c)α−1/2 = 17.2 km, where ~ p G indicate that at the densities achieved in neutron star α = m2G / c ' 0.589 × 10−38 is the gravitational fine G p N ~ interiors, n (3 − 6)n , a well defined expansion in structure constant. B & 0 terms of two-, three-, or more, body forces may not exist. Furthermore, beyond baryon densities a few times n0 the forces between particles should no longer be describ- C. Microscopic calculations of dense nucleonic matter able by static few-body potentials. At the same time, however, the density remains much too low to treat the matter as weakly interacting quark matter; indeed per- Microscopic calculations of the equation of state of turbative QCD (pQCD) begins to become applicable for neutron-star matter, ε(nB), have been based on a va- baryon chemical potentials µ (3-6) GeV, correspond- riety of inputs. The approach most firmly founded on B & ing to baryon densities nB &(10-100) n0. experiment in the region of nuclear saturation density, n0 ' 0.16 nucleons per fm3, or equivalently a mass density In addition to the limitations inherent in any few-body ' 2.7 × 1014 g cm−3, is to use nucleon-nucleon scattering potential model, equations of state based on nucleons data below 350 MeV and the properties of light nuclei alone fail to account for the rich variety of hadronic to determine two-body potentials together with a three- degrees of freedom that enter with increasing density, nucleon potential [52, 53]. Quantum Monte Carlo calcu- including ∆’s, strangeness [60–66], and meson conden- lations based on such potentials give an excellent account sates [67] including pionic [52, 68–72] and kaonic [73– of the binding energies as well as excitation energies of 76]. The presence of hyperons in neutron stars is dif- light nuclei [54]. Quantum Monte Carlo calculations have ficult to predict, owing to the uncertainty of the forces been applied to neutron star structure, and the radius in between hyperons and nucleons as well as between other particular, in Ref. [55]. hyperons. While elementary hadronic models indicate Calculations of dense matter, which have primarily the emergence of hyperons at densities ∼ (2-4) n0, mix- been carried out in the limits of pure neutron matter and ing of strange degrees of freedom at such low densities symmetric nuclear matter, have uncertainties owing both softens the equation of state; the presence of hyperons is not obviously compatible with the stiff equation of state to the interactions used, especially at densities above n0, and the calculational methods employed [56, 57]. As dis- required by the existence of & 2M stars [77]. Attempts cussed below, extensions to neutron star matter in beta have been made to avoid the softening of the equation equilibrium have been based on interpolation between of state by introducing repulsive forces between strange these two limits, which introduces further uncertainties hadrons and nucleons and between hyperons as well as for matter in neutron stars. Λ-nucleon-nucleon forces, which shift the emergence of Not only are the explicit three-body interactions in nu- strangeness to higher densities [64, 65, 78–80]. Lattice clear matter not well determined [55] [see Ref. [56] for a gauge theory – solving QCD on a space-time lattice us- detailed analysis of the significant uncertainties in pure ing Monte Carlo techniques [81] – will in the future be neutron matter introduced by three body forces], one able to provide first principles information on hyperon- must ask at higher density when higher body interac- nucleon and hyperon-hyperon interaction potentials [82]. tions, e.g., four body, become important. Most naively Another approach to high density matter has been one can argue that the relative importance of higher body based on nucleons interacting via elementary meson ex- forces is determined dimensionally, since the ratio of the changes (e.g., [83, 84]; see [85] for a general summary of energy density En+1 from n + 1 body forces to that from equations of state). In such models, multiple meson ex- 3 n body forces will be ∼ (4πr0/3)nB, where r0 is a char- changes and virtual baryonic excitations, which comprise acteristic length of order the range of the nuclear force, the intermediate states in such theories, again raise the 4

clear and electromagnetic interactions drive dense matter toward both local color and electrical neutrality. At low densities, this results in strong correlations between particles, with correlations weakening with increased density. In the hadronic regime, three quarks bind together to produce a color singlet object. In the regime between hadronic and quark matter, colored quarks and diquarks appear virtually during quark exchanges between baryons – essentially the baryon-baryon interactions. In the quark matter regime a diquark or a pair of quarks can FIG. 2: Schematic picture of the transition from nuclear to de- easily find an extra quark nearby to produce local color confined quark matter with increasing density. i) For nB . 2n0, neutrality, so that the extra quark is weakly correlated the dominant interactions occur via a few (∼1-2) meson or quark with the diquark or pair, as shown in Fig.2. exchanges, and description of the matter in terms of interact- With increasing baryon density or temperature, the ing nucleons is valid; ii) for 2n0 . nB . (4-7) n0, many-quark exchanges dominate and the system gradually changes from effective degrees of freedom of matter change, possibly hadronic to quark matter (the range (4-7) n0 is based on geo- accompanied by phase transitions. The phases of QCD metric percolation theory – see Sec.VE); and iii) for nB & (4-7) are characterized by a variety of condensates in which n0, the matter is percolated and quarks no longer belong to spe- a macroscopic number of particles (and antiparticles) cific baryons. A perturbative QCD description is valid only for are strongly correlated by the strong interaction [67]. n 10-100 n . B & 0 The emergence of condensates reduces the energy of the system, and in addition, condensates break symmetries in QCD, leading to states with lower symmetry than is present in the QCD Hamiltonian. The condensates, which depend on temperature and baryon density, play an important role in the structure of hadrons, as well as in neutron stars, since the condensation energies are a large fraction of the energy density in a neutron star core. Chiral symmetry breaking, caused by the chiral condensation of paired quarks and antiquarks with different chirality (or handedness) – characterized most simply by a non-vanishing chiral condensate hqq¯ i, where q is the FIG. 3: Schematic picture of the crossover transition from the quark field – is largely responsible for hadron masses and hadronic to quark-gluon plasma phase with increasing temper- the existence of the nearly massless Nambu-Goldstone ature. i) For T . Tc, the system is a dilute gas of hadrons; ii) bosons [89], e.g., the pion as well as the kaon. As reviewed for Tc . T . (2-3) Tc, thermally excited hadrons overlap and begin to form a semi quark-gluon plasma (see text below); and in [90], chiral condensation persists from the vacuum, to iii) for T & (2-3) Tc, the matter is percolated and a quasiparti- nuclear matter [91], to high density quark matter. At cle description of quarks and gluons, including effects of thermal high baryon density the condensate hqq¯ i is expected to go media, becomes valid. to zero; however, owing to the formation of further condensates, e.g., by diquark-anti diquark pairs (see below), chiral symmetry is expected to continue to be broken at question of whether well-defined “asymptotic” laboratory high densities [92–95]. hadrons are the proper degrees of freedom to describe the In addition, QCD color-magnetic interactions favor system at high density. the formation of a diquark condensate of quark pairs in quark matter at low temperatures, characterized by a non-vanishing expectation value hqqi, and similar to the D. Quark degrees of freedom Bardeen-Cooper-Schrieffer (BCS) condensate of electron pairs in a superconductor [96]. Such a condensate breaks More realistically, one expects in dense matter a grad- the U(1) symmetry associated with baryon conservation, ual onset of quark degrees of freedom, not accounted for and leads to low temperature quark matter being a color by nucleons interacting via static potentials. Indeed, as superconductor. illustrated in Fig.2, at a sufficiently high density the Finally, the breaking of scale symmetry in QCD is re- matter should percolate, in the sense that their quark lated to formation of a gluon condensate – characterized µν a constituents are able to propagate throughout the sys- by a non-vanishing expectation value hFa Fµν i, where a tem [86–88]. The deconfinement of nuclear matter with Fµν (with µ, ν the space-time and a the color indices) is increasing density has many similarities to the manner in the gluon field tensor [97]. which atomic gases, when compressed, become gases of The possible astrophysical role of quarks in stars, and itinerant electrons in a background of ions. Strong nu- neutron stars in particular, has been discussed ever since 5 the first proposal of the quark model of hadrons [98– 100]. Because of the difficulty in describing hadrons and quarks within a single framework, the conventional description of the onset of quark matter has been to regard (hadronic) nuclear and quark matter as distinct phases, to calculate their energy densities using very different models, and then choose the phase with the lower energy density at given baryon density. In order to guar- antee pressure and chemical potential continuity across the transition, one must make a bitangent Maxwell construction, which leads to a first order phase transition from nuclear to quark matter, e.g., [100–103]. Stars fab- ricated with such a “hybrid” equation of state – hybrid stars – consist typically of a small quark matter core sur- rounded by hadronic matter. Reference [104] presents an illuminating analysis of why the quark cores are typically small. The conventional picture is based on a thermodynamic comparison of hadronic matter and quark mat- FIG. 4: A unified equation of state using a nuclear equation ter at too high a density for the description of hadronic of state for nB . 2n0 and a quark matter equation of state for matter to be physical, and at too low a density to apply nB & (4 − 7)n0, interpolated in the intermediate region. The perturbative QCD to the quark phase. An alternative dotted curves, the extrapolations of the nuclear and quark matter equations of state, indicate how such extrapolations become description is that with increasing density, neutron star unreliable. matter undergoes an essentially continuous transformation from the hadronic to quark regimes, a scenario we refer to as hadron-quark continuity [49, 94, 105–111]. In by lattice Monte Carlo calculations [127–130]. At low this scenario, as the density increases, quark degrees of temperature, space is filled with a dilute gas of hadrons; freedom gradually emerge, with partial restoration of chi- as the temperature increases this hadronic gas gradually ral symmetry and the onset of color superconductivity. fills the space, until near T the hadrons begin to merge, In the picture of QCD exhibiting a continuous evolu- c continuously transforming the matter structure and lead- tion from hadronic to quark matter at low temperature ing to a breakdown of its description purely in terms of neutron star cores are composed of deconfined u, d, and hadrons. Yet, a quasiparticle description based on quarks s quarks. Strangeness, rather than appearing in matter and gluons does not apply either, for while the hadrons as hyperons, whose interactions are poorly known, it ap- have broken down, the strong correlations remain. Thus, pears as strange quarks. Importantly, equations of state the system may be described as a strongly correlated exhibiting hadron-quark continuity are consistent with quark-gluon plasma [131, 132] or a “semi-QGP” [133– current observational inferences of neutron star radii, as 135] in which both hadronic-like and quark-like degrees well as neutron star masses ∼ 2M . The emerging de- of freedom can exist. Indeed perturbative QCD (pQCD) scription of dense matter, on which we focus in this re- calculations indicate that the quasiparticle picture of a view, is that at low densities matter is hadronic, and QGP begins to apply only beyond (2-3) T [136, 137]. at high densities is quark matter; in the intermediate c This crossover behavior of low density QCD matter has regime, where one does not at present have tools to cal- a number of features that are likely to be found in the culate, one can make physically constrained plausible in- hadron-quark transition at low temperature. However, it terpolations between these two limits, arriving at what is important to note that while at zero baryon chemical we shall refer to as a unified equation of state2 does use potential physical gluons can be thermally excited as one a to describe matter across the entire range of densities approaches T , in cold dense matter gluons appear only found in the interior of neutron stars [116–122]. Figure4 c virtually, through quantum fluctuations. The properties illustrates the construction of a unified equation of state. of gluons at low temperatures, which are much less well The QCD phase transition at finite temperature and known than those at high temperature, is an important low baryon density [123–126], illustrated in Fig.3, is an subject on which light may be shed through studies of example of the continuity from hadrons to quarks and neutron stars [118–120, 138]. gluons. The smoothness of the evolution of the matter at the transition temperature T has been established Calculations of neutron star masses and radii require c integration over vastly different density scales, from the crust region to the core. The outer regions of a neutron star, which are better understood, consist of a solid crust 2 Our usage of this term should not be confused with its prior use and a nuclear matter liquid just inside the crust, span- in the literature, e.g., [112], to describe equations of state arising ning baryon densities up to ∼ 2n0. The neutron star from a consistent physical model in the crust and liquid interior, interior, for which densities are nB ∼ (2 − 10)n0 is the e.g., in Refs. [113–115]. most difficult to describe from first principles. In this 6 regime quarks and gluons begin to play an important adiabatically connected to those in the vacuum. The re- role, but the density is not high enough to apply results quirement that the neutron star equation of state be very from pQCD. Thus, one is required to introduce some de- stiff, i.e., has a sufficiently large pressure for given en- gree of phenomenological modeling. Understanding the ergy density – to allow 2M neutron stars – tightly con- role of confinement and percolation in this regime is a strains the range of effective parameters of these models. critical and outstanding problem, which renders practi- In this way neutron star constraints translate into con- cal modeling for nB ∼ (2-10)n0 difficult without intro- straints on model parameters, with possible density de- ducing assumptions on confining forces and percolation. pendence, for delineating properties of low energy QCD At higher densities, where matter is a well defined quark- matter [3, 110, 111, 119, 121, 122, 148]. gluon plasma, one faces the difficulty that lattice gauge theory for QCD is at present unable to describe matter at the relevant baryon chemical potentials [81]. Thus one must introduce models of interacting quark matter in this regime. Only at much higher densities, beyond those ex- F. Determining the mass-radius relation pected in neutron stars, can one apply perturbative QCD observationally for dense matter directly. We briefly discuss three current observational developments – determinations of the neutron star mass-radius E. Modelling hadronic and quark matter relation; binary neutron star mergers, as seen both in gravitational and electromagnetic radiation; and mea- To illustrate the evolving physics with increasing den- surements of the thermal properties of neutron stars – sity, we will generally work with a low density nuclear developments which are rapidly leading to a greater un- equation of state at nB . 2n0 and a high density quark derstanding of the properties of neutron star interiors. matter equation of state at nB & (4−7)n0. In actual cal- Accurate determination of the mass-radius relation of culations we take the specific value 5n0. For concreteness, neutron stars strongly constrains the equation of state of we adopt the Akmal-Pandharipande-Ravenhall (APR) neutron star matter. Beyond simple estimates of radius equation of state [52] for nucleons as a representative based on spectroscopic analyses of photons from neu- nuclear equation of state, while noting that its extremely tron stars in quiescence (see [20] and references therein), stiff character at densities well above saturation density further information on the relation can be gained from provides a rough upper limit for neutron star masses in a following the evolution of thermonuclear X-ray bursts purely hadronic description. Since cold dense matter in on the surfaces of accreting neutron stars in binaries QCD cannot be calculated directly in lattice gauge theory [5,6, 20]. In addition the NICER experiment [15–20] [81], one must resort to phenomenological models of QCD is currently directly measuring the mass and radii of a to describing interacting quarks in neutron stars. For a number of neutron stars. quark model at high density as a template, we adopt The photon flux, F∞ per unit area, from a neutron star the three quark-flavor Nambu-Jona-Lasinio (NJL) model at distance D, is related to the surface flux, Fs, of the 2 2 2 [139–143], which captures much of the physics needed in star of radius R by 4πD F∞ = 4πR Fs/(1 + z) , where 2 1/2 neutron stars, and which has been employed previously 1 + z = 1/(1 − 2GN M/Rc ) is the redshift. For black- in studies of dense QCD matter [47, 90, 108, 109, 144– body emission from the surface at an apparent temper- 4 147]. The fundamental actors in the model are quarks; ature Teff , Fs = σTeff , where σ is the Stefan-Boltzmann while the model does not manifestly take gluon degrees constant. In reality, Teff is not the true surface color tem- of freedom into account, their effects in cold matter are perature, Tcolor, deduced from thermal fits of the spectra, to a large degree modeled by letting the quarks inter- since it depends on the physics in the atmosphere, e.g., act via a number of effective interactions reflecting non- its composition and the surface gravity; the two temper- perturbative QCD processes involving quarks and gluons atures are phenomenologically related by Tcolor = fcTeff at low energy. We summarize in subsec.IVD the ef- where fc ∼ 1.3 − 2. The apparent angular size of the fective interactions relevant for QCD phenomenology in neutron star seen by the observer is [150], neutron stars. 2 Although we do not review models of the intermedi- F∞ R∞ Aapp ≡ 4 = 2 4 , (3) ate density nuclear matter lying between the low density σTcolor,∞ D fc hadronic and high density quark regimes, the equation of state in this confinement-dominated regime is, as we in terms of the color temperature seen at the point of ob- shall see, strongly constrained. At higher densities, where servation, Tcolor,∞ = Tcolor/(1 + z). The effective radius baryons merge with one another, a detailed description of determining the angular size is R∞ = R(1+z), which can confining forces becomes unnecessary, and the remaining be understood in terms of light bending by the neutron interactions can be inferred from hadronic and nuclear star [149]. From observations of F∞, D, and Tcolor,∞ one phenomenology, by applying the hadron-quark continuity can constrain the relation between M and R. The dis- picture in which the structure and model parameters are tance D typically has large uncertainties, ∼ 50 %, while 7 the uncertainty is the level of 5 − 10 % for neutron stars in globular clusters. There are two important caveats in this procedure. The first is that neutron stars in strong magnetic fields 12 & 10 G are not black-body emitters; atoms at the surface are expected to be highly distorted by the magnetic field, making the surface radiation dependent on its polarization [151–153]. The second is the assumption that the radiation is from the neutron star surface uniformly, not from a hot spot on the surface, nor from further out in the atmosphere. Observations of thermonuclear X-ray bursts on neutron star surfaces (see [5,6, 11–13, 20, 150], and references therein) further constrain the mass-radius relation. In bursts in which the radiation flux exceeds the local Eddington limit (where the radiative force on the atmosphere balances the gravitational force) the photosphere of the neutron star is lifted; when it returns to the neutron star surface, the flux at the “touchdown” point, seen FIG. 5: Gravitational lensing, which is stronger for neutron from infinity is, stars with larger compactness M/R, permits one to see part of the back side of a neutron star. The inset shows the X-ray flux G Mc from a single (red) hot spot on a rotating neutron star, the green N line, as a function of the phase of the rotation. In the absence of FEdd,∞ = 2 , (4) (1 + z)κeD gravitational lensing, the flux from a point hot spot, shown as the dotted red line, would vanish for part of a rotational cycle; lens- where κe is the electron-scattering opacity in the atmo- ing shrinks the invisible region, allowing the spot to be seen for a sphere. Combining measurements of the touchdown flux, larger fraction of the phase, and reduces the contrast between the flux from the brightest and darkest regions. Combining general which is independent of Teff , with information from the quiescent flux F leads to useful constraints on M vs. relativistic effects, the velocity deduced from Doppler shifts of the ∞ spot color, and the rotational frequency allows one to constrain R. the neutron star mass and radius. The ongoing NICER experiment [15] aims to determine the masses and radii of several nearby rotating neutron stars by accurately monitoring their X-ray pulse profiles ∼ 5-10%. in time. Neutron stars with strong magnetic fields have hot spots at the magnetic polar caps, heated by charged particles moving along the magnetic flux lines. The pulse profiles are periodically modulated by brightness changes G. Binary neutron star mergers associated with these temperature non-uniformities on the neutron star surface. The key is that gravitational The recent multi-messenger detection of the merger of bending of light by the strong gravitational field of neu- two neutron stars GW170817 [22, 23] opens a new path tron stars, dependent on M/R, allows one to partially to determining neutron star properties both by compar- “see” the back side of the neutron stars. Figure5 ing the measurable gravitational signals with those com- schematically illustrates the basic mechanism for a sin- puted by fully general relativistic simulations with given gle hot spot. With stronger bending, more of the star QCD equations of state [24–29], as well as utilizing the is visible, and thus the measurable ratio of the ampli- electromagnetic signals accompanying the merger [154– tudes between the brightest and darkest points in the 158, 160–162]. A binary neutron star system at large profile decreases. The bending for slowly non-rotating distance is described by Newtonian mechanics, but as stars, described by the Schwarzschild metric, depends the stars approach each other in the late inspiral phase only on M/R, and thus to get information on M and the stars deform each other, prior to merging in tens of R separately requires looking at sufficiently rapidly ro- seconds. The tidal deformability of the stars, discussed tating stars, above ∼300 Hz in practice, in which cor- below, is very sensitive to their compactness. When the rections from the Kerr metric of rotating neutron stars orbital frequency reaches ∼ 500 Hz (the current limit come into play. The pulse profiles reflect both general rel- of LIGO detectability), the detectable gravitational sig- ativistic and Doppler effects; comparison with waveform nals begin to distinguish between different equations of modeling, including as a function of color, allows one to state. The gravitational waveform just before the merger, extract M and R, as well as other parameters such as especially its frequency, is again strongly dependent on the quadrupole moment and moment of inertia, which the compactness of the stars; smaller stars can approach play a role in determining the metric of a strongly grav- more closely before crushing into each other, thus reach- itating rotating star [18, 19]. Eventually NICER should ing higher orbital frequency. determine neutron star radii and masses to an accuracy Neutron star–neutron star mergers explore the possi- 8 ble existence of very heavy neutron stars and light black bility can be written in terms of the dimensionless Love 4 holes. After the merger, the coalesced star can either number, k2, as collapse into a black hole, or, if the star is not too mas- 5 sive ( 3 M ) and spinning rapidly, it could remain as a 2 R . λ = k2 . (9) metastable hypermassive star, with high differential ro- 3 GN tation and temperature ∼ 10−100 MeV. The final object Gravitational radiation waveforms in neutron star produced by the merger GW170817 has a mass of order mergers depend on the dimensionless tidal deformabil- 2.7 M , which exceeds the cold non-rotating neutron star ity, Λ, defined by mass bound proposed in [158, 159]. The observed jets favor the object being a black hole [154–157], since jet 2 5 λGN 2 Rc formation takes place through conversion of rotational Λ = 32 = k2 . (10) R5 3 MG energy into magnetic [28, 29]; others argue for it to be a s N long-lived massive neutron star [160]. The event has also 2 where Rs = 2MGN /c is the Schwarzschild radius of the been used to place constraints on the radii of the initial star; see Sec.VII. The strong dependence of Λ ∼ (R/M)5 neutron stars [161], as well as to suggest a lower bound indicates that (for fixed k2) the dimensionless tidal de- on the tidal deformability of the stars [162]. formability is a good measure of the neutron star com- As two neutron stars approach each other in a merger pactness, M/R. For an introduction to calculations of λ each tidally deforms the other. For two stars, A and B for general relativistic stars see [166], and for the effects of mass MA and MB, whose centers are separated by a of tidal deformations on gravitational merger waveforms ~ ~ 2 2 large distance R0 (with |R0| GN MA/c ,GN MB/c ), see [167–169] and references therein. the quadrupolar tidal gravitational field from star B felt at position ~r in star A (measured from the center of A) is H. Neutrino cooling and transport GN MB 2 ˆ 2 1 i j Φtidal(~r ) = 3 r − 3(~r · R0 ) = Eijr r , 2R0 2 Understanding transport and neutrino cooling prop- (5) erties of cold dense matter is essential to determining the phase structure and low energy degrees of freedom where E = ∂2Φ /∂r ∂r is the quadrupolar tidal field ij tidal i j in neutron stars [170–172]. The interiors of stars older tensor. Such a tidal potential distorts star A, producing than a few hundred years are nearly isothermal; however a quadrupole moment, the surface temperatures depend on the thermal trans- Z 1 port from the interior through the crust to the surface Q = d3r ρ(r) r r − δ r2 , (6) ij i j 3 ij (see [171]). When external heating of a star, e.g., by related to the tidal field in linear order by accretion from a companion, is not large, the long-time evolution of neutron star temperatures is dictated by the Qij = −λAEij, (7) neutrino luminosity for the first ∼ 105 − 106 years, after which photon luminosity from the surface dominates. thus defining (relativistically) the tidal deformability λA of star A. Explicitly Neutrino cooling by hadronic matter depends strongly on the fraction of nucleons that are protons. If the proton G M N B fraction is less than some 10-15%, the proton and elec- Qzz = −2Qxx = −2Qyy ≡ −λAEzz = 2λA 3 ; R0 tron Fermi momenta are too small for the direct process − + (8) of neutrino emission, n → p + e +ν ¯e, p → n + e + νe, 3 5 to conserve both energy and momentum. Neutrino emis- dimensionally λA is of order R /GN . The tidal deforma- sion takes place rather with transfer of momentum to a bystander nucleon, via the modified URCA process, n + n → n + p + e− +ν ¯ , and p + n → n + n + e+ + ν 3 A simple, qualitative and instructive estimate of λ is that for e e a non-relativistic self-gravitating spherical star of uniform den- – leading to the standard, or minimal cooling scenario sity with mass MA and unperturbed radius R. A prolate de- [173–175]. For larger proton fraction, and hence Fermi − formation of the surface of the star by δR = P2(cos θ)R, momentum, the direct URCA processes, n → p+e +ν ¯e, produces a quadrupole moment, Q = (2/5)M R2, and in- − zz A p + e → n + νe, are kinematically allowed in cooling, creases the gravitational energy of the sphere by ∆Edef = 2 2 leading to a rapid or enhanced cooling scenario. Owing (3/25)GN MA /R. In addition the energy of the star in the external tidal field (5) is ∆E = −(3/5)G M M R2/R3. to the extra nucleons in the modified URCA process the tidal N A B 0 2 Minimization of ∆E = ∆Edef + ∆Etidal with respect to yields cooling rate is suppressed by a factor ∼ (T/Tf ) com- 3 5 3 = (5/2)(MB/MA)(R/R0) , so that Qzz = MBR /R0, and thus pared with the direct URCA process, typically some five 5 5 λA = R /2GN , k2 = 3/4 [163], and ΛA = 16(R/Rs) . Cf. [164] for a discussion of tidal deformation in terms of energetics. Fully relativistic calculations for stars with realistic equations of state yield much smaller Λ than predicted by this schematic calcula- 4 tion, since more mass lives at smaller radii than in a uniform The conventional geophysical definition of k2 is a factor of 2 density sphere. larger [165] than that here. 9 to six orders of magnitude, where Tf is the Fermi tem- first order hadron-quark phase transition, and another perature of the nucleons. realizing hadron-quark continuity. We then outline the BCS pairing of nucleons has important physical effects relationship between the structure of the QCD phase di- on the cooling of stars. By suppressing the density of agram and the neutron star equation of state, discussing states, the rates of neutrino emissions are reduced by a hadron-quark continuity. We review explicit construc- factor ∼ exp(−2∆/T ), with ∆ the pairing gap. With tions of unified equations of state valid at all densities pairing, processes involving pair formation or breaking, for a number of quark model parameter sets, in terms of such as n + n → [nn] + ν +ν ¯, also lead to neutrino emis- interpolating between a low density nuclear equation of sion, where the initial n are excited quasiparticles, and state at nB . 2n0 and a high density quark matter equa- [nn] indicates a condensate pair; these can become impor- tion of state at nB & 5n0 (Fig.4). In this construction tant in the regime where pairing suppresses the modified we take the crossover between the two regimes to occur URCA process. On the other hand, the heat capacity of around nB ∼ (2 − 3)n0, below which the strangeness paired matter is similarly suppressed. The net effect of density is small and various hadronic equations of state pairing on the cooling of neutron stars depends in detail exhibit very similar properties. In Sec.VII we turn to on models of the equation of state of matter and pairing, the astrophysical consequences of the unified equations as well as on their masses [171, 174]. of state, and indicate connections between effective quark “Exotic” matter, e.g., pion and kaon condensates, hy- model parameters and neutron star mass and radius ob- peron mixing as well as quark matter, can lead to en- servations. We compare several model equations of state hanced cooling via direct URCA processes. Owing to reflecting hadron-quark continuity to existing constraints the theoretical uncertainty of the possible phases inside obtained from astrophysical data and find consistency neutron stars and the uncertainties of the age and sur- with current inferences [11, 12]. Finally, in Sec.VIII face temperatures of observed neutron stars, it is not yet we summarize the impact of parameter variations on the possible at the moment to conclude whether fast cooling maximum stable neutron star mass and discuss how ob- by direct URCA processes as well as by exotic compo- servational data of massive neutron stars can provide ad- nents is taking place. In any case, the cooling of neutron ditional constraints on microscopic model parameters. In stars needs to be investigated using equations of state addition, we mention open problems, including the need compatible with observed ∼ 2M neutron stars. to develop finite temperature equations of state for mod- Accretion of matter onto neutron stars leads to ther- eling gravitational waveforms in neutron star–neutron monuclear bursts on the surface [6, 20, 150]; observation star and neutron star– black hole mergers. In Appendix of the cooling of accretion-heated neutron stars, from a A we review scaling properties of the TOV equation, and few days to a few tens of years after outbursts, is a in- in AppendixB we review effects of a repulsive quark vec- novative way to constrain neutrino luminosity and probe tor interaction on quark masses, chiral symmetry break- the outer regimes of neutron stars [176]. The basic idea ing, and pairing gaps. Lastly in AppendixC, we present is that, following the injection of heat, the neutron star the equations of state with hadron-quark crossover in a interior acts as a calorimeter. Using this approach, Cum- parametrized form – called QHC18 – useful for modeling ming et al. [177] have deduced a lower bound to the inte- neutron stars. We generally adopt units ~ = 1, and in grated heat capacity of neutron stars from observations discussing QCD will also generally take c = 1. of accretion outbursts, from which they argue against matter below 2n0 having a very low specific heat (e.g., as with a color-flavor locked quark core); this result is II. THE CRUST consistent with the present unified equations of state discussed below. The crust plays an important role in observable phenomena in neutron stars, even though its mass and thickness are relatively small. Such phenomena include ther- I. Outline mal conduction, which establishes the temperature drop between the core and surface; superfluid neutron dynam- In this review we first describe the more familiar prop- ics at densities above neutron drip, which have been in- erties of neutron stars, the crust in Sec.II, and the liq- voked to understand pulsar glitches and quasi-periodic uid nuclear matter in the outer core in Sec.III. Although oscillations; and more generally the elastic properties of these regions have been well studied for many years, open the crust, which play a role in oscillations of the star; see, questions remain, as we discuss. We then turn to describ- e.g., [20, 179–182] and references therein. Although as we ing quark matter at high density in Sec.IV, describing indicate below, the mass and especially the thickness of effective models for the quark and gluon sectors. In the the crust are insensitive to the details of the equation of following sectionV we discuss general aspects of a uni- state, the equation of state itself, as well as dynamical fied equation of state capable of connecting the low den- phenomena in the crust, remain an important problem. sity hadronic and high density quark matter regimes, and The crust can be divided into three regions: a sequence continue in Sec.VI with a more detailed analysis of two of nuclei below the surface which become increasingly separate possible unifications, one with a conventional neutron rich with depth into the star; the neutron drip 10

is simple in principle: given the pressure, P , as a function of the baryon chemical potential, µB, for the phase with nuclei and for the uniform liquid phase, the ground state for matter with a given µB is the phase with the higher pressure. The phase transition occurs when P and µB of the two phases are equal and, since nB = ∂P/∂µB, the density discontinuity across the transition is given by the difference of the slopes of the P versus µB curves for the two phases. This calculation, demanding detailed calculations of the pasta phases, is complex in reality. One simple approach to estimating the position of the bound- FIG. 6: The crust of a neutron star, between the stellar radius ary is to compare the nucleonic pressure in the liquid in- R and the radius of the liquid core Rc. The mass density and terior with the pressure of a pure neutron gas (in which properties of matter change vastly in the crust region. As the µB is the neutron chemical potential µn) at the same µB, density increases, the nuclei in the crust undergo neutron drip, and locating the baryon chemical potential where the two and then become unstable and forming various “pasta” phases. curves cross. In this comparison one holds the electron At higher density (nB & 0.4n0) the matter becomes liquid nuclear matter. density fixed, and the electron contribution to the pressures plays no role. Again the slopes of P vs. µB at this point give the densities across the transition. This pro- regime, in which the continuum neutron states in the cedure is simply equivalent to neglecting the surface and matter are occupied; and at the highest densities in the Coulomb energies of the nuclei in the phase with nuclei, crust, a sequence of “pasta” nuclei, including rods and so that its thermodynamics is determined only by the sheets, which account for essentially half the mass of dripped neutron fluid. Detailed calculations are given in the crust. Representative calculations of the equation Ref. [207]. of state in the crust, particularly below the pasta region, A second way to estimate the position of the transition include Refs. [114, 183, 184], while details of the pasta is again to start in the uniform phase and decrease the phases can be found in Refs. [185–195]. At a baryon density until matter becomes unstable to formation of small amplitude modulations of the proton and neutron density of order one half nuclear matter density, n0, the matter undergoes a first order phase transition from a densities. In the absence the Coulomb interactions, the solid crust to liquid nuclear matter. electron density remains uniform, and the conditions for stability to long-wavelength density fluctuations are first Up to densities of order 1011 g/cm3 one can use the that the partial neutron and proton compressibilites are properties of laboratory nuclei to deduce the sequence positive, of increasingly neutron rich nuclei with depth [196]. Be- yond this point, however, the details of the nuclei that are εnn > 0 and εpp > 0, (11) present, as well as the neutron drip point, at mass density ∼ 4 × 1011 g/cm3 where continuum nuclear states where ε(nn, np) is the energy per unit volume of uni- 2 are first occupied, are sensitive to the shell structure of form nuclear matter and εij = ∂ ε/∂ni∂nj = ∂µi/∂nj = very neutron-rich nuclei. The spin-orbit force, which is ∂µj/∂ni; these conditions are generally satisfied for en- critical in determining the usual closed shell structure ergy functions commonly employed. The second condi- at neutron or proton numbers 20, 28, 50, 82, . . . , is ex- tion is that pected to decrease in more neutron-rich nuclei [197]; the ε ε > ε2 . (12) details are still very much in flux, both experimentally nn pp np and theoretically [198]. When the two terms in Eq. (12) become equal, matter To determine the nature of the nuclei beyond neutron becomes unstable to long-wavelength sinusoidal proton drip one must generalize the description of isolated nuclei and neutron density waves. In nuclear matter at densi- to allow for the neutron gas outside the nuclei. Common ties . n0, εnp is negative as a consequence of the strong methods are microscopic calculations based on Skyrme attractive s-wave interaction between neutrons and pro- interactions [199, 200], on generalized liquid drop models tons. Thus the neutron density is large where the proton [113, 201], and on chiral effective field theory [57]. Even density is large. The instability argument can be ex- though the gross properties of matter in the crust are tended to finite wavelengths by including the Coulomb much better understood than they are at higher densi- interaction and surface energy [113]; the onset of insta- ties, there are still a number of uncertainties, typically bility occurs at a lower density when these effects are at the 10% level, in the structure of the nuclei owing to included. As found in Ref. [208], the density in the liq- uncertainties in the effective interactions used. The su- uid at which the instability occurs is in the range (0.075- −3 perfluid properties of the nuclei immersed in the neutron 0.088) fm ' (0.47-0.55) n0, similar to that found in fluid [200, 202–204] as well as the superfluid properties Ref. [207]; the estimate based on comparing pure neu- of the neutron fluid itself [205, 206] are also uncertain. tron matter with the matter in the liquid interior occurs Determining the position of the inner edge of the crust at a somewhat higher density [207] (but still below n0). 11

The thickness of the crust as well as its mass can be across a phase transition between different species of nu- deduced from the TOV equation (2) without having a clei, although the baryon and electron densities need not detailed knowledge of the equation of state in the crust. be continuous. We comment briefly on the situation Hartle [209] gave initial estimates of the mass and thick- when matter in the outer regions of the crust is not in ness of the crust, and we give improved versions of his full chemical equilibrium, as in accreting stars, where, arguments here [210, 211]. We focus first on the thickness e.g., the neutron star can have a layer of hydrogen above of the crust. For matter at zero temperature containing a the fully catalyzed matter. Across such a transition be- single component, or in full chemical equilibrium, one has tween elements the pressure is continuous; however, in 2 dP = nBdµB and ρc + P = µBnB; the TOV equation general, the baryon chemical potential can have a discon- can then be rewritten as tinuity, ∆µ in going from larger to smaller radii, since the baryons are bound in nuclei. For a single discontinuity, ∂ ln µ (r) G m(r) + 4πr3P (r)/c2 B N µ −µ (R) in Eq. (17) is replaced by µ −µ (R)−∆µ. In = − 2 2 . (13) c B i B ∂r c r(r − 2Gm(r)/c ) other words only the continuous changes in the chemical In evaluating the right side in the crust, we can to a potential from the edge of the core to the edge of the star good approximation replace m(r) by the total mass, Mc, determine the thickness of the crust. of the liquid core of the star, and neglect the pressure If the matter is fully catalyzed out to a pressure Pχ, 3 2 term 4πP (r)r /c compared with Mc. beyond which the star has a layer of atoms of a given When µB(r) is continuous, as in fully catalyzed matter, species (A, Z), then the thickness of this added layer is we can readily integrate Eq. (13) from the outer edge of simply the core, at radius Rc, to the surface of the star, at radius R, to find 2 Rc ∆R ' [µAZ (Pχ) − µAZ (0)] − 1 ; (18) mn Rsc 1/2 µB(Rc) 1 − Rsc/R = , (14) the chemical potential difference here is given by the µB(R) 1 − Rsc/Rc equation of state P (µ) in the added layer. where µB(R) is the chemical potential at the surface of To estimate the mass of the crust, we again neglect the 2 the star and Rsc = 2McGN /c is the Schwarzschild ra- pressure terms on the right of the TOV equation, replace dius of the core. Solving for R we find, m(r) by Mc, and integrate the resulting equation from R − R ζ Rc to R. Since the pressure vanishes at the stellar radius, c = , (15) we find Rc [Rsc/(Rc − Rsc)] − ζ Z R 2 where 2 ρ(r)c Pc = Rsc r dr 3 . (19) R r (r − Rsc) µ (R ) 2 c ζ ≡ B c − 1. (16) µB(R) The integral is dominated by the contributions from close to the inner radius of the crust, and thus we may to a 3 Both µB(Rc) and µB(R) are very close to the nucleon good approximation replace the factor r (r − Rsc) in the mass. For example in [113], µB(Rc) − mn ' 15 MeV, 3 denominator of the integrand by Rc (Rc −Rsc). The crust while at the stellar surface, µB(R) − mn ' −8 MeV, mass, the binding energy of nucleons in the outermost nuclei (57Fe, ideally). Thus ζ ' 2(µ − µ (R))/m c2, and for Z R i B n 2 the numbers just cited, ζ ' 0.05. On the other hand, Mcrust = 4πr ρ(r)dr, (20) Rs Rsc/Rc ∼ 0.3M/M , so that to a good approximation the ζ can be neglected in the denominator of Eq. (15), is then approximately and we find Pc Rc R − Rc µB(Rc) − µB(R) Rc − Rsc 3 Mcrust ' 4πRc 2 − 1 , (21) ' 2 2 ; (17) c Rsc Rc mnc Rsc this equation gives the thickness of the crust in terms of again not dependent on the details of the equation of the radius and Schwarzschild radius of the core, and the state in the crust. For a characteristic value of Pc ' 0.5 3 −2 baryon chemical potential difference between the inner MeV/fm , we find Mcrust ∼ 10 M . and outer radius of the crust. Remarkably, the thickness The baryon density at the transition, in the range of the crust is insensitive to the details of the equation of (0.47-0.55) n0 [208], is uncertain. The related uncertainty state in the crust. For the numbers taken above, we find in Pc at the transition leads, as one sees from Eq. (21), a crust thickness of order 0.5 km. to an uncertainty in the mass of the crust at a level ∼ The calculation above assumes that the baryon chem- 8%. Although the lattice appears to be stable [212, 213], ical potential is continuous in the crust, as it must be an additional subtlety in the physics of the crust is the for fully catalyzed matter. In chemical equilibrium the question of whether protons in the crust can also drip neutron chemical potential and pressure are continuous out of the nuclei [207]. 12

III. LIQUID NUCLEAR MATTER IN THE kinetic energy to be that of the free gas at given nB and OUTER CORE x, with possible effective mass corrections [52], and apply a quadratic interpolation only between the interaction- The historic approach to describing nuclear matter in energy density of pure neutron matter and symmetric the outer region of the core is first to calculate the energy nuclear matter [52, 208]. To determine the proton frac- densities of both symmetric nuclear matter and neutron tion in the liquid interior, one then imposes the condition matter and then interpolate between them to describe of beta equilibrium (22). Typically the proton fraction matter at the finite proton fraction expected for matter just within the liquid interior varies from about ∼ 3% in beta equilibrium between the protons, electrons and at the crust-core boundary to ∼ 5% at n0. Initial direct neutrons, viz., Monte Carlo calculations of asymmetric nuclear matter are given in Ref. [218]. µn = µp + µe, (22) It should be noted that BCS pairing of nucleons in the where the rest masses are included in the chemical po- outer regions of the core does not produce significant cor- tentials µ . As described briefly in the introduction, the rections to the equation of state there, since energy gaps i ∆ are ∼ MeV, so that condensation energies, of order basic calculational method is to first determine nucleon- 2 nucleon potentials from scattering data supplemented ∆ /EF , are small compared with the Fermi energy, EF . by three-body interactions, and to then employ large Such pairing is not taken into account in the APR equa- scale computational methods to solve the many-body tion of state. Uncertainties in the equation of state arise Schrödingerequation in the presence of these potentials; from uncertainties in the three-body interactions with in- a representative such calculation is that of Akmal, Pand- creasing density [208], and also from the onset of neutral pion condensation, which APR finds to occur at ∼ 0.2 haripande, and Ravenhall (APR) [52]. It is common −3 −3 to employ an energy density of symmetric matter con- fm in matter in beta equilibrium, and at ∼ 0.3 fm in matter in symmetric nuclear matter. The transition sistent with the empirical nuclear binding energies and 5 compressibilities, thus reducing theoretical uncertainties. to the pion condensed state found in APR is first order. A further approach is via chiral effective field theory However, APR did not consider spatially non-uniform [214], which provides a systematic development of two neutral condensates (e.g., ∼ cos kz for small amplitude and higher body interactions [208, 215, 216]. condensation), nor did they examine the order parame- Various microscopic calculations of (pure) neutron ter of the condensed state; as a consequence the energy matter, reviewed in Ref. [217], are in relatively good of the “pion condensed phase” reported in APR is an upper bound to the energy of the pion condensed state. agreement up to nuclear matter density n0, where the dominant two-body interaction between nucleons is well The transition in APR is to a uniform phase character- characterized by nucleon–nucleon scattering data, Zero ized by a large tensor correlation length and enhancement range three body interactions in neutron matter give no of spin-isospin correlations [72]. contributions to the energy density, as a consequence of the Pauli principle, which prevents three neutrons being IV. QUARK MATTER AT HIGH DENSITY at the same point, because two of them must be in the same spin state. However, realistic three-body interactions have a finite range, and thus at densities of interest As an introduction to more detailed applications in in neutron stars, they give comparable contributions to Sec.VC, we begin in this section by discussing the ele- the energy densities of neutron matter and symmetric mentary physics of quark matter, the expected form of nuclear matter, and greatly stiffen the neutron matter neutron star matter at densities well beyond n0. equation of state. While the limit of validity of nuclear matter calculations based on interacting nucleons is uncertain, it is not unreasonable to use them up to a density A. Noninteracting quark matter ∼ 2n0. To interpolate between pure neutron matter and sym- The simplest model of quark matter takes into account metric nuclear matter, one can to first approximation the bare quark kinetic energy density, εK , and the bag take the energy density of uniform nuclear matter to be constant, B, which is the energy density difference be- a quadratic function of the neutron excess, δ = 1 − 2x, tween the non-perturbative vacuum in QCD and the per- where x = np/nB is the proton fraction; then, turbative vacuum. The zero of energy is commonly taken to be that of the non-perturbative vacuum, compared to E(nB, x) ' E(nB, 0) − 4x(1 − x)nBS(nB), (23) which the perturbative vacuum has a positive energy, B. where S(nB) = (E(nB, 0) − E(nB, 1/2)) /nB is the density-dependent symmetry energy. This interpolation leads typically to a value of S(n0) at nuclear saturation 5 This result is consistent with a general argument of Dyugaev density close to the empirical symmetry energy for nu- [219, 220] that as condensation is neared the pion-pion scattering clei with roughly equal numbers of neutrons and protons, amplitude in the medium changes from repulsive to attractive ≈ 32 MeV [208]. An alternative approach is to take the owing to exchange of soft pion modes. 13

For illustration, let us consider a gas of massless quarks with Nf ﬂavors. At low temperature the quarks form a degenerate Fermi sea, with quark density

Z pF 3 3 d p pF nq = 2NcNf 3 = NcNf 2 , (24) 0 (2π) 3π where Nc = 3 is the number of quark colors, the 2 is for spin, and pF is the quark Fermi momentum. The bare quark kinetic energy density is then FIG. 7: Chiral symmetry breaking via quark-antiquark pairing. Z pF 3 d p 3NcNf 4 Condensation of pairs opens a gap M in the quark dispersion εK = 2NcNf 3 |p| = 2 pF , (25) 0 (2π) 4π relation, changing the structure of the Dirac sea. The energy density of the symmetry-broken Dirac sea is smaller than that of 4 Then the total energy density is ε = εK + B. Since the symmetric sea by B ∼ ΛQCD (see text). the quark chemical potential is µq = ∂ε/∂nq = pF , the baryon chemical potential µB = 3µq, the pressure and the energy density read

4 4 P (pF ) = apF − B , ε(pF ) = 3apF + B, (26) 2 where a = NcNf /4π . These expressions give the equation of state which is valid for non-interacting massless quarks at high density. The bag constant, B, is the diﬀer- ence in the energy densities of the perturbative vacuum – devoid of all particles and condensates – and the non- perturbative vacuum, which is the true ground state of QCD, including chiral and gluon condensates. FIG. 8: Quark-antiquark pairing at high baryon density. In With this equation of state, the maximum mass of a creating a hole in the Dirac sea, the quark in the Dirac sea must, hypothetical quark star made of free quarks, for Nc = by the Pauli principle, be outside the Fermi sea. Nf = 3, scales with B [61, 221] (see also Appendix A) as

155 MeV 2 Mmax ' 1.78 M , (27) C. Condensates in the QCD vacuum and quark B1/4 matter while the corresponding radius scales as The strong interactions of quarks and gluons in QCD 2 155 MeV cause a variety of condensation phenomena. Fundamen- R ' 9.5 km. (28) B1/4 tal is the formation of a chiral condensate, made of quark-antiquark pairs with different chirality, e.g., a left- handed (i.e., spin antiparallel to momentum) quark and B. The bag constant right-handed antiquark (Fig.7). The approximate chiral symmetry of QCD, as a consequence of which left Various estimates of B have been made, for a variety of and right handed quarks comprise essentially indepen- physics models and choices of which energies to include dent sectors, is broken by the chiral condensate because in B (and which to include in the quarks). As a re- it couples quarks of different chiralities. Since quarks sult of these variations, comparison of the precise values and antiquarks are bound in the condensed ground state, obtained in different models is not generally meaningful. the creation of a quark excitation requires the break- For example, in the early MIT bag model, B ' (145-155) ing of a pair, and the associated energy cost; as a con- MeV4 ' (60-80) MeV/fm3. The bag constant calculated in the NJL model (see Eq. (57) below) is ' (218 MeV)4 = 296 MeV/fm3 [101]. On the other hand, Novikov et al. [226], in a QCD based calculation, find B ' (250-300 MeV)4 ' (500-1000) MeV/fm3, an order of magnitude larger than the MIT value. We note here that it may not be possible to calculate the properties of the gluons, at least in the density range relevant for neutron stars, by applying perturbative QCD, i.e., the gluons might remain non-perturbative. The properties of such matter FIG. 9: Quark-quark pairing, which leads to color supercon- have been discussed in the context of quarkyonic matter ductivity at high baryon density. The condensation of pairs opens [227], discussed in Sec.VE. a gap ∆ near the quark Fermi surface. 14 sequence, a quark acquires an effective, or constituent, is given in Eq. (34), below. quark mass, M. This mass, dependent on the strength of the QCD interactions, has a characteristic magnitude, ΛQCD ∼ 200 MeV, the dynamical scale of QCD, much D. Nambu–Jona-Lasinio model for interacting quarks larger than the bare, or current, quark mass, mq, entering the QCD Hamiltonian. The emergence of the chiral condensate changes the structure of the Dirac sea, giving rise Owing to the inability of lattice gauge theory cal- to a non-perturbative QCD vacuum. The difference in culations to describe cold matter at finite baryon den- energy density between the (perturbative) chirally sym- sity, due the fermion sign problem [81], one needs to metric and (non-perturbative) symmetry-broken Dirac adopt phenomenological models of interacting quarks in seas, with different effective masses Meff , at zero tem- order to describe dense quark matter in neutron star perature and baryon chemical potential formally defines cores [47, 108, 109, 144–147]. We discuss here the the bag constant: frequently employed Nambu–Jona-Lasinio (NJL) model [139, 141, 142], which replaces the full QCD interactions B ≡ ε(Meff = mq) − ε(Meff = M). (29) with effective quark-quark interactions, while at the same time suppressing explicit gluonic degrees of freedom. In At high baryon density, quark-antiquark pairing is no this subsection we write down the NJL model and discuss longer energetically favored in the presence of the quark the physics of the effective interactions.6 In describing Fermi sea, Fig.8, since to create an antiquark (a hole in the NJL model we adopt standard Dirac notation with the Dirac sea), the quark originally occupying the Dirac the Minkowski metric g = diag(1,-1,-1,-1), and γ0 and sea must now occupy a previously unoccupied high en- µν γ both Hermitian. ergy state outside the Fermi sea. One could imagine at 5 The Lagrangian of the three-flavor NJL model is first sight that chiral symmetry would then be restored. µ 0 (4) (6) However, in the presence of a quark Fermi sea, pairings L = q(γ pµ − mˆ q + µqγ )q + L + L , (30) employing the degrees of freedom near the Fermi sur- where q is the quark field operator with color, flavor, and face become possible, and these can continue to break Dirac indices,q ¯ = q†γ0,m ˆ the quark current mass ma- chiral symmetry [228, 229]. Of particular importance is q trix, µq the (flavor dependent) quark chemical potential diquark pairing, in which two quarks (or two quark-holes) (4) (4) (4) (4) (6) (6) (6) near the Fermi surface are paired as electrons are Cooper and L = Lσ + Ld + LV and L = Lσ + Lσd are paired in an ordinary superconductor, Fig.9. The di- four and six-quark interaction terms, chosen to reflect the quark pairs, macroscopic in number, form a diquark con- symmetries of QCD. densate, and as a condensate of paired electrons gives rise The first of the four-quark interactions, a contact in- to electromagnetic superconductivity, a condensate of di- teraction with coupling constant G > 0, quarks, which has color charges, gives rise to color super- 8 (4) X 2 2 † conductivity. A quark excitation requires the breakup of Lσ = G (qτjq) + (qiγ5τjq) = 8Gtr(φ φ),(31) a condensed pair, costing energy 2∆, where ∆ is the pair- j=0 ing gap. Condensation reduces the energy density of the describes spontaneous chiral symmetry breaking, where system by the energy gain for a single pair, ∆, times the 2 τj (j = 0,..., 8) are the generators of the flavor-U(3) phase space available for such pairs, ∼ 4πpF ∆, where pF symmetries, and in Eq. (31), is the quark Fermi momentum. This energy reduction, 2 2 φ = (q )j (q )i (32) ∼ ∆ pF plays a very important role in quark matter ij R a L a equations of state, as we discuss below. The modeling of is the chiral operator with flavor indices i, j (with sum- these effects will be discussed in subsec.IVD. mation over the color index a); the right and left quark The diquark pairing interaction is most attractive in 1 chirality components are defined by qR,L = 2 (1 ± γ5)q. the color-antisymmetric, flavor-antisymmetric, and spin- The second of the four-quark terms describes the scat- singlet channel. Flavor asymmetry implies that quarks in tering of a pair of quarks in the s-wave, spin-singlet, a pair must have different flavors. At densities relevant flavor- and color-antitriplet channel; this interaction to neutron stars, this complicates the pairing, because leads to BCS pairing of quarks: the (u, d, s)-quarks have different quark masses and elec- (4) X T T tric charges, introducing a considerable imbalance in the Ld = H qiγ5τAλA0 Cq q Ciγ5τAλA0 q size of their Fermi seas. Many possibilities for the pre- A,A0=2,5,7 ferred pairing structures have been discussed (reviewed T T + qτAλA0 Cq q CτAλA0 q , in [96]), but here we discuss only the simplest candidates; † † two flavor, or 2SC pairing, in which u- and d-quarks = 2Htr(dLdL + dRdR), (33) pair at chemical potentials not large enough for significant strangeness to appear, and the color-flavor-locked, or CFL phase, in which the chemical potential is suffi- 6 Readers not well acquainted with the effective theories of QCD ciently high that the u, d, and s all quarks participate in described here can skip the details of this subsection and continue the pairing. The detailed structure of these condensates on to Sec.V. 15

0 with H > 0. Here τA and λA0 (A, A = 2, 5, 7) are the with matrices antisymmetric generators of U(3) flavor and SU(3) color, respectively, and (R1,R2,R3) ≡ (τ7λ7, τ5λ5, τ2λ2) . (41) The three diquark condensates correspond to (ds, su, ud) (d ) = (q )jC(q )k (34) L,R ai abc ijk L,R b L,R c quark pairings, respectively. In the 2SC phase, only ud- are diquark operators of left- and right-handed chirality, pairs condense, while in the CFL phase all (ds, su, ud)- with C = iγ0γ2 the charge conjugation operator. The pairs are present. diquark pairing interaction leads as well as an attractive The inverse of the mean field single particle propaga- correlation between two quarks inside confined hadrons tors can be read off from the mean field Lagrangian, and, in constituent quark models, plays a role in the ob- γν k − Mˆ +µγ ˆ 0 γ ∆ R served mass splittings of hadrons [233–236]. This interac- S−1(k) = ν 5 k k , − γ ∆∗R γν k − Mˆ − µγˆ 0 tion, in weak coupling, arises from single gluon exchange; 5 k k ν however at the densities of interest in neutron stars, the (42) non-linearities of QCD prevent direct calculation of this where the effective mass matrix is diagonal, with ele- interaction, and so one must treat it phenomenologically. ments In addition 0 K 2 (4) µ 2 Mi = mi − 4Gσi + K|ijk|σjσk + |di| , (43) LV = −gV (qγ q) , (35) 4 with gV > 0, is the Lagrangian for the phenomenolog- while the three diquark pairing amplitudes, ical vector interaction, as in Eq. (58), which produces K0 universal repulsion between quarks [230]. ∆k = −2dk H − σi , (44) The six-quark interactions represent the effects of the 4 instanton-induced QCD axial anomaly, which breaks the and the effective chemical potential matrix, U(1)A axial symmetry of the QCD Lagrangian. The resulting Kobayashi-Maskawa-’t Hooft (KMT) interaction µˆ = µq − 2gV nq + µ8λ8 + µQQ, (45) leads to an effective coupling between the chiral and di- are color and flavor dependent. quark condensates of the form [237, 238]: The inverse propagator (42) is a 72×72 matrix at each momentum, whose eigenvalues, can be calculated by L(6) = −8K(det φ + h.c.), (36) j σ numerical inversion [144]. The eigenvalues are four-fold (6) 0 † Lσd = K (tr[(dRdL)φ] + h.c.), (37) degenerate (2 for spin times 2 from the Nambu-Gor’kov pairing structure). The single particle contribution to 0 where K and K are positive constants. Provided that the thermodynamic potential density, K0 ' K (which one expects on the basis of the Fierz transformation connecting the corresponding interaction Ω(µq,T ) = ε − T s − µqnq, (46) vertices) the six-quark interactions encourage the coexis- which is the negative of the pressure, P , is then tence of the chiral and diquark condensates. 18 Z Λ 3 X d k ∆j Ω = −2 T ln 1 + e−|j |/T + , single (2π)3 2 E. Mean field equation of state j=1 (47)

Having reviewed the structure of the NJL model, we free free now discuss its application to constructing the equation where ∆j = j − j , with j the eigenvalues in the of state for dense quark matter. For simplicity, we re- non-interacting quark system; here Λ is an ultraviolet strict our considerations to mean-field theory. The quark cutoff. The dependence on µq is hidden in the eigenvalues density is j. As in mean field treatments, “condensate” terms, here 3 0 X † X 2 K 2 nq = hqi qii , (38) Ω = 2Gσ + H − σ |d | , cond i 2 i i i=u,d,s i=1 2 with an implicit sum over color and spin. Similarly the −4Kσ1σ2σ3 − gV nq (48) flavor-dependent chiral condensate is appear in Ω (as a consequence of avoiding double count- ing). σ = hq q i , (39) i i i While the quark matter thermodynamic potential is bare with i = u, d, s; with our conventions σ is generally neg- formally Ωq = Ωsingle +Ωcond, one must further choose ative. The diquark mean fields are the “zero” of Ω so that it vanishes in the vacuum, µq = T = 0. This choice is important because the ab- T dj = hq Cγ5Rjqi , (40) solute energy, not just energy differences, directly enters 16

† † in the general relativistic TOV equation (2) for neutron densities except n3 = hq λ3qi and n8 = hq λ8qi automat- star structure. The correctly normalized quark-matter ically vanish. Thus, including the term [145, 241–243] thermodynamic potential is thus † † L3,8 = µ3q λ3q + µ8q λ8q, (53) bare bare Ωq(µq,T ) ≡ Ω (µq,T ) − Ω (µq = T = 0). (49) q q is suﬃcient ensure color neutrality. In addition one must include eﬀects of leptons – electrons and muons – that may be present in the matter; the τ lepton’s large mass prevents it from playing a role G. Minimizing the thermodynamic potential in dense matter. The Lagrangian for the leptons is

X ν The total thermodynamic potential is Ω = Ωq + Ωl. Ll = ψl(γ pν − ml)ψl, (50) The thermodynamic state of the system is determined l=e,µ by minimizing the free energy with respect to the six condensates {σi, dk} and the quark density nq, under with ψl the lepton field operator and ml the lepton mass. The lepton contribution to the thermodynamic potential the conditions of electrical and color charge neutrality is expressed by the conditions Z d3k ∂Ω X X −(El+λµQ)/T n = − = 0, (54) Ωl = −2T ln 1 + e , j (2π)3 ∂µj l=e,µ λ=± (51) with j = Q, 3, 8. In addition the condensates are determined by the six “gap equations,” with E = pk2 + m2. l l ∂Ω ∂Ω 0 = − = − , (55) ∂σi ∂di F. Electric and color neutrality constraints where all derivatives are taken at fixed quark chemical potential. Finally the quark density is The matter comprising neutron stars must be both ∂Ω electrically and color neutral, since any long range charge nq = − . (56) or color imbalance would be prohibitively expensive ener- ∂µq getically. To achieve electrical neutrality in quark matter In order to construct the equation of state, one can solve one must allow for the possibility of electrons and muons these 10 equations self-consistently, using the method being present. outlined in [244], to construct first the energy density In modeling the equation of state, neutrality is sim- ε(µq,T ) and the pressure P (µq,T ), then finally the de- plest to achieve by introducing the charge chemical po- sired P (ε). tential, µQ, and color chemical potentials, and tuning these chemical potentials to keep the color and charge densities zero. The charge chemical potential couples to H. Parameter sets the charge density in the Lagrangian through a term   The NJL model for dense quark matter contains two † X † LQ = µQ q Qq − ψ ψl , (52) distinct sets of parameters: {Λ, mu, md, ms, G, K} and  l  0 l=e,µ {gV ,H,K }. The first set is fixed by matching to QCD vacuum phenomenology. To be specific, we consider pri- where Q = diag(2/3, −1/3, −1/3) is the quark charge marily the set by Hatsuda and Kunihiro (HK) [142] (Ta- operator in flavor space. bleI), which gives the vacuum effective masses for the Dependent on the particular diquark pairing scheme, light quarks, Mu,d ' 336 MeV, and the strange quark, the diquark pairing interaction, (33), can lead to a viola- Ms ' 528 MeV. The NJL model with this set of pa- tion of color neutrality. For example, pairing of only red rameters yields (e.g. Ref. [101]) the “bag constant” [cf. and green quarks (as in the 2SC phase) leads to a de- Eq. (29)], creased energy per particle of these colors, and thus an B ≡ [ ε(M = m ) − ε(M = M)] increase of red and green quarks compared with unpaired NJL eff q eff T =µq =0 blue quarks, leaving the system with a net anti-blue color ' (218 MeV)4 = 296 MeV/fm3 . (57) density. In realistic QCD this color imbalance is exactly cancelled by the appearance of a non-zero coherent gluon The second set of parameters is not well fixed by QCD field [239, 240]. However, this mechanism is outside the vacuum phenomenology, but it is natural to expect their scope of the NJL model, and thus one must restore color values to be characterized by the QCD momentum scale neutrality by hand [239–243], most generally by introduc- ΛQCD, in the absence of anomalous mechanisms to sup- ing eight independent color chemical potentials. For the press these couplings. In addition, the K0 terms couple diquark pairing structures discussed in Eq. (40) all color the diquark condensate to the chiral condensate, and thus 17

TABLE I: Three common parameter sets for the three-flavor phase maximizes the pressure at given µB, and thus in NJL model: the average up and down bare quark mass mu,d, the presence of competing phases the one with the higher strange bare quark mass ms, coupling constants G and K, and pressure is favored. In addition, a kink, or discontinuous three-momentum cutoff Λ [140, 142, 143]. change of slope, in P (µB) indicates a first order transi- 2 5 tion (illustrated in Fig. 18 below), while a sudden change Λ (MeV) mu,d (MeV) ms (MeV) GΛ KΛ in curvature (with continuous slope) indicates a second HK 631.4 5.5 135.7 1.835 9.29 RHK 602.3 5.5 140.7 1.835 12.36 order transition. LKW 750.0 3.6 87.0 1.820 8.90

A. The stiffness of the quark matter equation of the value of K0 is strongly correlated with the values of state gV and H; most, if not all, of the effects of its variation can be absorbed into variations of gV and H. In We ask now the effects of the bag constant, pairing, the present analysis we assume as a first orientation that and the vector repulsion on the stiffness of the quark the values of these coefficients are of the same order of matter equation of state. To see the physics we write the magnitude as the first set of parameters. total energy density in the schematic form

4/3 2/3 2 ε = AnB + B − CnB + DnB, (58) Attempts have been made to estimate the magnitude where the first term is the kinetic energy for massless as well as the medium dependence of gV in the NJL model quarks (we ignore corrections to the kinetic energy from 7 (with and without coupling to a Polyakov loop ) by us- finite quark masses), B is the bag constant, C ∝ ∆2 [231] ing lattice QCD inputs or other phenomenological con- measures the energy contribution of pairing, and D ∝ gV siderations, see e.g. [138, 230, 249, 250]. In the follow- measures the strength of the density-density repulsion, ing we describe only the situation with g constant in 2 V gV nq [230]. We assume here that B and ∆ are den- the regime where quarks are relevant for neutron stars, sity independent for simplicity√ (although at high density, nB 5n0, since dependence of gV on the baryon density −5 2 & ∆ ∼ µqg exp(−3π /g 2) [232], where g is the scale arises primarily from the modifications of the gluons by dependent QCD coupling constant); then differentiating the matter, which is argued to be small [251]. (58) yields the baryon chemical potential, Detailed results for the equation of state in the quark phase are given in Sec.VI and in AppendicesB andC. ∂ε 4 1/3 2 −1/3 µB = = AnB − CnB + 2DnB, ∂nB 3 3 (59) V. CONSTRUCTING THE NEUTRON STAR EQUATION OF STATE: GENERAL and the pressure, CONSIDERATIONS 2 ∂(ε/nB) 1 4/3 1 2/3 2 P = nB = AnB + CnB + DnB − B. Having laid out the basic physics in both the lower den- ∂nB 3 3 sity hadronic regime and the higher density quark regime, (60) we turn now to an examination of the general character- We note that in general a term in the energy density of istics of the equation of state, paying particular atten- γ γ tion to its stiffness and the corresponding implications the form αnB, leads to a term in the pressure (γ−1)αnB. for neutron star structure. The most convenient ther- To see how the pairing interaction and the repulsive inter- modynamic potential for studying the equation of state action affect the stiffness of the equation of state, P (ε), is the pressure, P , as a function of the baryon chemi- that is, the magnitude of the pressure for a given energy density, we must evaluate their pressure at a fixed energy cal potential, µB. The pressure must be a continuous density. Thus we ask how the pressure varies as α is var- function of µB, and since the baryon density is given by ied at fixed energy density. In order to keep ε fixed with nB = ∂P/∂µB, the pressure is monotonically increas- 2 2 varying α one must vary the density nq; thus ing; furthermore the curvature, ∂ P/∂µB = ∂nB/∂µB, must be positive, else the system is unstable against ∂P ∂P ∂ε 2 2 γ density fluctuations. The thermodynamically preferred = − cs = (γ − 1 − cs)nB, (61) ∂α ε ∂α nB ∂α nB

2 where cs = ∂P/∂ε|α is the square of the thermodynamic sound speed. 7 The Polyakov loop [123, 126, 246], which is essentially an order Since for the repulsive interaction Dn2 , one has γ = 2 parameter for confinement, plays an important role in determin- B ing the structure of the QCD phase diagram at finite tempera- and α = D, we conclude that increasing the strength of ture [144, 145]. The effective potential for this order parameter this interaction will always lead to higher pressure and fitted to the lattice QCD data at zero density [247, 248], vanishes thus a stiffer equation of state, as long as the causal- in the zero temperature limit. 2 ity condition, cs ≤ 1 is satisfied. We also observe that 18

P P 2 P1

P* stiffer softer

µB ε 2* ε 1* FIG. 10: Graphical analysis of the equation of state, P (µB ). The slope of the tangent line a given point (µB∗,P∗), is the baryon density, nB∗ = ∂P/∂µB |µB∗ , and its intercept on the P axis is the negative of the energy density ε . ∗ FIG. 11: Comparison of two equations of state in which the pressure curves have the same shape, but P2 is shifted toward lower chemical potential, relative to P1. The equation of state increasing a density-independent pairing gap, ∆ (corre- P2 is stiffer than P1 because ε1∗ < ε2∗ . A reduction of the bag sponding to γ = 2/3, with α = −C) increases the pres- constant, B, would lead precisely to such a shift from P1 to P2. 1 2 2/3 sure: ∂P/∂C|ε = 3 + cs nB , and stiffens the equation of state. We see from this exercise that the effect of interactions on the stiffness of the equation of state depends not only on the sign of the interaction, but also on the power of the P density it involves. Generally, as indicated in Eq. (61), P1 P3 increasing the strength of a repulsive interaction with 2 γ > 1+cs stiffens the equation of state, as does increasing P* 2 the strength of an attractive interaction with γ < 1 + cs. softer stiffer A larger value of the bag constant, for which γ = 0, softens the equation of state. µB ε B. Graphic determination of the stiffness of the 3* equation of state ε1* Figure 10 demonstrates how to determine graphically the relative stiffness of an equation of state from its P (µB) curve [119]. The slope of the curve at a given FIG. 12: Increasing the vector repulsion decreases the pressure point (µB∗,P∗) is the baryon density, nB, for the spec- at fixed µB and decreases the slope at fixed P . The result is that ified µB. Thus, from the zero temperature thermody- P3, with larger vector repulsion, is a stiffer equation of state than P . namic identity, P = µBnB − ε, we find that the tangent 1 curve intercepts the P axis at the point −ε∗, the negative of the energy density at chemical potential µB∗.A stiff equation of state is characterized by a large pressure for given energy density ε (or mass density ρ = ε/c2), or equivalently by a small energy density for given pressure. where we regard nB as a function of µB. Thus increasing The smaller the slope of P at given µB, the stiffer the the bag constant (γ=0) decreases the pressure at fixed equation of state. Similarly, the smaller is µB for a given µB, and softens the equation of state, as illustrated in P and slope, as illustrated by the two curves P1 and P2 Fig. 11. Similarly, increasing the strength of the repulsive in Fig. 11, the stiffer is the equation of state. To see the vector interaction (γ = 2) leads to a lower slope and a effects of the various terms in Eq. (58) graphically, we stiffer equation of state, as shown by the shift from P1 to γ P3 in Fig. 12. On the other hand increasing the pairing write the terms generically, as before, as αnB and note the thermodynamic identity (cf. Eq. (61)), strength increases the pressure at fixed µB, and at the same time it increases the slope of P vs. µB, with the ∂P ∂P ∂µ net effect of stiffening the equation of state, as illustrated B γ = − nB = −nB, (62) by the shift from P to P in Fig. 13. ∂α µB ∂α nB ∂α nB 1 4 19

matter at low density. (In the strange matter hypothe- sis, where three-flavor quark matter is assumed more sta- P P4 ble than nuclear matter at low density [61, 62, 252, 253] P 1 the equation of state would be of the form P2.) The relatively stiff quark equation of state P would also be P* 3 rejected in constructing a hybrid equation of state be- stiffer softer cause it does not intersect the hadronic pressure PH , and therefore would not in this construction be considered an µ acceptable model of quark matter at high density. Thus, B large classes of stiff quark matter equations of state – precisely those consistent with stable massive neutron stars ε ε 4* – must be rejected in a conventional hybrid equation of 1* state construction. Quark equations of state consistent with such a construction are generally soft, so that within the conventional description, one concludes that massive FIG. 13: Increasing the pairing increases the pressure at fixed neutron stars can at most have a small quark matter core µB and increases the slope at fixed P . The net result is that P4, (e.g., see Fig. 18 of Ref. [52]); Ref. [254] summarizes qual- with larger pairing, is stiffer than P1. itatively possible conditions on the quark matter equation of state that would support neutron stars of two solar masses. P In the conventional construction of a hybrid equation P5 P1 of state one looks for an intersection of the pressures of the hadronic and quark matter equations of state as functions of the baryon chemical potential (or equivalently PH the energy per baryon, ε/nB, as functions of 1/nB, the volume per baryon), and makes a Maxwell construction P3 to equate pressures and baryon chemical potentials between the two phases. An implicit assumption in this procedure is that both equation of states are reliable in ~ M µ µ N H→Q B the vicinity of the intersection. However, the typical intersection, corresponding to nB ∼ (2−5)n0, is exactly in FIG. 14: Conventional construction of a hybrid equation of the region where the hadronic equation of state becomes state from independent hadronic (PH ) and quark (P1) equations uncertain due to many-body forces and hyperon forces, of state. In such a construction a first order phase transition oc- and the quark equation of state becomes uncertain due curs at µH→Q, and stiff quark equations of state, e.g., P5 (cf. P2 in Fig. 11 and P4 in 13) and P3 (cf. Fig. 12), which are incom- to the effects of confinement. The unified approach, to patible with such a construction are excluded. which shortly we turn in subsec.VF, allows one to relax the requirement on the intersection and obtain a description of dense matter which permits certain classes of stiff C. Hybrid equations of state equations of state with quark matter. The fundamental problem with conventional hybrid Hybrid equations of state assume that matter can be equation of state constructions is that they assume that in one of two distinct phases, hadronic or quark. The both hadronic and quark matter equations of state are favorable phase, hadronic at low densities and quark at reliable near their intersection. When the intersection oc- high densities, has the higher pressure at fixed chemical curs at small chemical potential, one implicitly assumes potential, with a first order transition between the two that the quark pressure at low density is reliable; but no phases. Figure 14 shows the construction of a hybrid viable quark model calculations are available in the low equation of state in P vs. µB. To ensure the existence density regime, due to the difficulty of modeling critically of hadronic matter at low density one demands that the important confining effects. Similarly, when the intersec- quark pressure, PQ, intersects the hadronic pressure, PH , tion occurs at large chemical potential, one accepts the from below, while to ensure the existence of quark matter hadronic equation of state at high density, where many- at high density, one requires that PQ be greater than body forces are rapidly enhanced, rendering the equation PH above the chemical potential where the two curves of state seriously uncertain, even if one continues to as- intersect. sume that hadronic degrees of freedom correctly describe In Fig. 14 we also show two quark equations of state, the matter. One cannot reliably compare hadronic and P3 and P5, which are stiffer than P1 (P5 corresponds to quark matter pressures across the entire density domain. either P2 in Fig. 11 or P4 in Fig. 13, while P3 corresponds These considerations suggest that the conventional hy- to the curve in Fig. 12). The conventional hybrid con- brid construction places overly stringent requirements on struction rejects P5 because it does not allow hadronic the form of quark matter equation of state by accepting 20 the predictions of hadronic models above their regimes of validity. The unified approach, to which shortly we turn P/T4 + hadronic in subsec.VF, allows one to relax these requirements interactions and obtain a description of dense matter which permits certain classes of stiff quark equations of state. + confining pQCD gas effects

D. Thermodynamics of finite temperature QCD at zero baryon density HRG lattice QCD T To further motivate the unified construction of the ~T ~2-3T equation of state of cold dense matter, and the emerg- c c ing role of hadron-quark continuity in the phase diagram, we briefly consider implications of the lattice re- 4 sults [129, 130, 255, 256] for the structure of the equation P/µB + many-body of state at zero chemical potential, P (T ) = T s − ε, with forces s the entropy density, above and below the crossover at zero chemical potential, Fig.3. These calculations in- quark matter dicate that matter composed of light quarks with finite masses undergoes a rapid continuous crossover with increasing temperature from a hadronic to a quark-gluon Interpolated phase at the pseudocritical or “deconfinement” tempera- + confining hadronic matter ture Tc ∼(150-155) MeV, with smooth restoration of ap- forces (with 3-body) proximate chiral symmetry. (Were the u, d, s-quarks all ~ M massless, then at very high temperatures chiral symmetry N µB would be completely restored in a first order phase transition, distinguishing the symmetry broken phase and the symmetry restored phase (see e.g., [47]). FIG. 15: Schematic representations of the pressure of dense matter at finite temperature and zero baryon chemical potential At zero baryon density and low temperatures, matter (upper panel) and finite baryon density and zero temperature is well described by the non-interacting hadron resonance (lower panel). The bold lines represent model predictions within gas (HRG) model [129, 130, 257]. However, as the tem- the domain of applicability of each picture, while dotted lines show extrapolations beyond the domains of validity. Upper panel: perature approaches Tc the hadron resonance gas model pressure of the hadron resonance gas (HRG), the perturbative strongly overestimates the pressure compared with lat- quasiparticle gas of quarks and gluons (pQCD gas), and lattice tice calculations. Physically, this discrepancy arises from QCD calculations. Lower panel: pressure of hadronic matter with the large overlap of thermally excited hadrons, whose in- two- and three-body forces, and of a deconfined quark gas. The teractions can no longer be neglected. On the other hand, interpolated pressure is constructed using the hadronic gas pressure below µBL and the quark gas pressure above µBU . at temperatures & (2 – 3) Tc, the matter can be described reasonably well by a weakly interacting quasiparticle picture of quarks and gluons, a perturbative or pQCD gas.8 However, with decreasing temperature the gas pressure the perturbative QCD gas models [258]. Matter in this calculated by taking interaction effects into account per- region is a strongly correlated quark-gluon plasma [131– turbatively is well above that calculated on the lattice, 135]. The message from understanding finite tempera- since the lack of confining effects allows an artificially ture matter at n = 0 is that were one simply to adopt enhanced population of quarks and gluons without trig- B the resonance gas picture at low temperature and the gering the kinetic energy cost of their confinement into perturbative QCD picture at high temperature and then hadrons or glueballs. The behavior of the extrapolated apply a Maxwell construction to find the phase transi- pressures are illustrated in the upper panel of Fig. 15. tion between the two phases, the transition would nec- While we understand both the low and high temper- essarily be first order and the resulting hybrid equation ature limits qualitatively, the intermediate temperature of state near the critical point would depend highly on regime, T T (2 − 3)T , has a qualitative trend c . . c model artifacts (compare the intersection of the dotted considerably different from the hadron resonance gas or lines in the top panel of Fig. 15 with the smooth lattice results). However, if one instead restricts the use of each model equation of state to its domain of applicability, and interpolates between the two pictures one can obtain a 8 The lattice results do not appear to reproduce the non- interacting Stefan-Boltzmann limit, which one sees only at tem- physically sensible pressure. By analogy, in finite den- peratures beyond those considered in the calculations. One can sity matter, as depicted in the bottom panel of Fig. 15, a understand this difference by considering terms in pQCD to or- smooth interpolation between hadronic and quark matter der g5, where g is the QCD coupling constant. allows a wider range of equations of state, while avoid- 21 ing the artifacts introduced by applying models beyond A critical aspect of hadron-quark continuity is the pos- their regimes of validity. One should keep in mind, how- sibility that the quark and hadron phases have the same ever, that such construction depends on the choice of the symmetry structure. While phases with different (exact) interpolating functions, which brings its own source of symmetries are separated by a first or second order phase ambiguity. transition, the symmetries of the superfluid baryon phase in hadronic matter can, as Schäferand Wilczek [105] el- egantly discussed, be smoothly connected to those in the E. Hadron-quark continuity and percolation CFL phase in quark matter; thus there need not be a sharp phase transition separating the hadron and quark matter phases.10 The existence of the crossover from the hadronic to In contrast, the conventional picture of dense nuclear the quark phase at zero baryon chemical potential raises matter is that there exists a first order (chiral) phase a very instructive question [259], as it seems to imply transition beginining at zero temperature and terminat- that since quarks are free in the plasma phase above the ing at a high temperature the critical point [48, 268] (see crossover at low µ and finite T , by continuity free quarks B Fig.1), as in a liquid-gas phase transition where one can would have a probability, albeit small, to be present go continuously from liquid to gas around the critical in matter below the crossover, e.g., there could be free point (in water at 373 C). If the low temperature part of quarks running around in air. This situation is reminis- the first order line is in fact a crossover, there must ex- cent of the very tiny possibility of finding free electrons ist a second, low temperature critical point (as shown in in air, as a result of thermal ionization. But since there Fig.1)[49, 269, 270]. It is also possible that the crossover cannot be free quarks in confined low density matter, the at high temperature and low baryon density is directly correct conclusion is that even above the crossover, there connected to the low temperature crossover, without the are no free quarks (except in the very high T asymp- need for the conventional first order line.11 totically free regime); rather the matter must consist of complicated clusters of gluons and quarks both above and below the crossover, as illustrated in Fig.2 for matter at finite µ and zero temperature, and Fig.3 for matter at B bility of quarks being able to traverse the system, as in Fig.2. µB = 0 and finite temperature. One must take into account the quantum nature of the hadrons in The crossover, and deconfinement more generally, can discussing their overlap. For example even though the electron be characterized as a percolation transition – in which wavefunction in a simple hydrogen atom has an exponentially the region in which the quarks can roam freely changes small tail extending to infinity, one would not claim that any from finite in extent to the system size – as first proposed two hydrogen atoms in a gas at whatever distance are always overlapping. Similarly, in hadrons, the pion and otherqq ¯ struc- in [86] in terms of classical percolation theory for dense tures (as well as the wee partons) extend well beyond the hadron matter at zero temperature and further amplified by Satz core. Understanding the quantum percolation transition in dense and coworkers [87, 88] in terms of quark mobilities; also matter and its relation to Anderson localization of particles in a disordered system remains an open question [261]. [260]. As seen in Fig.2, percolation at finite µB and zero 10 Hadron-quark continuity in states of finite angular momentum temperature proceeds as baryons exchange many quarks between hadronic matter and superfluid quark matter in the CFL and lose their identities. Eventually baryons overlap and phase is more subtle. While superfluid hadronic matter and quark matter is formed. The regions of space in which quark matter each carry angular momentum in quantized vor- quarks can move around are color singlets, from nuclear tices, owing to quarks having baryon number 1/3, a triply quan- to quark matter domain. Similarly at finite temperature tized vortex in the hadronic regime would carry the same angular momentum per baryon as a singly quantized U(1)B vortex in the (for µB small compared with the nucleon mass) the clus- quark regime [262]. Thus in a rotating neutron star in which the ters, Fig.3, are isolated as single thermal pions which be- nuclear superfluid evolves with depth into a CLF quark phase, come more and more connected as T increases, through one would at first expect that at some point a surface of boo- the gluon and quark exchanges responsible for the inter- jums (where three low density hadronic vortices merge into one actions of the pions, until the clusters fill enough of space high density CFL vortex) between the low density hadronic and high density quark matter regions. However, in the CFL phase, that a single quark can propagate from one end to the a single U(1)B vortex is unstable against transforming into three other. As at finite density, the regions of space in which color flux tubes [263–265], suggesting that at the boojum three quarks can move around are always net color singlets. low density vortices transform into three color flux tubes [266]. At the percolation transition the sizes of the color singlet In fact, however, each low density vortex can transform directly regions change from always being finite in the hadronic to a single such non-Abelian vortex without a boojum, consistent with hadron-quark continuity [267]. regime to being the size of the entire system, e.g., the 11 The Asakawa-Yazaki critical point is being searched for in exper- collision volume in a heavy-ion collision.9 imental programs at the RHIC heavy ion collider at Brookhaven National Laboratory [271], the SPS at CERN [272], SIS at GSI in Germany, and will be searched in the future program at FAIR at GSI, NICA at JINR in Dubna [273], and J-PARC in Japan [274]. Recently possible experimental signatures for the (conventional) 9 While it is easiest to visualize the transition as classical perco- critical point were found in analyses for the critical fluctuations lation in which regions of space available to the quarks overlap, [275] and the finite volume scaling [276]. Owing to controversies a more precise picture of the transition is in terms of the proba- in the interpretation of those results [277–279], further studies 22

As matter goes from having hadronic to quark degrees that the interpolated pressure matches the hadronic and L U of freedom, it may pass through spatially inhomogeneous quark values at µB and µB, while satisfying the thermo- 2 2 phases [281, 282]. While this is an intriguing possibility, dynamic constraint ∂nB/∂µB = ∂ P/∂µB > 0, as well we will not treat it in this review, except to mention one as the (reasonable) causality condition that the adiabatic 2 unconventional state, quarkyonic matter, which has both speed of sound at zero frequency, cs = ∂P/∂ε not exceed aspects of nuclear and quark matter [227]. This state is the speed of light.12 These conditions place significant re- conjectured from studies of dense matter in QCD with a strictions on the acceptable interpolations of the pressure large number of colors Nc; in this limit screening of gluons in the intermediate density regime, and provide insights by quarks is suppressed by a numerical factor 1/Nc, and into the qualitative properties of this critical domain in thus the gluons remain confined until the quark chemical neutron star structure. 1/2 potential µq = µB/Nc reaches ∼ Nc ΛQCD MN /Nc. As noted, the primary distinction between hybrid and In the limit of a large number of colors, the dominant unified constructions of the equation of state is that in pairing in the quarkyonic matter is, instead of diquark the latter no direct comparison of the hadronic and quark pairing, the formation of a spatially inhomogeneous chi- pressures is made, since the domains of validity of the ral condensate of quark particle-hole pairs, called chiral hadronic and quark descriptions do not overlap. Ac- spirals. This idea is applied to neutron stars in Ref. [138]. cordingly, a number of the stiff quark matter equations In the real world with Nc = 3, the extent to which glu- of state excluded by the conventional construction (see ons are screened due to quark excitations remains unclear Fig. 14 and related discussion above) are allowed within a [251, 283]. unified construction. Furthermore, the unified construction can encompass hadron-quark continuity. A simple but reasonably general function to interpolate F. Unified construction the equation of state between the hadronic and the quark matter regimes is a polynomial which smoothly joins the In the unified procedure to construct the equation of hadronic and quark pressure curves between µ = µBL state one explicitly restricts the hadronic and quark mat- and µ = µBL, ter equations of state to their respective domains of va- N X m lidity, avoiding the potentially unphysical implications P(µB) = CmµB for µBL < µB < µBU ,(63) of the conventional construction. The hadronic equa- m=0 tion of state is used only at low densities, n < n B BL where µ and µ are chosen so that n (µ ) ∼ 2n (“L” for lower), where two- and three-body forces dom- BL BU B BL 0 and n (µ ) ∼ 5n . The coefficients C are chosen inate and the composite nature of hadrons is not mani- B BU 0 m to satisfy matching conditions at the boundaries of the fest. A reasonable choice of the maximum density n is BL interpolating interval. In general, we require that ∼ 2n . We denote the corresponding chemical potential 0 as µBL. Similarly, the deconfined quark-matter equation ∂P ∂PH P(µBL) = PH (µBL) , = , ··· of state is used only at relatively high densities, beyond ∂µ ∂µ B µBL B µBL where baryons first percolate and quarks can no longer ∂P ∂PQ be thought of as belonging to specific baryons. For a P(µBU ) = PQ(µBU ) , = , ··· . ∂µB ∂µB typical baryon radius of rB ∼ 0.5 fm, geometric percola- µBU µBU 3 tion should occur at a baryon density ∼ 0.08/rB ∼ 4n0 (64) [86], and thus a reasonable choice of the lowest density at which to use a quark matter equation of state The number of derivatives to be matched at each boundary is a matter of choice. Matching up to the second is nBU ∼ (4 − 7)n0; we label the corresponding chemical potential µ (“U” for upper). [In the calculations in derivative at each boundary ensures that the pressure, BU baryon number density, and baryon number compress- Sec.VI we choose nBU = 5n0 as a specific illustrative value.] ibility (or susceptibility), ∂nB/∂µB, are continuous. In In the density range nBL < nB < nBU neither a purely hadronic nor quark matter picture is applicable. Given the present intractability of directly calculating the equa- 12 However, Refs. [284–287] indicate that this requirement of causal- tion of state in this domain, a simple approximate ap- ity on cs is still suggestive; we are not aware of a rigorous proof that c ≤ c is necessary for causal propagation of signals and proach is to interpolate P (µB) between the two limiting s regimes in a thermodynamically consistent way, requiring information. In particular, Lorentz invariance itself does not impose such a constraint [284, 287], and it is possible to devise models that exhibit superluminal sound speed cs (and sometimes even superluminal group velocity as well), yet with causal propagation of signals [284–286]. The first√ argument that the sound are called for. It should be emphasized that the current state-of- speed in dense matter can exceed c/ 3 is given by Zel’dovich the-art lattice QCD studies based on a Taylor expansion in µB /T [288]. The assumption of cs ≤ c was used early on to obtain a around µB = 0 disfavor a critical point in the region where the maximum possible neutron star mass of 3.2 M [289], compared expansion is trustworthy, µB /T . 2 [280]. So far the existence with a mass of ∼ 5M in [290] which did not assume cs ≤ c; of the first order phase transition has not been established. refined upper bounds were given in [209, 291]. 23

FIG. 16: Schematic interpolation of the hadronic (PH ) and FIG. 17: Schematic interpolation violating the causality condi- quark (PQ1, PQ2) equations of state. For PQ1 the interpolated ton. The linear interpolated pressure implies a constant baryon pressure is physically acceptable. However, for PQ2 one cannot and energy density (see Fig. 10); the latter condition leads to the 2 construct an interpolated pressure without introducing an in- unphysical result cs = ∂P/∂ε → ∞. flection point. Such an unphysical feature implies a mechanical instability and PQ2 must therefore be discarded from considera- tion. this case one has six boundary conditions so one needs to include polynomials up to N = 5. As discussed in Sec.VF, the interpolated pressure as a function of µB is constrained by the stability condition that P (µB) be without an inflection 2 2 point, and the requirement that cs/c = ∂P/∂ε = 2 (1/c )∂ ln µB/∂ ln nB ≤ 1, so that the thermodynamic sound speed does not exceed the speed of light. The stability condition is depicted in Fig, 16, where one can interpolate smoothly between the schematic hadronic equation of state PH and the quark matter equation of state 2 2 PQ1, with ∂ P/∂µB > 0; however, an interpolation between PH and the quark matter equation of state PQ2 FIG. 18: Interpolated equation of state for both hadron-quark necessarily has an inflection point, violating the stability continuity (smooth curve) and a first order phase transition criterion. A situation in which the speed of sound ex- (kinked curve). In the latter case, for which P (µ) must lie below ceeds the speed of light is shown in Fig. 17, where the the smooth curve, the slope increases discontinuously at the tran- slope is constant; in this region P varies but ε remains sition. In the small panel we show the corresponding dependence 2 of the sound speed on the energy or mass density; generally, the constant, so that cs = ∂P/∂ε → ∞. In general a P (µB) kinked curve yields a larger sound speed in the hadronic regime, which grows too slowly violates the causality condition. with the sound speed vanishing in the transition region.

G. First order phase transitions

The possibility of a first order transition is not ex- an unphysical inflection point. In addition, the slope of cluded in the unified construction. However, such a tran- the hadronic pressure curve is smaller in the case of a first sition more severely constrains the interpolated pressure order transition than for a continuous evolution, giving than does a continuous hadron-quark evolution. Fig- rise to a correspondingly larger sound speed. Thus, for ure 18 compares the interpolated pressure curves for specified hadronic and quark matter equations of state, a continuous (smooth curve) and a first-order (kinked an interpolation with a kink, a first order transition, has curve) hadron-quark phase transition. The boundary a greater chance of violating the causality constraint. matching conditions severely restrict the strength of a These conditions strongly restrict both the location and possible first order phase transition, as measured by the strength of a possible first order hadron-quark transition change in slope of the hybrid pressure curve at the tran- [231]. This constraint, a consequence of accounting for sition. In particular, except for relatively small changes massive neutron stars, would indicate that there should in slope it is impossible to match the slopes of both the not be a strong first order chiral restoration transition in hadronic and quark matter curves without introducing QCD at low temperature (see subsec.VE). 24

VI. EXPLICIT CONSTRUCTION OF UNIFIED 0.7 NJL , g /G= 0.0 EQUATIONS OF STATE nB>5n0 v 0.6 0.5 1.0 0.5 APR The two fundamental ingredients in constructing a uni- nB<2n0 ] 3 fied equation of state that can explain neutron stars of 0.4 masses & 2M are first a large vector repulsion gV and second a large diquark pairing interaction H. As we will 0.3 P [GeV/fm discuss, the characteristic diquark coupling, H, is larger 0.2 than the conventionally assumed H/G = 3/4 expected 0.1 from the Fierz transform of the one-gluon-exchange in- H=K’=0 teraction. In fact, en route with decreasing density to 0 the strong two and three quark correlations that eventu- 0 0.2 0.4 0.6 0.8 1 1.2 1.4 3 ally become-well defined nucleons, the system may have ε [GeV/fm ] larger pairing correlations than at higher density, so that 0.7 H/G > 1 (and ∆ ∼ 100 − 200 MeV) at densities of inter- NJL , g /G= 0.0 nB>5n0 V est. 0.6 0.5 1.0 The vector repulsion (35) plays a major role in stiff- APR 0.5 nB<2n0 ]

ening the equation of state calculated in the NJL model, 3 as shown in Fig. 19. While the equation of state is con- 0.4 siderably softer than the nucleon-based APR equation of 0.3

state for small vector couplings, for sufficiently large gV , P [GeV/fm it can be as stiff as APR across a wide range of densities, 0.2 H=K’=0 thus enabling NJL equations of state at sufficiently large 0.1 gV to explain massive neutron stars. However, increasing 0 gV moves the NJL pressure curve, as a function of µB, 1 1.2 1.4 1.6 1.8 2

away from the APR pressure curve in the positive µB µB [GeV] direction, making it harder to interpolate between the two phases without introducing an unphysical inflection FIG. 19: Effects of the vector repulsion on the pressure vs. point, as seen in the P (µ ) curve for H = 0 in Fig. 20. B energy density (top) and pressure vs. quark chemical potential This figure shows the interpolated P (µB) for H = 0 and (bottom) for vector couplings gV /G = 0, 0.5, 1.0, without pairing, 0 0 1.5 G, with gV = 0.8 G and K = 0. For H = 0, the APR H = K = 0. The NJL curves are shown as bold lines for nB > and NJL curves are rather widely separated in µB and it 5n0, and as thin lines below. The APR equation of state (solid is difficult to construct a sensible interpolated equation of line for nB < 2n0 and double dotted line above) is also plotted for comparison. One sees here clearly in the upper panel how state; Fig. 21 shows the corresponding plots of nB vs. µB, increasing gV stiffens the equation of state. while Fig. 19 shows the pressure as a function of energy or mass density. With increasing H the NJL pressure toward lower chemical potential, enabling one, with large gV and H, to construct a physical interpolation between 13 the hadronic and quark regions. 0.7 To demonstrate explicitly the construction of a unified (gV,H)=(0.8, 0.0) G equation of state, we assume the APR hadronic equation 0.6 interpolated (gV,H)=(0.8, 1.5) G of state below a baryon density, nL ≡ nB(µBL) ' 2n0, 0.5 interpolated ]

above which the underlying hadronic description begins 3 APR to break down. We also assume a quark matter equation 0.4 of state above a density nU ≡ nB(µBU ) ' 5n0, to choose 0.3

a speciﬁc representative value, and carry out a polyno- P [GeV/fm mial interpolation for µB between µBL and µBU . Within 0.2 the range of the NJL model parameters we discuss, the 0.1 variation of nU from 4n0 to 10n0 does not produce sig- K’=0 niﬁcant qualitative changes in the resulting equation of 0 state. 1 1.2 1.4 1.6 1.8 2 µB [GeV]

FIG. 20: Pressure vs. baryon chemical potential for interpo- 13 Were we to include in the quark phase a possible negative residual lated equations of state, for parameters gV = 0.8G, with H = 0 pressure, Pg, originating from a (non-perturbative) condensate and 1.5 G, and K0 = 0. At H = 0 the interpolated equation of 4 3 of gluons, for Pg as large as ∼ ΛQCD ∼ 200 MeV/fm , the total state has an unstable region, ∂nB /∂µB < 0. pressure of the quark matter would become too soft to allow a sensible interpolation between the APR and NJL models. 25

10 10 (g ,H)=(0.8, 0.0) G V H=K’=0 interpolated 8 8 (gV,H)=(0.8, 1.5) G interpolated APR 6

6 0 0 /n /n B B n n 4 4 NJL , g /G= 0.0 nB>5n0 V 2 0.5 2 1.0 APR K’=0 nB<2n0 0 0 1 1.2 1.4 1.6 1.8 2 1 1.2 1.4 1.6 1.8 2 µB [GeV] µB [GeV]

FIG. 23: Eﬀects of the vector repulsion on the baryon density Baryon density vs. chemical potential for interpolated −3 FIG. 21: nB , in units of nuclear matter density, n0 = 0.16 fm , as a func- equations of state for the same parameters as in Fig. 20. The tion of the baryon chemical potential, for vector couplings gV /G discontinuity in the APR curve at µB ∼ 1 GeV is a result of = 0, 0.5 and 1.0. The baryon density in the APR nucleon equa- the onset of neutral pion condensation, and that in the dotted tion of state is shown for comparison. The bold line at nB > 5n0 curve at µ ∼ 1.2 GeV is the transition from 2SC to CFL color B is the quark pressure, and at nB < 2n0 the APR pressure. The superconductivity. discontinuity in the solid part of the APR curve is at their onset of pion condensation. The sudden rise of the curve for gV = 0 indicates chiral symmetry restoration transition in the NJL model. 0.7 (gV,H)=(0.8, 1.5) G 0.6 interpolated APR separated in µB and it is diﬃcult to construct a sen- 0.5 sible interpolated equation of state; Fig. 21 shows the ] 3 corresponding nB vs. µB. With increasing H the NJL 0.4 pressure curve shifts toward lower chemical potential, en- 0.3 abling one to construct a physical interpolation between

P [GeV/fm the hadronic and quark regions by employing large gV 0.2 and H. 0.1 The diquark pairing interaction significantly affects the K’=0 pressure; a larger H leads to larger pairing gaps, which 0 as discussed in subsec.VA, leads to a stiffer equation 0 0.2 0.4 0.6 0.8 1 1.2 1.4 2 of state (for cs > 1/3, as is the case in the high den- ε [GeV/fm3] sity quark regime for gV > 0). This effect is illustrated in Fig. 24, where with increasing H the pressure as a FIG. 22: Pressure vs. energy density for interpolated equations function of µB is shifted toward lower chemical poten- 0 of state, with parameters. gV = 0.8G,, H = 1.5G, K = 0. tial, tending to eliminate the unphysical inflection point. The flat region in the pressure reflects the transition from 2SC to In addition, increasing H increases the baryon density CFL color pairing. at given µB. Both behaviors result from the reduction, with increasing density, of the average single quark energy by the pairing, or color-magnetic, interaction. Such The detailed effects of the vector repulsion on nB as a a reduction is expected from constituent quark models function of µB are shown in Fig. 23; as expected, increas- [233], where the color-magnetic interaction reduces the ing the repulsion decreases nB for given µB. To inves- baryon mass from approximately three times the con- tigate the effects of vector repulsion on the equation of stituent quark mass (∼ 3×336 MeV) down to the nucleon state, we explore a range of couplings, 0.5G ≤ gV ≤ 1.0G. mass, 938 MeV; in uniform quark matter pairing near the 2 2 Increasing the vector repulsion also reduces the tendency Fermi surface leads to an energy reduction δε ∼ −pF ∆ . of high densities to restore chiral symmetry, an effect We examine here a range of diquark pairing strengths, discussed in AppendixB. When gV exceeds a critical 0.8G ≤ H ≤ 1.5G, notably larger as mentioned in sub- value, dependent on H, one cannot interpolate between sec.IVH, than in previous studies of hybrid equations of the hadronic and quark regimes without introducing a state. Note that the pressure P (µB) of the quark phase mechanical instability, as seen in Fig. 20. for H = 1.5 G with gV = G (top panel, Fig. 24) is larger We next discuss how the interpolation depends on the than that of the hadronic APR equation of state. In the model parameters. Figure 20 shows the interpolated hybrid construction, these pressures must be rejected, 0 P (µB) for H = 0 and 1.5 G, with gV = 0.8 G and K = 0. thus restricting H to a smaller range [101]. The direct For H = 0, the APR and NJL curves are rather widely effects of increasing H on the stiffness of the equation of 26

0.7 state P (ε) are shown in Fig. 25, for H/G = 0, 1.0, 1.3, 1.5, NJL , H/G= 0.0 0 nB>5n0 with gV = G and K = 0. The flat line, which appears at 0.6 1.0 1.3 nB . 5n0, reflects the first order phase transition from 0.5 1.5 the 2SC pairing state to CFL pairing, with accompany- ]

3 APR nB<2n0 ing softening of the equation of state. In typical NJL 0.4 studies with small H, the softening associated with the 0.3 appearance of condensates leads to a smaller maximum

P [GeV/fm neutron star mass [101]. As seen in Fig. 25, the paired 0.2 phase becomes stiffer than the unpaired phase at high 0.1 gv/G=1.0, K’=0 density, a behavior consistent with the schematic discussion in subsec.VA; at larger H, the stiffening occurs at 0 1 1.2 1.4 1.6 1.8 2 lower density. In particular, at H ≥ 1.5 G, the paired µB [GeV] phase at nB & 5n0 is stiffer than the phase at H = 0. As noted in [118], increasing K0 slightly stiffens the 8 NJL , H/G= 0.0 quark matter equation of state; however, its impact is nB>5n0 7 1.0 much smaller than those of gV and H, as long as we con- 1.3 6 0 1.5 sider a reasonable value of K ∼ K [144, 145]. Thus, for APR 5 nB<2n0 the sake of simplicity, we restrict the present considera-

0 tions to K0 = 0, but note that a non-zero K0 allows one /n 4 B

n to choose larger gV . 3 Figure 26 shows the squared sound speed as a function 2 gv/G=1.0, K’=0 of ε; in this figure the interpolation region is roughly from 1 ε ∼ 0.3-1 Gev/fm3. The large sound velocity in the high 0 density quark regime is driven both by the large gV and 0.8 1 1.2 1.4 1.6 1.8 2 H. µB [GeV] In AppendixB we review the effects of gV and H on the quark effective masses generated by chiral symmetry FIG. 24: Effects of diquark pairing on the pressure P (top breaking, on the restoration of chiral symmetry, and on panel) and normalized baryon density nB /n0 (bottom panel) as functions of the baryon chemical potential, for vector coupling the pairing gaps generated by the diquark condensation. 0 gV = 1.0 G, axial anomaly coupling K = 0, and diquark couplings H/G = 0, 1.0, 1.3, and 1.5. With increasing H, the P and As Fig. 20 shows, for large gV with large H, the equa- nB curves shift toward lower chemical potential. tion of state satisfies the stability constraint, and in addition can support neutron stars with masses above 2.0 M , with a subluminal sound velocity. Figure 22 shows the interpolated P (ε) for parameters, gV = 0.8 G, 0 0.7 H = 1.5 G, K = 0, with beta equilibrium included; the NJL , H/G= 0.0 nB>5n0 equation of state in this form directly enters the TOV 0.6 1.0 1.3 equation. For the given interpolation range, we note that 0.5 1.5 the interpolated pressure P (ε) increases rather rapidly ]

3 APR nB<2n0 to merge into PNJL(ε). This rapid stiffening is a rather 0.4 generic feature of hadron-quark interpolations that yield 0.3 equations of state stiff enough to satisfy the 2M con- P [GeV/fm straint. However the causality constraint, ∂P/∂ε . 1, re- 0.2 stricts the rate at which matter can stiffen, so the freedom 0.1 to choose the model parameters is significantly limited. g /G=1.0, K’=0 v In fact, model studies show that the physical interpola- 0 0 0.2 0.4 0.6 0.8 1 1.2 1.4 tion almost uniquely fixes the value of H for a given gV . ε [GeV/fm3] For gV /G = 0.5, 0.8, and 1.0, we are required to choose H/G ' 1.4, 1.5, and 1.6, respectively. Below, we show the results for these sets of parameters. FIG. 25: Effects of diquark pairing on the equation of state, P (ε), for the same parameter sets as in Fig. 24. The discontinu- In AppendixC we give parametrized forms for rep- ous change of ε at fixed P reflects the first order 2SC-CFL phase resentative unified equations of state in terms of sim- transition. For 0 < H < 1.5 G, the equation of state is softened ple functions. We call the set of such equations of state immediately following the 2SC-CFL phase transition, but as density increases, the equation of state eventually stiffens relative to QHC18 – for quark-hadron crossover (2018 version) – and that at H = 0. For H = 1.5 G, the equation of state is stiffer than give them together with instructions for use on the web- in unpaired matter for all densities. site: Home Page of Relativistic EOS table for supernovae (http://user.numazu-ct.ac.jp/∼sumi/eos/index.html). 27

2.5

1.4 (gV,H)=(0.5, 1.4) G (0.8, 1.5) G 1.2 (1.0, 1.6) G 2 MJ1614-2230 = 1.93Mʘ 1 1.5 ʘ 2 0.8 s M = 1.4M

c ʘ

0.6 M / M 1 0.4 0.5 0.2 (g V,H)/G=(0.5,1.4) K’=0 (0.8,1.5) 0 (1.0,1.6) 0.2 0.4 0.6 0.8 1 1.2 1.4 0 0 1 2 3 4 5 6 7 8 9 ε [GeV/fm3] nBc /n 0

2 Neutron star mass as a function of central baryon FIG. 26: The squared speed of sound cs = ∂P/∂ε as a function FIG. 27: of ε. We compare (gV ,H)/G = (0.5, 1.4), (0.8, 1.5), and (1.0, density, nBc, using the uniﬁed equation of state (QHC18) for 0 1.6). For each gV , the choice of H is ﬁxed to within ∼ 10% to gV = 0.8G, H = 1.5G and K = 0, with nU = 5n0. produce physical interpolations. The dip at ε ∼ 0.2 GeV/fm3 is at the onset of pion condensation in APR. 2.5

2 VII. NEUTRON STARS WITH UNIFIED MJ1614-2230 EQUATIONS OF STATE 1.5 ʘ We turn now to the implications of the unified con- M=1.4Mʘ M/M struction of the equation of state (QHC18) on astrophys- 1 ical properties of neutron stars, and the constraints indicated by current observations, notably the measurements 0.5 (g ,H)/G=(0.5,1.4) of the two neutron stars with masses ∼ 2M [41, 42], V (0.8,1.5) early inferences of the mass-radius relation [5, 11, 12], and (1.0,1.6) the tidal deformability of neutron stars bounded from 0 6 7 8 9 10 11 12 13 above by the binary neutron star merger, GW170817 [22]. R [Km] We integrate the TOV equation (2) for a given value of the central baryon density to construct a family of FIG. 28: Mass-radius relation for neutron stars with the same equations of state (QHC18) as in Fig. 27. stars whose masses and radii are functions of nBc. As illustration we use the interpolated equations of state in the liquid interior for three sets of parameters (gV ,H) = (0.5, 1.4) G, (0.8, 1.5) G, and (1.0, 1.6) G, for which we time with improving certainty in our theoretical under- we able to construct sensible interpolated equations of standing of the nuclear matter equation of state, the state (see Sec.VI) with the interpolation window from quark matter parameters – and indeed the quark model n = 2.0n to n = 5.0n . L 0 U 0 itself – as well as in the interpolation from nuclear matter At densities n from 0.26 n to 2.0 n in the liquid in- B 0 0 to quark matter. terior we use the APR equation of state, and in the crust The neutron star mass as a function of central baryon for nB . 0.26n0, we take the Togashi equation of state [115], which includes the same detailed physics for the density is shown in Fig. 27. The choice gV = 0.5 G is inner crust nuclei as APR includes in the nuclear matter not stiff enough to satisfy the 2M constraint, and we liquid in the interior. Combining these two equations of are required to take larger values of gV . At the maximal state provides a consistent physical description from the mass the baryon density is & 5n0 or higher, where we inner crust into the liquid interior.14 expect a quark matter description to be valid. The unified equations of state and hence the neutron Figure 28 shows the neutron star mass-radius relation. star models constructed from them will be refined over Since the overall radii of neutron stars are primarily determined by the equation of state at nB . 2n0, neutron stars with masses ∼ 1.4 M have similar radii, 11.3- 11.5 km, for the three parameter sets. The relative small- 14 As discussed in AppendixC, the SLy(4) equation of state for the ness of the radii, compared to those found with typical crust, based on Skyrme effective interactions, does not join in a relativistic mean field equations of state, e.g., [84], re- thermodynamically consistent manner onto the APR equation of flects the relative stiffness of mean-field equations of state state. at low density. The mass-radius relations of the unified 28 equations of state reviewed here are reasonably similar to that obtained with a pure nuclear matter equation of state, e.g., APR. The similarity of radii in hybrid and pure nuclear matter stars is analyzed in [231]. 1000 While the masses of certain neutron stars, e.g., those in binary orbits with another neutron star or a white dwarf,

can be inferred observationally to good accuracy, obser- ] 3 vational determination of neutron star radii is much less 100 precise, and does not at this stage permit a detailed comparison with the neutron star models shown in Fig. 28.

Data from the NICER experiment [15] will make such P [MeV/fm a comparison more feasible. A complete and accurate Ozel et al. mass vs. radius curve would allow a determination of 10 Steiner et al. the equation of state [292]; in this spirit, observations of APR (g ,H)/G=(0.5,1.4) burst and quiescent low mass X-ray binaries in globu- V (0.8,1.5) lar clusters have been used to infer masses and radii of (1.0,1.6) some fourteen neutron stars [5, 11–13] (and references 1 therein); see discussion in subsec.IF. These measure- 1 2 3 4 5 6 7 8 9 ments, while not sufficiently accurate to determine the nB/n0 equation of state, do indicate constraints on it. Refer- ence [293] discusses the Bayesian analysis and accuracy of inference of an equation of state from M vs. R data FIG. 29: Pressure vs. baryon density for hadronic APR and three versions of the unified hadron-quark equations of state sets. (QHC18). All the equations of state include beta equilibrium. Figure 29 shows the pressure vs. baryon density cor- The shaded regions are those inferred from M-R measurements of responding to the three curves in Fig. 28. As implied neutron stars, at an effective 2σ level [11, 12]. by the arguments in subsec.VA, increasing gV as well as H tends to increase the pressure at fixed baryon density. The unified equations of state with quark matter at 2000 (gv,H)/G=(0.5,1.4) high density are softer than the APR equation of state; (0.8,1.5) such asymptotic softening is expected from consistency (1.0,1.6) 1500 with perturbative calculations valid at nB & 100n0. The shaded regions in this figure show a range of equations of state [5, 11, 12] that are compatible with the available mass-radius data inferred from bursts in low mass X-ray Λ 1000 binaries; overall, the unified equations of state are consistent with this range. While there are possible discrepan- 500 cies using the equations of state of Ref. [11] in the vicinity of n0, analysis of the differences in the inferences from the data is beyond the scope of this review. Nonetheless, re- 0 solving discrepancies with observation both through bet- 1 1.2 1.4 1.6 1.8 2 ter mass-radius determinations, as NICER will provide, M/Mʘ as well as through improvements to the theory of nuclear matter in beta equilibrium remains an open challenge. FIG. 30: The dimensionless tidal deformability Λ of a neutron Figure 30 shows the dimensionless neutron star tidal star as a function of the neutron star mass M, for the same equa- deformability Λ, Eq.(10), as a function of the neutron tions of state (QHC18) as in Fig. 27. star mass for the three parameter sets used in computing the M-R relations; Λ was calculated using the procedure summarized in [166]. For a 1.4M star, the range of The GW170817 merger measured the chirp mass, Λ is 240-270, similar to what one finds using the APR 3/5 −1/5 +0.004 Mchirp = (M1M2) (M1 + M2) ' 1.188 M . equation of state throughout the star. −0.002 The assumption of small spin (. 0.05) for each neu- The gravitational waveforms as detected in binary neutron star, which is probable from the population anal- tron star mergers are sensitive to the combination, yses, weakly constrains the mass ratio η = M1/M2 (for 4 4 M ≤ M ) to 0.7-1.0, but even with this uncertainty, η 16 (M1 + 12M2)M Λ1 + (12M1 + M2)M Λ2 1 2 Λ˜ = 1 2 , together wtih M tightly constrains the total mass, 13 (M + M )5 chirp 1 2 M + M ' 2.74+0.04M . Figure 31 shows the result of (65) 1 2 −0.01 our equations of state for Λ˜ as a function of η with fixed of the masses and Λ’s of the individual neutron stars, as Mchirp = 1.188M . The resulting Λobs depends weakly derived in a post-Newtonian calculation [168]. on η and is ' 290-320, consistent with the upper bound 29

(g v,H)/G=(0.5,1.4) region accurately and further refine neutron star mod- 1000 (0.8,1.5) els, one needs better understanding of the interactions, (1.0,1.6) e.g., the axial anomaly-induced Kobayashi-Maskawa-’t 800 GW170817 Hooft (KMT) interaction, that drive the hadron-quark crossover. Third, much work remains to be done in ad- 600 dressing the possibility of additional quark pairing struc- ~ Λ tures and inhomogeneous phases which may exist in the 400 cores of the densest neutron stars (e.g., pion condensation, quarkyonic matter, etc.), and how such structures 200 can span the hadronic to quark matter crossover. Even- tually one would like to go beyond simple NJL mod- 0 els of quark matter. Finally, as both the quantity and 0.4 0.5 0.6 0.7 0.8 0.9 1 quality of observational neutron star data are improved, η constraints on quark models and their parameters will continue to become more stringent. It is encouraging FIG. 31: The combined dimensionless tidal deformability that current observational inferences are consistent with Λ˜ of a pair of neutron stars as a function of the mass ratio hadron-quark continuity. Through precise astrophysical η = M1/M2(≤ 1), for the same equations of state (QHC18) as in Fig. 27, with Mchirp = 1.188M . The upper bound from observation, assessment of such continuity can be contin- GW170817 is shown. The curves terminate at low η when M2 ually refined. reaches the maximum mass for the given equation of state. The effects of the underlying quark picture on the dynamical properties of neutron stars have yet to be fully explored. The color-magnetic interaction and the result- Λ˜ ≤ 800 (90% confidence level) found from analysis of ing quark pairing correlations are expected to play a sig- GW170817 [22]. nificant role in the non-equilibrium properties in neutron star interiors, e.g., in neutrino emission and thermal transport processes involved in cooling of stars. It is im- VIII. SUMMARY portant to delineate possible signatures of large quark matter cores in the cooling. As we have reviewed here, the unified construction of A major issue is to determine the effects of finite tem- equations of state (as exemplified by QHC18) avoids ar- perature on the equation of state of neutron stars, in tifacts arising from extrapolating the hadronic and non- order to understand how signals of neutron star struc- confining quark matter equations of state outside their ture are produced in the gravitational radiation emerg- ranges of validity, as in the conventional hybrid construc- ing from binary neutron star and neutron star–black hole mergers [25–30]. Neutron star temperatures in mergers tion of the equation of state. The unified construction 2 allows quark matter to be sufficiently stiff to produce can be as large as 10 MeV [27, 30, 294]. A initial estimate of the effects of finite temperature on the equation at least 2M neutron stars, and at the same time takes strange quarks into account. The requirement of stiffness of constraint consistent with the present point of view of disfavors a strong first order phase transition (although the hadronic to quark matter crossover is given in Ref. a weak one in the interpolated domain cannot be ruled [295]. out), indicating that hadronic and quark matter are likely not distinctly different, but rather smoothly connected, with the quark interactions as strong as those in the QCD IX. ACKNOWLEDGMENTS vacuum. In the quark matter equations of state discussed here, First and foremost we thank Chris Pethick for his in- both the repulsive vector and attractive color-magnetic valuable input, insights, and support during the writ- interactions play important roles in constructing phys- ing of this review, including his substantial contributions ically acceptable unified equations of state. The M > to the discussion of neutron star crusts and the nuclear 2M constraint requires those interactions to be as matter liquid interior. We thank Feryal Ozel,¨ Andrew strong as the NJL scalar interaction responsible for chiral Steiner, Hajime Togashi for providing us with the nu- symmetry breaking. merical tables for their equations of state, Dimitrious As discussed throughout this review, many outstand- Psaltis and Sanjay Reddy for very helpful input, and ing questions remain regarding neutron star properties in Charalampos Markakis and Michael O’Boyle for their general, and in the context of hadron-quark continuity in calculations of tidal deformability. Research reported particular. First, an exhaustive analysis of the parame- here was supported in part by NSF Grants PHY0969790 ter ranges of interactions in quark matter that is con- and PHY1305891. In addition, authors TH and TT were sistent with both hadron-quark continuity and current supported by JSPS Grants-in-Aid for Scientific Research astrophysical data has yet to be performed. Second, in (B) No. 24340054 and No. 25287066; GB and TH were order to describe the location and width of the crossover partially supported by the RIKEN iTHES Project and 30 iTHEMS Program; and TK was supported by NSFC 1 M , g /G= 0.0 grant 11650110435. Authors GB, TH, and TK are grate- H=K’=0 u V 0.5 ful to the Aspen Center for Physics, supported in part by 0.8 1.0 NSF Grants PHY1066292 and PHY1607611, where parts Ms, gV/G= 0.0 0.5 of this review were written. 0.6 1.0

0.4

Appendix A: Scaling the TOV equation Mass [GeV]

0.2 Here we derive a simple expression for the scale of masses and radii of neutron stars, as well as quark stars 0 made purely of quarks. We assume that the equation of 1 1.2 1.4 1.6 1.8 2 state is governed by a basic energy scale, 0, which is if µB [GeV] the order of the QCD scale parameter ΛQCD ∼ 200 MeV, and rescale the mass density and pressure by FIG. 32: The constituent masses of the u and s quarks, in the ρ = 4ρ,˜ P = 4P,˜ (A1) pure quark regime in the absence of diquark pairing. The d-quark 0 0 mass has essentially the same behavior as that of the u quark. whereρ ˜ and P˜ are dimensionless in units with = c = 1, The region µB below ∼1.5 GeV is in the intermediate regime, ~ where one must interpolate between the nucleonic and quark In addition we let ζ be the scale of the radius, writing pictures; the pure quark behavior shown in this regime is only to illustrate the restoration of chiral symmetry. r = ζr˜; (A2)

R r 2 Then the mass within radius r, m(r) = 0 4πr ρ(r)dr, scales as ζ34. Clearly by choosing Gζ24 = 1, or restor- also consistent with expected neutron star radii, ∼ 10−12 0 0 2 ing and c, km. Note also that the compactness M/R scales as c /G ~ 27 = 6.9 M /10 km, where M /10 km = 2.0 × 10 g/cm. 3/2 7/2 2 2 ~ c mpc ~ For the particular choice of the free quark equation ζ = = , (A3) 1/4 1/2 1/2 of state, Eq. (26), the natural scale is = B , and 2G 0 m c α 0 0 N p G we ﬁnd that the mass and radius scale as 1/B1/2 as in all the dimensional factors in the TOV equation cancel Eqs. (27) and (28); the prefactors there result from actual out; here integration of the TOV equation.

2 mpGN −38 αG = ' 0.589 × 10 (A4) ~c Appendix B: Effects of vector repulsion and the is the gravitational fine structure constant, with m the pairing on the constituent quark masses in quark p matter proton mass. After rescaling, the TOV equation (2) reduces to the dimensionless form: The phenomenological vector repulsion between ∂P˜(r) 1 (˜ρ + P˜)(m ˜ (˜r) + 4πr˜3P˜) = , (A5) quarks in quark matter has significant effects not only ∂r˜ r˜2 1 − 2m ˜ (˜r)/r˜ on the pressure but also on the broken chiral symmetry. For simplicity, we consider quark matter alone, and R r˜ 2 wherem ˜ (˜r) = 0 4πr˜ ρ˜(˜r)dr˜. neglect the transition to hadronic matter at low densi- We see then that the total mass scales as ties. We also neglect the anomaly-induced coupling be- 2 2 tween the chiral and diquark condensates, K0, for the m m c2 m c2 M ∝ p p = 1.86 p M . (A6) moment. Then with increasing quark chemical potential 3/2 0 0 αG µq = µB/3, a non-zero quark density begins to develop when µ exceeds the constituent u and d quark masses, (Note that M /m = 1.189 × 1057.) Similarly the radius q p M ,M ' 336 MeV. scales as u d For gV = 0, the quark density increases rapidly and 2 2 mpc causes the chiral condensate σ and hence the constituent R ∝ ζ = 17.2 km. (A7) quark mass M to decrease rapidly, as illustrated in 0 Fig. 32 for H = 0. Such a rapid change results in a The actual masses and radii found from integrating the first order chiral phase transition in which both the quark TOV equation are given by the scales set by Eqs. (A6) number and the chiral condensate become discontinuous; and (A7) times numerical factors of order unity; M˜ = see Figs. 23 and 32. On the other hand, increasing gV R 4πr˜2ρd˜ r˜ for the mass andr ˜ for the radius. With the from zero makes it more energetically costly for addi- 2 choice 0 = mpc the scale of neutron star masses is tional quarks to enter the system; the vector repulsion close to the expected maximum neutron mass, somewhat requires a larger chemical potential in order to achieve above two solar masses, while the scale of the radius is a specified baryon density. As a result, the slope of 31

0.8 0 M 10 g /G=1.0, H/G=1.5, K’=0 u 0.7 v M Outer crust d Intermediate crust Ms 0.6 10-2 Drip ∆ds Pasta 0.5 ∆su ∆ ud 10-4

0.4 ) 3 0.3 10-6 Gap/Mass [GeV] 0.2 P (MeV/fm 0.1 10-8 0 0.8 1 1.2 1.4 1.6 1.8 2 10-10 µB [GeV]

10-12 FIG. 33: The constituent masses and the pairing gaps as a 0.4

0 ) - 1 function of µB for H = 1.5G, gV = 1.0G, and K = 0, With 0.2 increasing density, the system undergoes a ﬁrst order phase tran- data 0 sition from the 2SC phase to the CFL phase. Just prior to this

/ P -0.2 phase transition, the strange quarks begin to appear. Thick lines fit -0.4 (P denote the density region nB > 5n0 which enters the construction -9 -8 -7 -6 -5 -4 -3 -2 -1 of the unified equation of state. 10 10 10 10 10 10 10 10 10 nB/n0 nB(µB) decreases, as seen in Fig. 23, and the melting FIG. 34: Pressure P vs. nB /n0 in the crust from nB = of the chiral condensate proceeds more smoothly, as seen −9 10 n0 to nB = 0.26n0. The quality of fit, δ ≡ Pfit/Pdata − 1, in Fig. 32. For gV exceeding a critical value ' 0.4G, the is shown in the lower panel. The parameterizations for the four first order chiral transition becomes a smooth crossover regions, outer crust, intermediate crust, drip regime and nuclear [48, 249, 250, 269]. pasta regime, are given in TableII. The constituent quark masses Mu,d,s and the pairing gaps ∆ud,ds,su for gV /G = 1 and H/G = 1.5 are shown To satisfy thermodynamic relations in numerical com- in Fig. 33. The first order transition from the 2SC phase putation, and to avoid artifacts, we parametrize only the to the CFL phase occurs around µ ∼ 1.3 GeV in the B single thermodynamic function, ε(nB), writing figure. If the anomaly-induced coupling K0 is increased ν from zero, the coexistence of the chiral condensate and max X lν diquark condensates becomes energetically favorable. As ε(ξ) = aξ + d0 + d1ξ ln ξ + dν ξ , (C1) a result, the chiral condensate survives to higher chem- ν=2 −3 ical potential, while diquark pairing begins to develop where ξ ≡ nB/n0, with n0 = 0.16 fm , and a, dn’s, at lower chemical potential. The quantities Mu,d,s and νmax, and lν ’s fitting parameters dependent on the do- ∆ud,ds,su in the region nB > 5n0 are denoted by the thick mains. Then the chemical potential, µB = ∂ε/∂nB, and 2 lines. All the details associated with the onset of the pressure, P = nB(∂ε/nB)/∂nB are given by strange quarks and the 2SC-CFL transition at nB < 5n0 " ν # 1 Xmax are not relevant for the unified equation of state, since µ (ξ) = a + d (1 + ln ξ) + l d ξlν −1 ,(C2) they occur in the interpolation region. B n 1 ν ν 0 ν=2 and Appendix C: Parameterized equations of state νmax X lν P (ξ) = −d0 + d1ξ + (lν − 1)dν ξ . (C3) ν=2 In this Appendix we present parameterized forms of 2 equations of state, ε(nB), where ε = ρc is the energy Note how the linear term in the pressure (∼ d1) corre- 15 density, and ρ is the mass density, as well as numerical sponds to a logarithmic term in ε and in µB. In prac- tables, to use in numerical modeling of neutron stars. tice, we use the pressure, rather than energy density, to Separating the star into domains – the outer and the determine the di because it is more sensitive to the values intermediate crust regions below neutron drip, the neu- of fitting parameters. tron drip region, and the nuclear pasta phase region; the low and high density regions of the liquid nuclear matter; the crossover; and higher density quark matter – we 15 The reader may be concerned that the parametrization of pres- use accurate polynomial fits of the numerical equations sure in the interpolation region Eq. (63) and that given here of state, with the coefficients in the polynomials tuned to are not obviously consistent. However, we use the present each domain. parametrization only to fit numerically the interpolated P (µB ). 32

−9 −6 The region 10 n0 . nB . 10 n0 contributes ∼ 20- 20 −9 APR LDP 100 m to the radii and ∼ 10 M to the masses. At −7 7 3 HDP nB & 10 n0 (∼ 3 × 10 g/cm ) the equations of state Phys are unaﬀected by magnetic ﬁelds as strong as 1014 G, and 9 15 by temperatures . 10 K[1]. ) 3

10 TableII summarizes the crustal ﬁtting coeﬃcients. To

P (MeV/fm obtain good ﬁtting quality within the present simple parameterization, we divide the crust into four regions: the 5 −9 −7 outer crust, from nB = (10 − 5 · 10 )n0, the inter- −7 mediate crust from nB = 5 · 10 n0 to the neutron drip −7 −3 regime nB = (5 · 10 − 1.5 · 10 )n0; the neutron drip −3 −2 0 regime, nB = (1.5 · 10 − 5 · 10 )n0; and the fourth −2 0.4 interval, nB = (5 · 10 − 0.26)n0, containing the various ) - 1 0.2 nuclear pasta phases.

data 0

/ P -0.2 fit -0.4 (P 1400 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2 (gV,H)/G = (0.5, 1.4) (0.8, 1.5) nB/n0 1200 (1.0, 1.6)

1000 FIG. 35: Fitting of P vs nB /n0 for the APR equation of state ) in β-equilibrium in the range nB = 0.26n0 to 2n0. At the phase 3 transition from low to high density the APR density changes 800 −3 discontinuously from nB ' 1.29n0 ' 0.21 fm to ' 1.50n0 ' 0.24 fm−3. The purple squares show the physical equation of 600 state. The parameterization is given in TableII. P (MeV/fm 400

1. Crust 200

In the crust we compare the SLy(4) equation of state 0 0.4 [114, 184] as given in the numerical tables of Haensel ) - 1 0.2 and Potekhin [112], and the Togashi equation of state data 0

[115] determined via variational calculations based on / P -0.2 fit -0.4

the microscopic AV18 two-body plus the UIX three-body (P potentials (numerical tables can be found in CompOSE 2 3 4 5 6 7 8 9 10 data base, http://compose.obspm.fr/eos/105/). These nB/n0 equations of state are consistent below the nuclear drip regime, while at higher densities the two equations of FIG. 36: Pressure vs. nB /n0 for the interpolated and state start to deviate because of the differences in inter- quark matter equations of state. The parameter sets are 0 actions and treatments of the nuclei in a neutron gas. (gV /G, H/G) = (0.5, 1.4), (0.8, 1.5), (1.0, 1.6), all with K = 0. The difference is important for matching the crust equation of state to that in the nuclear liquid. The Togashi equation of state, being based on the same input physics, can be matched with APR in a thermodynamically con- 2. APR in beta equilibrium sistent manner, while SLy(4) cannot, in the absence of ad-hoc smoothening and reduction of data points. We detail here the APR equation of state in the range Taking APR to describe the nuclear liquid, we adopt 0.26n0 < nB < 2n0. The fitting coefficients are sum- the Togashi equation of state for the crust. We chose the marized in TableII. From the parametrizations of the matching point with APR to be nB = 0.26n0, around equation of state for pure neutron matter and symmetric which the two equations of state overlap fairly well. The nuclear matter in the APR paper [52], one can approx- fit (C3) is compared with the tabulated equation of state imately calculate the equation of state in beta equilib- −9 in Fig. 34, where only the data points from nB = (10 − rium by quadratic interpolation, Eq. (23); in interpolat- −6 5 8 10 )n0 are used (corresponding to ∼ (3 × 10 − 3 × 10 ) ing with the APR parametrized forms we use the aver- 3 −9 g/cm ). The region nB . 10 n0 does not affect the age of the proton and neutron masses, (mp + mn)/2 ' overall structure of neutron stars and can be ignored here. 938.92 MeV. 33

TABLE II: Fitting parameters for the crust and nuclear liquid regions (see text). The unit of the coeﬃcients are MeV/fm3. Each domain starts at the threshold ξth = (nB /n0)th shown in the table.

ξth a d0 d1 d2 d3 l2 l3 Outer crust 10−9 149.9 − −7.112 · 10−2 9.168 −1.522 1.271 0.97508 Intermediate crust 5 · 10−7 148.2 − −3.391 · 10−2 6.922 −3.591 · 10−9 1.224 1.0 · 10−2 Drip 1.5 · 10−3 150.9 − 9.940 · 10−2 3.941 −6.422 · 10−4 2.205 9.353 · 10−2 Pasta 5 · 10−2 150.1 − 0.9845 1.443 1.861 5.059 0.7864 APR (LDP) 0.26 151.5 −6.690 · 10−2 4.914 · 10−2 1.320 − 3.056 − – 1st order – 1.29 − − − − − − − APR (HDP) 1.50 151.7 1.220 2.461 0.2666 − 3.856 −

0 TABLE III: Fitting parameters for the unified equations of state for various sets of (gV ,H)/G with K = 0 (see text). The unit of 3 coefficients are MeV/fm . The parameters lν ’s are fixed to integers, lν = ν. To have a good quality fit in the interpolated domain, it is important to include all the digits shown.

ξth a d0 d1 d2 d3 d4 d5 d6 (0.5,1.4) 514.390 1038.33 2088.56 −1688.70 350.282 −51.8325 4.42276 −0.162045 (0.8,1.5) 2.0 −495.283 −715.928 −1761.03 1722.13 −428.157 76.4943 −7.78341 0.335614 (1.0,1.6) 1575.17 3029.17 6352.80 −5460.10 1230.94 −204.048 19.9068 −0.843217 (0.5,1.4) 116.0 54.43 −1.989 10.44 − − − − (0.8,1.5) 5.0 102.0 55.85 0.6177 13.15 − − − − (1.0,1.6) 87.28 54.58 3.639 15.03 − − − −

TABLE IV: Fitting parameters for the quark matter equations of state at ξ = nB /n0 ≥ 5.0 for various sets of (gV ,H)/G with 0 3 K = 0 (see text). The coeﬃcients are in units of MeV/fm , and lν = ν.

gV /G H/G a d0 d1 d2 1.4 116.0 54.31 −2.020 12.06 0.7 1.5 103.6 53.07 −0.1710 12.39 1.6 88.36 52.73 3.047 12.65 1.4 115.9 54.50 −2.000 12.87 0.8 1.5 102.0 55.85 0.6177 13.15 1.6 85.49 57.77 4.537 13.35 1.4 116.8 52.79 −2.422 13.71 0.9 1.5 105.3 50.06 −1.081 14.08 1.6 87.03 55.01 3.749 14.22 1.4 114.4 57.00 −1.146 14.42 1.0 1.5 99.61 60.04 1.892 14.68 1.6 87.28 54.58 3.640 15.03

The APR equation of state has a low and a high den- exponents in Eq. C1 to be integers, lν = ν = 2, . . . , νmax, sity regime, distinguished by a first order phase transition with νmax = 6. In the quark matter domain, we fit associated with the onset of neutral pion condensation; the data for 5n0 ≤ nB ≤ 10n0 and find that taking in matter in beta equilibrium the density changes discon- only one term, νmax = 2, yields adequate fits (Fig. 36). −3 tinuously from nB ' 1.29n0 ' 0.21 fm to ' 1.50n0 ' TableIII summarizes the fitting parameters for uni- 0.24 fm−3. The low density parametrization (LDP) is for fied equations of state for K0 = 0 and parameter sets the range 0.26n0 . nB . 1.29n0, and the high density (gV /G, H/G) =(0.5,1.4), (0.8,1.5), and (1.0,1.6), as used parametrization (HDP), the range 1.50n0 . nB . 2.0n0. in the calculations shown in Fig. 28. The coefficients for The error estimator δ ≡ Pfit/Pdata − 1, shown in Fig. 35, the interpolated domain strongly correlate one another. does not exceed ±0.01, for the fits. To maintain the good accuracy of fits as in Fig. 28, we must keep 5-6 digits in the fitting parameters.

The quark model parameters (gV ,H) chosen here are 3. Unified equations of state (QHC18) suitable for interpolating with the APR equation of state. For interpolation with other nuclear equations of state or In the interpolation region nL = 2n0 ≤ nB ≤ nU = other possible interpolation schemes, it would be better 5n0, we parametrize the equations of state by taking the to use slightly different sets of parameters. For this rea- 34 son we have generated numerical equations of state for a wider range of (gV ,H) which are fitted by Eq. C1. Table IV summarizes the fitting parameters. The present equation of state QHC18, for several choices of QCD parameters, is given together with instructions for use on the website: Home Page of Rela- tivistic EOS table for supernovae (http://user.numazu- ct.ac.jp/∼sumi/eos/index.html). 35

[1] See, e.g., P. Haensel, A. Y. Potekhin, and D. G. 8 (2016). Yakovlev, Neutron stars 1: Equation of state and struc- [19] S. Bogdanov, J.E. Grindlay, and G.B. Rybicki, “Ther- ture (Springer-Verlag, New York ) 2007). mal X-rays from millisecond pulsars: constraining the [2] G. Baym and C. J. Pethick, “Neutron Stars,” Annu. fundamental properties of neutron stars,” Astrophys. J. Rev. Nucl. Sci. 25, 27-77 (1975); “Physics of neu- 689, 407-415 (2008). tron stars,” Annu. Rev. Astron. Astrophys. 17, 415-43 [20] A. L. Watts, N. Andersson, D. Chakrabarty, M. Fe- (1979). roci, J. Hebeler, G. Israel, F. K. Lamb, M. C. Miller, S. [3] T. Hell and W. Weise, “Dense baryonic matter: con- Morsink, F. Ozel,¨ A. Patruno, J. Poutanen, D. Psaltis, straints from recent neutron star observations,” Phys. A. Schwenk, A. W. Steiner, L. Stella, L. Tolos, and M. Rev. C 90 (2014), 045801 (2014). van der Klis, “Measuring the neutron star equation of [4] J. M. Lattimer, “The nuclear equation of state and neu- state using x-ray timing,” Rev. Mod. Phys. 88, 021001: tron star masses,” Ann. Rev. Nucl. Part. Sci. 62, 485- 1-23 (2016). 515 (2012). [21] K. D. Kokkotas, T. A. Apostolatos, and N. Andersson, [5]F. Ozel¨ and P. Freire, “Masses, radii, and the equation “The inverse problem for pulsating neutron stars: a ‘fin- of state of neutron stars, Annu. Rev. Astron. Astrophys. gerprint analysis’ for the supranuclear equation of state. 54, 401-440 (2016). Mon. Not. R. Astron. Soc. 320, 307-315 (2001). [6] C. M. Miller, and F. K. Lamb, “Observational con- [22] B. P. Abbott et al. (LIGO Scientific Collaboration straints on neutron star masses and radii,” Eur. Phys. and Virgo Collaboration), “GW170817: Observation of J. A 52, 69, 63: 1-20 (2016). gravitational waves from a binary neutron star inspiral,” [7]F. Ozel,¨ G. Baym, and T. Güver, “Astrophysical mea- Phys. Rev. Lett. 119, 161101:1-18 (2017). surement of the equation of state of neutron star mat- [23] B. P. Abbott et al., “Multi-messenger observations of a ter,” Phys. Rev. D 82, 101301(R) (2010). binary neutron star merger,” Astrophys. J. Letters 848, [8] A. W. Steiner, J. M. Lattimer, and E. F. Brown, “The L12:1-59 (2017), and following papers in Astrophys. J. equation of state from observed masses and radii of neu- Letters 848. tron stars,” Astrophys. J, 722, 33-54 (2010). [24] K. Hotokezaka, K. Kyutoku, H. Okawa, M. Shibata, and [9] A. W. Steiner, J. M. Lattimer, and E. F. Brown, “The K. Kiuchi, “Binary Neutron Star Mergers: Dependence neutron star mass-radius relation and the equation of on the Nuclear Equation of State,” Phys. Rev. D 83, state of dense matter,” Astrophys. J. Letters 765, L5: 124008 (2011); arXiv:1105.4370 [astro-ph.HE]. 1-5 (2013). [25] K. Takami, L. Rezzolla, and L. Baiotti, “Constraining [10] J. M. Lattimer and A, W. Steiner, “Neutron star masses the equation of state of neutron stars from binary merg- and radii from quiescent low-mass X-ray binaries, ”As- ers,” Phys. Rev. Letters 113, 091104 (2014). trophys. J, 784, 123-137 (2014). [26] L. Baiotti and L. Rezzolla, “Binary neutron-star [11]F. Ozel,¨ D. Psaltis, T. Güver, G. Baym, C. Heinke, and mergers: a review of Einstein’s richest labo- S. Guillot, “The dense matter equation of state from ratory,” Repts. Prog. Phys. (in press) (2017), neutron star radius and mass measurements,” Astro- https://doi.org/arXiv:1607.03540 [gr-qc]. phys. J. 820, 28:1-25 (2016). [27] M. Shibata, ”Merger of binary neutron stars: Gravita- [12] A. W. Steiner, S. Gandolfi, F. J. Fattoyev and tional waves and electromagnetic counterparts,” Nucl. W. G. Newton, “Using neutron star observations to de- Phys. A 956, 225-232 (2016). termine crust thicknesses, moments of inertia, and tidal [28] V. Paschalidis, M. Ruiz and S. L. Shapiro, “Relativis- deformabilities,” Phys. Rev. C 91, 015804:1-7 (2015). tic simulations of black hole-neutron star coalescence: [13] A. W. Steiner, J. M. Lattimer, and E. F. Brown, “Neu- the jet emerges,” Astrophys. J. Letters 806, L14:1-5, tron star radii, universal relations, and the role of prior (2015). distributions,” Euro. Phys.. J, A52, 18:1-16 (2016). [29] M. Ruiz, R. Lang, V. Paschalidis and S. L. Shapiro, [14] D. Alvarez-Castillo, A. Ayriyan, S. Benic, D. Blaschke, Binary neutron star mergers: a jet engine for short H. Grigorian and S. Typel, “New class of hybrid EoS gamma-ray bursts, Astrophys. J. Letters 824, L1:1-5 and Bayesian M-R data analysis,” Eur. Phys. J. A 52, (2016). 69: 1-11 (2016). [30] S. Bernuzzi, D. Radice, C. D. Ott, L. F. Roberts, [15] K. C. Gendreau et al., “The Neutron star Interior Com- P. Moesta and F. Galeazzi, “How loud are neutron star position Explorer (NICER): design and development,” mergers?,” Phys. Rev. D 94, 024023:1-6 (2016). Proc. SPIE 9905, Space Telescopes and Instrumenta- [31] B. P. Abbott et al., ”GW150914: The advanced LIGO tion 2016: Ultraviolet to Gamma Ray, 99051H (July detectors in the era of first discoveries, Phys. Rev. Lett. 22, 2016). 116, 131103:1-12 (2016). [16] M. Bauböck, D. Psaltis, and F. Ozel,¨ “Effects of spot [32] B. P. Abbott et al. (LIGO Scientific Collaboration size on neutron-star radius measurements from pulse and Virgo Collaboration), “GW151226: Observation of profiles, Astrophys. J. 811, 144:1-8 (2015). Gravitational Waves from a 22-Solar-Mass Binary Black [17] M. C. Miller, “The case for PSR J1614-2230 as a NICER Hole Coalescence,” Phys. Rev. Letters 116, 241103 target,” Astrophys. J. 822, 27:1-7 (2016). (2016). [33] B. P. Abbott et al. (LIGO Scientific Collaboration and [18]F. Ozel,¨ D. Psaltis,, Z. Arzoumanian, S. Morsink, and Virgo Collaboration), “Astrophysical implications of the M. Bauböck, “Measuring neutron star radii via pulse binary black hole merger GW150914,” Astrophys. J. profile modeling with NICER,” Astrophys. J. 832, 92:1- Letters 818, L22:1-15 (2016). 36

[34] F. Acernese et al., “Advanced Virgo: A second- Monte Carlo methods for nuclear physics,” Rev. Mod. generation interferometric gravitational wave detector,” Phys. 87, 1067-118 (2015). Class. Quantum Grav. 32, 024001:1-55 (2015). [55] S. Gandolfi, J. Carlson, and S. Reddy, “Maximum mass [35] J. Hough et al. “Proposal for a Joint German-British and radius of neutron stars, and the nuclear symmetry Interferometric Gravitational Wave Detector,” available energy,” Phys. Rev. C 85, 032801 (2012). at eprints.gla.ac.uk/114852/7/114852.pdf. [56] S. Gandolfi, J. Carlson, S. Reddy, A. W. Steiner and [36] Y. Aso et al. “Interferometer design of the KAGRA R. B. Wiringa, “The equation of state of neutron mat- gravitational wave detector,? Phys Rev D 88, 043007:1- ter, symmetry energy, and neutron star structure,” Eur. 15 (2013). Phys. J. A 50 10:1-11 (2014). [37] B. Iyer et al. “LIGO-INDIA: Proposal for [57] See, e.g., K. Hebeler, S. K. Bogner, R. J. Furnstahl, A. an interferometric gravitational-wave observa- Nogga, and A. Schwenk, Phys. Rev. C 83, 031301(R) tory,” LIGO Public Documents, Available at (2011). https://dcc.ligo.org/public/0075/M1100296/002/LIGO- [58] N. Kaiser, “Chiral four-body interactions in nuclear India lw-v2.pdf (2011). matter,” Euro. Phys. J. A 48, 135-45 (2012). [38] See, Proc. 11th Int, LISA Symposium, J. Phys. Conf. [59] S. Fritsch, N. Kaiser and W. Weise, “Chiral approach Series 840, 2017. to nuclear matter: Role of two-pion exchange with vir- [39] M. McLaughlin, “The international pulsar timing array: tual delta-isobar excitation,” Nucl. Phys. A 750, 259-93 a galactic scale gravitational wave observatory,” Gen (2005). Relativ Gravit. 46 1810:1-22 (2014). [60] A. Bodmer, “Collapsed Nuclei,” Phys. Rev. D 4, 1601-6 [40] E. Fonseca et al., “The NANOGrav nine-year data set: (1971). mass and geometric measurements of binary millisecond [61] E. Witten, “Cosmic separation of phases,” Phys. Rev. pulsars,” Astrophys. J. 832, 167:1-22 (2016). D 30, 272-85 (1984). [41] P. Demorest, T. Pennucci, S. M. Ransom, R. M. S. E., [62] E. Farhi and R. L. Jae, “Strange matter,” Phys. Rev. and J. W. T. Hessels, “A two-solar-mass neutron star D 30, 2379-90 (1984). measured using Shapiro delay,” Nature (London) 467, [63] M. Baldo, G. Burgio, and H. Schulze, “Hyperon stars 1081-3 (2010). in the Brueckner-Bethe-Goldstone theory,” Phys. Rev. [42] J. Antoniadis, et al., “A massive pulsar in a compact C 61, 055801 (2000). relativistic binary,” Science 340, 1233232 (2013). [64] S. Nishizaki, Y. Yamamoto, and T. Takatsuka, “Effec- [43] M. H. van Kerkwijk, R. P. Breton, and S. R. Kulka- tive YN and YY interactions and hyperon-mixing in rni, “Evidence for a massive neutron star from a radial- neutron star matter – Y ≡ Λ case,” Prog. Theor. Phys. velocity study of the companion to the black widow pul- 105, 607-26 (2001). sar PSR B1957+20,” Astrophys. J. 728, 95:1-8 (2011). [65] S. Nishizaki, Y. Yamamoto, and T. Takatsuka, [44] Joshua Schroeder and Jules Halpern, “Observations and “Hyperon-mixed neutron star matter and neutron modeling of the companions of short period binary mil- stars,” Prog. Theor. Phys. 108, 703-18 (2002). lisecond pulsars: Evidence for high-mass neutron stars,” [66] H. Schulze and T. Rijken, “Maximum mass of hyperon Astrophys. J. 793, 78:1-12 (2014). stars with the Nijmegen ESC08 model,” Phys. Rev. C [45] R. W. Romani, A. V. Filippenko, J. M. Silverman, S. B. 84, 035801 (2011). Cenko, J. Greiner, A. Rau, J. Elliott, and H. J. Pletsch, [67] C. J. Pethick, T. Schäfer, and A. Schwenk, “Bose- “PSR J1311-3430: A heavyweight neutron star with Einstein condensates in neutron stars,” in in Univer- a flyweight helium companion,” Astrophys. J. Letters sal Themes of Bose-Einstein Condensation, D. Snoke, 760, L36: 1-6 (2012). N. Proukakis, and P. Littlewood, eds, Cambridge Univ. [46] R. W. Romani, A. V. Filippenko, and S. B. Cenko, ”A Press, Cambridge 2017). pp. 573-592; arXiv:1507.05839. spectroscopic study of the extreme black widow PSR [68] A. B. Migdal, “Stability of vacuum and limiting fields,” J1311-3430,” Astrophys. J. 804, 115R: 1-10 (2015). Zh. Eksp. Teor. Fiz. 61, 2209-24 (1971) [Sov. Phys. [47] K. Fukushima and T. Hatsuda, “The phase diagram of JETP 34, 1184-91 (1972)]; “Pion Fields in Nuclear Mat- dense QCD,” Repts. Prog. Phys. 74, 014001 (2011). ter,” Rev. Mod. Phys. 50, 107-72 (1978). [48] M. Asakawa and K. Yazaki, “Chiral restoration at finite [69] S. Barshay, G. Vagradov, and G. E. Brown, “Possibility density and temperature,” Nucl. Phys. A 504, 668-84 of a phase transition to a pion condensate in neutron (1989). stars,” Phys. Letters B 43 5, 359 201361 (1973). [49] T. Hatsuda, M. Tachibana, N. Yamamoto, and G. [70] G. Baym, “Pion condensation in nuclear and neutron Baym, “New critical point Induced by the axial anomaly star matter,” Phys. Rev. Letters 30, 1340-2 (1973). in dense QCD,” Phys. Rev. Letters 97, 122001 (2006). [71] R .F. Sawyer and D. J. Scalapino, “Pion condensation [50] R. C. Tolman, “Static solutions of Einstein’s field equa- in superdense nuclear matter,” Phys. Rev. D 7 4, 953-64 tions for spheres of fluid,” Phys. Rev. 55, 364-73 (1939). (1973). [51] J. Oppenheimer and G. Volkoff, “On massive neutron [72] A. Mukherjee, “Variational theory of hot nucleon mat- cores,” Phys. Rev. 55, 374-81 (1939). ter. II. Spin-isospin correlations and equation of state of [52] A. Akmal, V. R. Pandharipande, and D.G. Ravenhall, nuclear and neutron matter,” Phys. Rev. C 79, 045811 “Equation of state of nucleon matter and neutron star (2009). structure,” Phys. Rev. C 58, 1804-28 (1998). [73] D. B. Kaplan and A. E. Nelson, “Strange goings on in [53] J. Morales, Jr., V.R. Pandharipande, and D.G. Raven- dense nucleonic matter,” Phys. Letters B 175 1, 57-63 hall, “Improved variational calculations of nucleon mat- (1986). ter,” Phys. Rev. C 66, 054308 (2002). [74] V. R. Pandharipande, C. J. Pethick, and V. Thorsson, [54] J. Carlson, S. Gandolfi, F. Pederiva, S. C. Pieper, R. “Kaon energies in dense matter,” Phys. Rev. Letters 75, Schiavill, K.E. Schmidt, and R. B. Wiringa, ”Quantum 4567-70 (1995). 37

[75] H.A. Bethe and G.E. Brown, “Evolution of binary com- “Color superconductivity in dense quark matter,” Rev. pact objects that merge,” Astrophys. J. 506, 780-9 Mod. Phys. 80, 1455-515 (2008). (1998), and references therein. [97] M. A. Shifman, A. I. Vainshtein and V. I. Zakharov, [76] N.K. Glendenning and J. Schaffner-Bielich, “First order “QCD and resonance physics. Theoretical foundations,” kaon condensate,” Phys. Rev. C 60, 025803 (1999). Nucl. Phys. B 147 385-447 (1979). [77] H. Dapo, B.-J. Schaefer, and J. Wambach, “Appearance [98] D. Ivanenko and D. F. Kurdgelaidze, “Remarks on of hyperons in neutron stars,” Phys. Rev. C 81, 035803 quark stars,” Lett. Nuovo Cim. 2. 13-16 (1969), and (2010). references therein. [78] S. Weissenborn, D. Chatterjee and J. Schaffner-Bielich, [99] N. Itoh, ‘Hydrostatic equilibrium of hypothetical quark “Hyperons and massive neutron stars: Vector repulsion stars,” Prog. Theor. Phys. 44, 291-292 (1970). and SU(3) symmetry,” Phys. Rev. C 85, 065802 (2012). [100] G. Baym and S. A. Chin, “Can a neutron star be a giant [79] S. Weissenborn, D. Chatterjee and J. Schaffner-Bielich, MIT bag?,” Phys. Letters B 62, 241-4 (1976). “Hyperons and massive neutron stars: the role of hy- [101] M. Buballa, “NJL-model analysis of dense quark mat- peron potentials,” Nucl. Phys. A 881, 62-77 (2012). ter,” Phys. Rep. 407, 205-376 (2005). [80] D. Lonardoni, A. Lovato, S. Gandolfi, and F. Pederiva, [102] S. Benic, D. Blaschke, D. E. Alvarez-Castillo, T. Fischer “Hyperon puzzle: hints from quantum Monte Carlo Cal- and S. Typel, “A new quark-hadron hybrid equation culations,” Phys. Rev. Letters 114, 092301 (2015). of state for astrophysics - I. High-mass twin compact [81] A. Ukawa, “Kenneth Wilson and lattice QCD,” J. stars,” Astron. Astrophys. 577, A40: 1-8 (2015). Statist. Phys. 160, 1081-1124 (2015). [103] D. Alvarez-Castillo, S. Benic, D. Blaschke, S. Han and [82] T. Inoue (HAL QCD Collaboration), “Hyperon single- S. Typel, “Neutron star mass limit at 2M supports particle potentials from QCD on lattice,” Proc. 26th the existence of a CEP,” Eur. Phys. J. A 52, 232: 1-8 Int. Nucl. Phys. Conf. (INPC2016), Adelaide, Australia, (2016). arXiv:1608.02425 [nucl-th]. Sept. 2016, arXiv:1612.08399. [104] I. F. Ranea-Sandoval, S. Han, M. G. Orsaria, G. A. Con- [83] S.A. Chin and J.D. Walecka, “ An equation of state for trera, F. Weber and M. G. Alford, “Constant-sound- nuclear and higher-density matter based on a relativistic speed parametrization for Nambu-Jona-Lasinio models mean-field theory,” Phys. Letters B52 24-28 (1974). of quark matter in hybrid stars,” Phys. Rev. C 93, [84] H. Müllerand B.D. Serot, “Relativistic mean-field the- 045812 (2016). ory and the high-density nuclear equation of state,” [105] T. Schäferand F. Wilczek, “Continuity of quark and Nucl. Phys. A 606, 508-37 (1996). hadron matter,” Phys. Rev. Letters 82, 3956-9 (1999). [85] J. M. Lattimer and M. Prakash, “Neutron star obser- [106] N. Bratovic, T. Hatsuda, and W. Weise, “Role of vector vations: Prognosis for equation of state constraints,” interaction and axial anomaly in the PNJL modeling of Phys. Rep. 442, 109-65 (2007). the QCD phase diagram,” Phys. Letters B 719, 131-5 [86] G. Baym, “Confinement of quarks in nuclear matter,” (2012). Physica 96A, 131-135 (1979). [107] O. Lourenco, M. Dutra, T. Frederico, A. Delfino, and [87] T. Celik, F. Karsch, and H. Satz, “A percolation ap- M. Malheiro, “Vector interaction strength in Polyakov- proach to strongly interacting matter,” Phys. Letters B Nambu-Jona-Lasinio models from hadron-quark phase 97, 128-30 (1980). diagrams,” Phys. Rev. D 85, 097504 (2012). [88] H. Satz, “Deconfinement and percolation,” Nucl. Phys. [108] K. Masuda, T. Hatsuda, and T. Takatsuka, “Hadron- A 642, c130-c142 (1998); Repts. Prog. Phys. 63 ,1511 quark crossover and massive hybrid stars with (2000); Extreme States of Matter in Strong Interaction strangeness,” Astrophys. J. 764,12: 1-5 (2013). Physics (Springer 2012). [109] K. Masuda, T. Hatsuda, and T. Takatsuka, “Hadron- [89] Y. Nambu and G. Jona-Lasinio, “Dynamical model of quark crossover and massive hybrid stars,” Prog. Theor. elementary particles based on an analogy with super- and Exp. Phys. 7, 073D01 (2013). conductivity. 1,” Phys. Rev. 122, 345-58 (1961). [110] T. Klähn, R.Lastowiecki, and D. Blaschke, “Impli- [90] J. W. Holt, M. Rho and W. Weise, “Chiral symme- cations of the measurement of pulsars with two solar try and effective field theories for hadronic, nuclear and masses for quark matter in compact stars and heavy-ion stellar matter,” Phys. Rept. 621, 2-75 (2016). collisions: A Nambu-Jona-Lasinio model case study,” [91] T. Krüger,I. Tews, B. Friman, K. Hebeler, and A. Phys. Rev. D 88, 085001 (2013). Schwenk, “The chiral condensate in neutron matter, [111] K. Masuda, T. Hatsuda and T. Takatsuka, “Hyperon Phys. Letters B726, 412-416 (2013). puzzle, hadron-quark crossover and massive neutron [92] D.T. Son and M. A. Stephanov, “Inverse meson mass stars,” Eur. Phys. J. A 52, 65-79 (2016). ordering in the color-flavor-locking phase of high-density [112] P. Haensel and A. Y. Potekhin, “Analytical representa- QCD,” Phys. Rev. D 61, 074012 (2000). tions of unified equations of state of neutron-star mat- [93] K. Fukushima, “Quark description of the Nambu- ter,” Astron. Astrophys. 428 ,191-197 (2004); tabula- Goldstone Bosons in the color-flavor-locked phase,” tions given at http://www.ioffe.ru/astro/NSG/NSEOS. Phys. Rev. D 70, 094014 (2004). [113] G. Baym, H. A. Bethe, and C. J. Pethick, “Neutron star [94] N. Yamamoto, M. Tachibana, T. Hatsuda, and G. matter,” Nucl. Phys. A 175, 225-71 (1971). Baym, “Phase structure, collective modes, and the ax- [114] F. Douchin and P. Haensel, “A unified equation of state ial anomaly in dense QCD,” Phys. Rev. D 76, 074001 of dense matter and neutron star structure,” Astron. (2007). Astrophys. 380, 151-67 (2001). [95] Y. Song and G. Baym, “Generalized Nambu-Goldstone [115] H. Togashi, K. Nakazato, Y. Takehara, S. Yamamuro, pion in dense matter: a schematic NJL model,” Phys. H. Suzuki, and M. Takano, “Nuclear equation of state Rev. C (submtted 2017); arXiv:1703.08236 [nucl-th]. for core-collapse supernova simulations with realistic [96] M. G. Alford, A. Schmitt, K. Rajagopal and T. Schäfer, nuclear forces,” Nucl. Phys. A 961, 78-105 (2017). 38

(http://www.np.phys.waseda.ac.jp/EOS/) the quark gluon plasma. 1. Entropy and density,” Phys. [116] E. S. Fraga, A. Kurkela and A. Vuorinen, “Interacting Rev. D 63 065003 (2001). quark matter equation of state for neutron stars,” As- [137] J. O. Andersen, L. E. Leganger, M. Strickland and trophys. J. Letters, 781, L25: 1-5 (2014). N. Su, “Three-loop HTL QCD thermodynamics,” JHEP [117] A. Kurkela, E. S. Fraga, J. Schaffner-Bielich and 2011, 53-77 (2011). A. Vuorinen, “Constraining neutron star matter with [138] K. Fukushima and T. Kojo, “The quarkyonic star,” As- Quantum Chromodynamics,” Astrophys. J. 789, 127- trophys. J. 817, 180-8 (2016). 33 (2014). [139] U. Vogl and W. Weise, “The Nambu and Jona Lasinio [118] T. Kojo, P. D. Powell, Y. Song, and G. Baym, “Phe- model: Its implications for hadrons and nuclei,”Prog. nomenological QCD equation of state for massive neu- Part. Nucl. Phys. 27, 195-272 (1991). tron stars,” Phys. Rev. D 91, 045003 (2015). [140] M. Lutz, S. Klimt, and W. Weise, “Meson properties at [119] T. Kojo, “Phenomenological neutron star equations of finite temperature and baryon density,” Nucl. Phys. A state: 3-window modeling of QCD matter,” Eur. Phys. 542, 521-58 (1992). J. A 52, 51-69 (2016). [141] S. P. Klevansky,“The Nambu-Jona-Lasinio model of [120] T. Kojo, P. D. Powell, Y. Song, and G. Baym, “Phe- quantum chromodynamics,” Rev. Mod. Phys. 64, 649- nomenological QCD equations of state for neutron 708 (1992). stars,” Proc. QM2015 Kobe, Nucl. Phys. A 956, 821-5 [142] T. Hatsuda and T. Kunihiro, “QCD phenomenology (2016). based on a chiral effective Lagrangian,” Phys. Rep. 247, [121] D. L. Whittenbury, J. D. Carroll, A. W. Thomas, 221-367 (1994). K. Tsushima and J. R. Stone, “Quark-Meson Coupling [143] P. Rehberg, S. P. Klevansky and J. Hufner, “Hadroniza- Model, Nuclear Matter Constraints and Neutron Star tion in the SU(3) Nambu-Jona-Lasinio model,” Phys. Properties,” Phys. Rev. C 89, 065801 (2014). Rev. C 53, 410-29 (1996). [122] D. L. Whittenbury, H. H. Matevosyan and [144] P. D. Powell and G. Baym, “Axial anomaly and the A. W. Thomas, “Hybrid stars using the quark- three-flavor Nambu–Jona-Lasinio model with confine- meson coupling and proper-time Nambu-Jona-Lasinio ment: Constructing the QCD phase diagram,” Phys. models,” Phys. Rev. C 93, 035807 (2016). Rev. D 85, 074003 (2012). [123] A. M. Polyakov, “Thermal properties of gauge fields and [145] P. D. Powell and G. Baym, “Asymmetric pairing of real- quark liberation,” Phys. Letters B 72, 477-80 (1978). istic mass quarks and color neutrality in the Polyakov– [124] L. Susskind, “Lattice models of quark confinement at Nambu–Jona-Lasinio model of QCD,” Phys. Rev. D 88, high temperature,” Phys. Rev. D 20, 2610-8 (1979). 014012 (2013). [125] T. Banks and E. Rabinovici, “Finite temperature be- [146] H. Abuki, G. Baym, T. Hatsuda, and N. Yamamoto, havior of the lattice Abelian Higgs model,” Nucl. Phys. “Nambu–Jona-Lasinio model of dense three-flavor mat- B 160, 349-79 (1979). ter with axial anomaly: The low temperature critical [126] B. Svetitsky, “Symmetry aspects of finite temperature point and BEC-BCS diquark crossover,” Phys. Rev. D confinement transitions,” Phys. Rep. 132, 1-53 (1986). 81, 125010 (2010). [127] Y. Aoki, G. Endrodi, Z. Fodor, S. D. Katz and K. K. Sz- [147] T. Takatsuka, T. Hatsuda, and K. Masuda, “Massive abo, “The order of the quantum chromodynamics tran- hybrid stars with strangeness,” [arXiv:1402.4677 [nucl- sition predicted by the standard model of particle th]] (2013). 7th Intl. Symp. on Chiral Symmetry in physics,” Nature 443 675-678 (2006). Hadrons and Nuclei (CHIRAL 13) Oct. 2013, Beijing [128] A. Bazavov et al. (HotQCD Collaboration), “Chiral and (World Scientific Publ. Co., 2015). deconfinement aspects of the QCD transition,” Phys. [148] M. G. Alford and S. Han, “Characteristics of hybrid Rev. D 85 054503 (2012). compact stars with a sharp hadron-quark interface,” [129] A. Bazavov (HotQCD collaboration), “Equation of state Eur. Phys. J. A 52, 62-76 (2016). in ( 2+1 )-flavor QCD,” Phys. Rev. D 90, 094503 (2014). [149] D. Psaltis, F. Ozel,¨ and S. DeDeo, ”Photon propagation [130] S. Borsanyi, Z. Fodor, C, Hoelbling, S. D. Katz, S. around compact objects and the inferred properties of Krieg, and K. K. Szabo, “Full result for the QCD equa- thermally emitting neutron stars,” Astrophys. J. 544 tion of state with 2+1 flavors”, Phys. Letters B 370, 390-6 (2000). 99-104 (2014). [150]F. Ozel,¨ “Surface emission from neutron stars and impli- [131] C. E. Detar, “A conjecture concerning the modes of cations for the physics of their interiors,” Repts. Prog. excitation of the quark-gluon plasma,” Phys. Rev. D Phys. 76, 016901 (2013). 32, 276-83 (1985). [151] D. Lai, E. E. Salpeter, and S. L. Shapiro, “Hydrogen [132] T. Hatsuda and T. Kunihiro, “Fluctuation effects in hot molecules and chains in a superstrong magnetic field, quark matter: precursors of chiral transition at finite ”Phys. Rev. A 45 4832-4847 (1992). temperature,” Phys. Rev. Letters 55 158-61 (1985). [152]F. Ozel,¨ “Surface emission properties of strongly mag- [133] R. D. Pisarski, “Quark gluon plasma as a condensate of netic neuton stars,” Astrophys. J. 563, 276-288 (2001). SU(3) Wilson lines,” Phys. Rev. D 62, 111501 (2000). [153] W. DeGottardi, Q. Meng, S. Vishveshwara, and G. [134] A. Dumitru, Y. Guo, Y. Hidaka, C. P. K. Altes and Baym, “Surface radiation from magnetized neutron R. D. Pisarski, “How wide is the transition to decon- stars,” to be submitted. finement?,” Phys. Rev. D 83, 034022 (2011). [154] S. L. Shapiro, “Black holes, disks, and jets following [135] R. D. Pisarski and V. V. Skokov, “A chiral matrix model binary mergers and stellar collapse: the narrow range of of the semi-quark gluon plasma in QCD,” Phys. Rev. D electromagnetic luminosities and accretion rates,” Phys. 94, 034015 (2016). Rev. D 95, 101303(R):1-6 (2017). [136] J. P. Blaizot, E. Iancu and A. Rebhan, “Approximately [155] B. D. Metzger, “‘Welcome to the multi-messenger era! selfconsistent resummations for the thermodynamics of 39

Lessons from a neutron star merger and the landscape [174] D.G. Yakovlev, K.P. Levenfish, Yu.A. Shibanov, “Cool- ahead,” arXiv:1710.05931. ing neutron stars and superfluidity in their interiors,” [156] B. Margalit and B. D. Metzger, “Constraining the max- Uspekhi Fizicheskikh Nauk 169 825-68 (1999); Physics imum mass of neutron stars from multi-messenger ob- Uspekhi 42, 737-778 (1999). servations of GW170817,” arXiv:1710.05938. [175] D. Page, J. M. Lattimer, M. Prakash and A. W. Steiner, [157] M. Ruiz, S. L. Shapiro, and A. Tsokaros, “GW170817, “Stellar superfluids” in Novel superfluids, Vol. 2, Ed. General relativistic magnetohydrodynamic simula- by K-H. Bennemann and J. B. Ketterson (Int. Series tions, and the neutron star maximum mass,” Monographs on Phys., Oxford Univ. Press). arXiv:1711.00473. [176] R. Wijnands, N. Degenaar and D. Page, “Cooling of [158] L. Rezzolla, E. R. Most, and L. R. Weih, “Using accretion-heated neutron stars,” J. Astrophys. Astron. gravitational-wave observations and quasi-universal re- 38, 49 (2017). lations to constrain the maximum mass of neutron stars, [177] A. Cumming, E. F. Brown, F, J. Fattoyev, C. J. arXiv:1711.00314. Horowitz, D. Page, and S, Reddy, “Lower limit on the [159] S. Lawrence, J. G. Tervala, P. F. Bedaque, and M. C. heat capacity of the neutron star core,” Phys. Rev. C Miller, “An upper bound neutron star masses from 95, 025806 (2017). models of short gamma-ray bursts,” Astrophys. J. 808, [178] D. Psaltis and F. Ozel,¨ “Pulse profiles from spinning 186:1-7 (2015). neutron stars in the Hartle-Thorne approximation,” As- [160] M. Shibata, S. Fujibayashi, K. Hotokezaka, K. Kiuchi, trophys. J, 792, 87-93 (2014). K. Kyutoku, Y. Sekiguchi, and M. Tanaka, “GW170817: [179] L. Rezzolla, F. K. Lamb, S. L Shapiro, “R-mode oscil- Modeling based on numerical relativity and its implica- lations in rotating magnetic neutron stars,” Astrophys. tions,” arXiv:1710.07579. J. Letters 531, L139-L142 (2000). [161] A. Bauswein, O. Just, H.-T. Janka, and N. Stergioulas, [180] M. Alford, “Analysis of how pulsar timing data can con- “Neutron-star radius constraints from GW170817 and strain the composition of a neutron star via r-mode os- future detections,” arXiv1710.06843. cillations,” Phys. Rev. Letters 113 251102:1-5 (2014). [162] D. Radice, A. Perego, F. Zappa, and S. Bernuzzi, [181] A.L. Watts, and T. E. Strohmayer (2006), “Detection “GW170817: Joint contstraint on the neutron star with RHESSI of high-hrequency X-ray oscillations in equation of state from multimessenger observations,” the tail of the 2004 hyperflare from SGR 1806-20,” As- arXiv:1711.03647. trophys. J. 637, L117-L120 (2006). [163] E. Poisson and C. M. Will, Gravity: newtonian, post- [182] M. C. Miller, “QPO constraints on neutron stars, New newtonian, relativistic, Cambridge Univ. Press, Cam- Astronomy Reviews 54, 128-134 (2010). bridge, 2014, Eq. (2.249). [183] G. Baym, C. J. Pethick and P. Sutherland, “The ground [164] K. S. Thorne, “Tidal stabilization of rigidly rotat- state of matter at high densities: Equation of state and ing, fully relativistic neutron stars,” Phys. Rev. D 58, stellar models,” Astrophys. J. 170, 299-317 (1971). 124031(1998). [184] P. Haensel, J. L. Zdunik, and F. Douchin, “Equation [165] W. H. Munk and G. J. F. MacDonald, The rotation of of state of dense matter and the minimum mass of cold the earth: a geophysical discussion, Cambridge Univ. neutron stars,” Astron. Astrophys. 385, 301-307 (2002). Press, Cambridge, 1960, Eq. (5.7.1). [185] D. G. Ravenhall, C. J. Pethick, and J. R. Wilson, [166] T. Hinderer, “Tidal Love numbers of neutron stars,” “Structure of matter below nuclear saturation density,” Astrophys. J. 677 1216-20 (2008); erratum Astrophys. Phys. Rev. Letters 50, 2066-9 (1983). J. 697, 964 (2009). [186] M. Hashimoto, H. Seki, and M. Yamada, “Shape of nu- [167] E.´ E.´ Flanagan and T. Hinderer, “Constraining neutron- clei in the crust of neutron star,” Prog. Theor. Phys. star tidal Love numbers with gravitational-wave detec- 71, 320-6 (1984). tors,” Phys. Rev. D77, 021502(R) (2008). [187] K. Oyamatsu, M. Hashimoto, and M. Yamada, “Fur- [168] T. Hinderer, B. D. Lackey, R. N. Lang and J. S. Read, ther study of the nuclear shape in high-density matter,” “Tidal deformability of neutron stars with realistic Prog. Theor. Phys. 72, 373-5 (1984). equations of state and their gravitational wave sig- [188] R. D. Williams and S. E. Koonin, “Sub-saturation natures in binary inspiral,” Phys. Rev. D81, 123016 phases of nuclear matter,” Nucl. Phys. A 435, 844-58 (2010). (1985). [169] J S.Read, L. Baiotti, J. D. E. Creighton, J .L. Friedman, [189] G. Watanabe, K. Iida, and K. Sato, “Thermodynamic B. Giacomazzo, K. Kyutoku, C Markakis, L. Rezzolla, properties of nuclear ’pasta’ in neutron star crusts,” M, Shibata, and K. Taniguchi, “Matter effects on bi- Nucl. Phys. A 676, 455-73 (2000); ibid. A 687, 512-31 nary neutron star waveforms,” Phys. Rev. D88, 044042 (2001); Erratum-ibid. A 726, 357-65 (2003) . (2013). [190] C. J. Horowitz, M. A. Perez-Garcia and J. Piekarewicz, [170] S.Tsuruta, “Thermal properties and detectability of “Neutrino-pasta scattering: The opacity of nonuniform neutron stars, II. Thermal evolution of rotation-powered neutron - rich matter,” Phys. Rev. C 69 , 045804 (2004). neutron stars,” Phys. Rep. 292, 1 (1998). [191] G. Watanabe, T. Maruyama, K. Sato, K. Yasuoka and [171] D. G. Yakovlev and C. J. Pethick, “Neutron star cool- T. Ebisuzaki, “Simulation of transitions between ‘Pasta’ ing,” Annu. Rev. Astron. Astrophys. 42 169-210 (2004). phases in dense matter,” Phys. Rev. Letters 94, 031101 [172] A. Y. Potekhin, J. A. Pons and D. Page, “Neutron (2005). stars - cooling and transport,” Space Sci. Rev. 191, 239 [192] T. Maruyama, T. Tatsumi, D. .N. Voskresensky, (2015). T. Tanigawa and S. Chiba, “Nuclear pasta structures [173] C. J. Pethick, “Cooling of neutron stars,” Rev. Mod. and the charge screening effect,” Phys. Rev. C 72, Phys. 64, 1133-40 (1992). 015802 (2005). 40

[193] T. Maruyama, G. Watanabe and S. Chiba, “Molecular structures from neutron star compactness,” arXiv dynamics for dense matter,” Prog. Theor. Exp. Phys. 1706.04736. 2012 01A201. [212] D. Kobyakov and C. J. Pethick, “Towards a metallurgy [194] K. Iida and K. Oyamatsu, “Symmetry energy, unstable of neutron star crusts,” Phys. Rev. Letters 112, 112504 nuclei, and neutron star crusts,” Eur. Phys. J. A 50 (2014). 42-57 (2014). [213] D. Kobyakov and C. J. Pethick, “Nucleus-nucleus inter- [195] N. Martin and M. Urban, “Liquid-gas coexistence ver- actions in the inner crust of neutron stars,” Phys. Rev. sus energy minimization with respect to the density pro- C 94, 055806:1-10 (2016). file in the inhomogeneous inner crust of neutron stars,” [214] E. Epelbaum, H.-W. Hammer, and U.-G. Meissner, Phys. Rev. C 92, 015803 (2015). “Modern theory of nuclear forces,” Rev. Mod. Phys. 81, [196] P. Haensel and B. Pichon, “Experimental nuclear 1773-1825 (2009). masses and the ground state of cold dense matter,” As- [215] I. Tews, S. Gandolfi, A. Gezerlis, and A. Schwenk, tron. Astrophys. 283, 313-8 (1994). “Quantum Monte Carlo calculations of neutron matter [197] B. S. Pudliner, A. Smerzi, J. Carlson, V. R. Pandhari- with chiral three-body forces,” Phys. Rev. C 93, 024305 pande, S. C. Pieper, and D. G. Ravenhall, “Neutron (2016). drops and Skyrme energy density functionals,” Phys. [216] C. Drischler, K. Hebeler, and A. Schwenk, “Asymmetric Rev. Letters 76, 2416-19 (1996). nuclear matter based on chiral two- and three-nucleon [198] H. Sakurai, Repts. Prog. Phys. to be published. interactions,” Phys. Rev. C 93, 054314 (2016). [199] F. Douchin , P. Haensel, and J. Meyer, “Nuclear surface [217] S. Gandolfi, A. Gezerlis, and J. Carlson, “Neutron mat- and curvature properties for SLy Skyrme forces and nu- ter from low to high density,” Ann. Rev. Nucl. Part. Sci. clei in the inner neutron-star crust,” Nucl. Phys. A 665, 65, 303-328 (2015). 419-46 (2000). [218] S. Gandolfi, A. Lovato, J. Carlson, and K. E. Schmidt, [200] J. M. Pearson, N. Chamel, S. Goriely, and C. Ducoin, “From the lightest nuclei to the equation of state of “Inner crust of neutron stars with mass-fitted Skyrme asymmetric nuclear matter with realistic nuclear inter- functionals,” Phys. Rev. C 85, 065803 (2012). actions,” Phys. Rev. C 90, 061306 (2014). [201] D. G. Ravenhall, C. D. Bennett, and C. J. Pethick, “Nu- [219] A. M. Dyugaev, “Character of phase transition in case clear surface energy and neutron-star matter,” Phys. of π condensation,” Pisma Zh. Eksp. Teor. Fiz. 22, 181- Rev. Letters 28, 978-81 (1972). 5 (1975). JETP Lett., 22, 83-5 (1975). [202] P. Avogadro, F. Barranco, R. A. Broglia, and E. Vigezzi, [220] A. B. Migdal, E. E. Saperstein, M. A. Troitsky and “Vortex-nucleus interaction in the inner crust of neutron D. N. Voskresensky, “Pion degrees of freedom in nuclear stars,” Nucl. Phys. A 811, 378-412 (2008). matter,” Phys. Rept. 192, 179-437 (1990). [203] P. Avogadro, F. Barranco, A. Idini, and E. Vigezzi, [221] K. Yagi, T. Hatsuda, and Y. Miake, Quark-Gluon “Medium polarization effects on the superfluidity of fi- Plasma: From Big Bang to Little Bang, Cambridge nite nuclei and of the Inner crust of neutron stars,” in Univ. Press, Cambridge 2005). Fifty Years of Nuclear BCS: Pairing in Finite Systems, [222] B. A. Freedman and L. D. McLerran, “Fermions and R. A. Broglia and V. Zelevinsky, eds. (World Scientific gauge vector mesons at finite temperature and density. Publ. Co., 2013). 3. The ground state energy of a relativistic quark gas,” [204] N. Chamel, J.M. Pearson, and S. Goriely, “Pairing: Phys. Rev. D 16, 1169-85 (1977). from atomic nuclei to neutron-star crusts,” in Fifty years [223] B. Freedman and L. D. McLerran, “Quark star phe- of nuclear BCS: Pairing in finite systems, R. A. Broglia nomenology,” Phys. Rev. D 17, 1109-22 (1978). and V. Zelevinsky, eds. (World Scientific Publ. Co., [224] E. S. Fraga, R. D. Pisarski and J. Schaffner-Bielich, 2013). “Small, dense quark stars from perturbative QCD,” [205] T. Ainsworth, D. Pines, and J. Wambach, “Effective Phys. Rev. D 63, 121702 (2001). interactions and superfluid energy gaps for low density [225] A. Kurkela, P. Romatschke and A. Vuorinen, “Cold neutron matter,” Phys. Letters B 222, 173-8 (1989). quark matter,” Phys. Rev. D 81,105021 (2010). [206] A. Gezerlis, C. J. Pethick, and A. Schwenk, “Pairing [226] V. A. Novikov, M. A. Shifman, A. I. Vainshtein and and superfluidity of nucleons in neutron stars,” in Novel V. I. Zakharov, “Are all hadrons alike?,” Nucl. Phys. B Superfluids, v. 2, K. H. Bennemann and J. B. Ketter- 191, 301-69 (1981). son, eds. (Oxford University Press, Oxford, 2014) p. 580; [227] L. McLerran and R. D. Pisarski, “Phases of cold, dense arXiv:1406.6109. quarks at large N(c),” Nucl. Phys. A 796, 83-100 (2007). [207] C. J. Pethick, D. G. Ravenhall, and C. Lorenz, “The [228] M. Alford, K. Rajagopal, and F. Wilczek “Color-flavor inner boundary of a neutron star crust,” Nucl. Phys. A locking and chiral symmetry breaking in high density 584, 675-703 (1995). QCD,” Nucl. Phys. B537, 443-458 (1999). [208] K. Hebeler, J. M. Lattimer, C. J. Pethick, and A. [229] R. D. Pisarski and D. H. Rischke, “Why color-flavor Schwenk, “Equation of state and neutron star proper- Locking is just like chiral symmetry breaking, Proc. Ju- ties constrained by nuclear physics and observation,” dah Eisenberg Memorial Symposium, ‘Nuclear matter, Astrophys. J. 773, 11-24 (2013); arXiv:1303.4662. hot and cold’, Tel Aviv, April 14 - 16, 1999; arXiv:nucl- [209] J. Hartle, “Bounds on the mass and moment of inertia of th/9907094. non-rotating neutron stars,” Phys. Reports 46, 201-47 [230] T. Kunihiro, “Quark number susceptibility and fluctua- (1978). tions in the vector channel at high temperatures,” Phys. [210] J. L. Zdunik, M. Fortin and P. Haensel, ‘Neutron star Letters B 271, 395-402 (1991). properties and the equation of state for the core,” As- [231] M. Alford, M. Braby, M. W. Paris and S. Reddy, “Hy- tron. and Astrophysm 599 A119:1-8 (2017). brid stars that masquerade as neutron stars,” Astro- [211] H. Sotani, K.Iida, and K. Oyamatsu, “Probing crustal phys. J. 629, 969-78 (2005). 41

[232] D. T. Son, “Superconductivity by long-range color mag- Katz, S. Krieg, C. Ratti and K. K. Szabo, “The QCD netic interaction in high-density quark matter,” Phys. equation of state with dynamical quarks,” JHEP 77: Rev. D 59, 094019 (1999). 1-32 (2010). [233] A. De Rujula, H. Georgi and S. L. Glashow, “Hadron [257] V. Vovchenko, D. V. Anchishkin, and M. I. Gorenstein, masses in a gauge theory,” Phys. Rev. D 12 147-62 “Hadron resonance gas equation of state from lattice (1975). QCD,” Phys.. Rev. C 91, 024905 (2015). [234] M. Anselmino, E. Predazzi, S. Ekelin, S. Fredriksson [258] M. Asakawa and T. Hatsuda, “What thermodynamics and D. B. Lichtenberg, “Diquarks,” Rev. Mod. Phys. tells about QCD plasma near phase transition,” Phys. 65, 1199-1233 (1993). Rev. D 55, 4488-91 (1997). [235] A. Selem and F. Wilczek, “Hadron systematics and [259] G. Baym, “Ultrarelativistic heavy ion collisions: the emergent diquarks,” hep-ph/0602128. first billion seconds,” Proc. QM2015 Kobe, Nucl. Phys. [236] R. L. Jaffe, “Exotica,’; Phys. Rept. 409 1-45 (2005). A 956, 1-10, (2016). [237] M. Kobayashi and T. Maskawa, “Chiral symmetry and [260] S. Lottini and G. Torrieri, “Quarkyonic percolation and eta-X mixing,” Prog. Theor. Phys. 44, 1422-4 (1970). deconfinement at finite density and number of colors,” [238] G. ’t Hooft, “How instantons solve the U(1) problem,” Phys. Rev, C 88 024912 (2013). Phys. Rept. 142, 357-387 (1986). [261] C. M. Soukoulis, Q. Li, and G. S. Grest, “Quantum [239] A. Gerhold and A. Rebhan, “Gauge dependence identi- percolation in three-dimensional systems,” Phys. Rev. ties for color superconducting QCD,” Phys. Rev. D 68, B 45, 7724-7729 (1992); A. Kaneko, D. Uema, and 011502 (2003). T. Ohtsuki, “Quantum percolation and the Anderson [240] D. D. Dietrich and D. H. Rischke, “Gluons, tadpoles, transition,” J. Phys. Soc. Jpn. 72, 141-142 (2003); K. and color neutrality in a two flavor color superconduc- Fukushima, private communication. tor,” Prog. Part. Nucl. Phys. 53, 305-316 (2004). [262] K. Iida and G. Baym, “Superfluid phases of quark mat- [241] M. Buballa and I. A. Shovkovy, “A Note on color neu- ter, III. Supercurrents and vortices,” Phys. Rev. D66, trality in NJL-type models,” Phys. Rev. D 72, 097501 014015, 1-15 (2002). (2005). [263] A. P. Balachandran, S. Digal and T. Matsuura, “Semi- [242] K. Iida and G. Baym, “The superfluid phases of quark superfluid strings in high density QCD,” Phys. Rev. D matter: Ginzburg-Landau theory and color neutrality,” 73, 074009:1-0 (2006). Phys. Rev. D 63, 074018 (2001). [264] E. Nakano, M. Nitta, and T. Matsuura, “Interactions [243] A. W. Steiner, S. Reddy, and M. Prakash, “Color neu- of non-Abelian global strings, Phys. Lett. B 672, 61- tral superconducting quark matter,” Phys. Rev. D 66 64 (2009); “Non-Abelian strings in high density QCD: 094007 (2002). zero modes and interactions, Phys. Rev. D 78, 045002 [244] H. Abuki, M. Kitazawa, and T. Kunihiro, “How do chi- (2008). ral condensates affect color superconducting quark mat- [265] M. G. Alford, S. K. Mallavarapu, T, Vachaspati, and ter under charge neutrality constraints?,” Phys. Letters A, Windisch, “Stability of superfluid vortices in dense B 615, 102-110 (2005). quark matter,” Phys. Rev. C 93, 045801 (2016). [245] J. R. Oppenheimer and G. M. Volkoff, “On massive neu- [266] M. Cipriani, W. Vinci, and M. Nitta, “Colorful boojums tron cores, Phys. Rev. 55, 374-381 (1939). at the interface of a color superconductor, Phys. Rev. D [246] P. M. Lo, B. Friman, and K. Redlich, “Polyakov loop 86, 121704(R) (2012). fluctuations and deconfinement in the limit of heavy [267] G. Baym, K. Fukushima, T. Hatsuda, M. Alford, and quarks,” Phys. Rev. D90, 074035 (2014). M. Tachibana, “Vortex continuity from the hadronic to [247] K. Fukushima, “Chiral effective model with the the color-flavor locked phase,” in preparation. Polyakov loop,” Phys. Letters B 591, 277-284 (2004). [268] M. A. Stephanov, “QCD phase diagram and the critical [248] S. Rössner, C. Ratti, and W. Weise, “Polyakov loop, di- point,” Prog. Theor. Phys. Suppl. 153, 139-156 (2004) quarks and the two-flavour phase diagram,” Phys. Rev. [Int. J. Mod. Phys. A 20, 4387-92 (2005)]. D 75, 034007 (2007). [269] M. Kitazawa, T. Koide, T. Kunihiro and Y. Nemoto, [249] N. M. Bratovic, T. Hatsuda and W. Weise, “Role of Vec- “Chiral and color superconducting phase transitions tor Interaction and Axial Anomaly in the PNJL Model- with vector interaction in a simple model,” Prog. Theor. ing of the QCD Phase Diagram,” Phys. Letters B 719, Phys. 108, 929-951 (2002). 131-135 (2013). [270] Z. Zhang, K. Fukushima and T. Kunihiro, “Number of [250] J. Steinheimer and S. Schramm, “The problem of re- the QCD critical points with neutral color superconduc- pulsive quark interactions - Lattice versus mean field tivity,” Phys. Rev. D 79, 014004 (2009). models,” Phys. Letters B 696, 257-261 (2011). [271] M. M. Aggarwal et al. [STAR Collaboration], “An [251] T. Kojo and G. Baym, “Color screening in cold quark experimental exploration of the QCD phase diagram: matter, Phys. Rev. D 89, 125008 (2014). the search for the critical point and the onset of de- [252] C. Alcock, E. Farhi, and A. Olinto, “Strange stars,” confinement,” arXiv:1007.2613 [nucl-ex]. Astrophys. J. 310, 261-272 (1986). [272] N. Abgrall et al. [NA61 Collaboration], “NA61/SHINE [253] M. A. Alpar, “Comment on strange stars,” Phys. Rev. facility at the CERN SPS: beams and detector system,” Letters 58, 2152 (1987). JINST 9 P06005 (2014). [254] M. G. Alford, S. Han, and M. Prakash, “Generic condi- [273] NICA white paper, tions for stable hybrid stars, Phys. Rev. D 88, 083013 http://theor0.jinr.ru/twiki-cgi/view/NICA/WebHome. (2013). [274] J-PARC white paper, [255] M. Cheng, et al., “Equation of state for physical quark http://asrc.jaea.go.jp/soshiki/gr/hadron/jparc-hi/. masses,” Phys. Rev. D 81 054504 (2010). [275] X. Luo [STAR Collaboration], “Energy dependence of [256] S. Borsányi, G. Endrodi, Z. Fodor, A. Jakovac, S. D. moments of net-proton and net-charge multiplicity dis- 42

tributions at STAR,” PoS CPOD 2014 019 (2015). 1286-90 (1968). [276] R. A. Lacey, “Indications for a critical end point in the [286] G. Caporaso and K. Brecher, “Must ultrabaric matter phase diagram for hot and dense nuclear matter,” Phys. be superluminal,” Phys. Rev. D 20, 1823-31 (1979). Rev. Letters 114, 142301 (2015). [287] G. F. R. Ellis, R. Maartens, and M. A. H. MacCal- [277] A. Bzdak and V. Koch, “Local efficiency corrections lum, “Causality and the speed of sound,” Gen. Relativ. to higher order cumulants,” Phys. Rev. C 91, 027901 Gravit. 39, 1651-60 (2007). (2015). [288] Ya. B. Zeld’ovich, “The equation of state at ultra- [278] A. Bzdak, R. Holzmann and V. Koch, “Multiplicity high densities and its relativistic limitations,” ZhETF dependent and non-binomial efficiency corrections for (USSR) 41 1609-15 (1961) [JETP (Sov. Phys.) 14, 1143- particle number cumulants,” Phys. Rev. C 94, 064907 7 (1962)]. (2016). [289] C. E. Rhoades and R. Ruffini, “Maximum mass of a [279] M. Kitazawa, “Efficient formulas for efficiency correc- neutron star,” Phys. Rev. Letters 32, 324-7 (1974). tion of cumulants,” Phys. Rev. C 93, 044911 (2016). [290] A. G. Sabbadini and J. B. Hartle, “Upper bound on the [280] H. T. Ding, F. Karsch and S. Mukherjee, “Thermody- mass of non-rotating neutron stars,” Astrophys. Space namics of strong-interaction matter from Lattice QCD,” Sci. 25, 117-31 (1973). Int. J. Mod. Phys. E 24, 1530007 (2015). [291] V. Kalogera and G. Baym, “The maximum mass of [281] K. Fukushima and C. Sasaki, “The phase diagram of a neutron star,” Astrophys. J. Letters 469, L61-L64 nuclear and quark matter at high baryon density,” Prog. (1996). Part. Nucl. Phys. 72, 99-154 (2013). [292] L. Lindblom, “Determining the nuclear equation of state [282] M. Buballa and S. Carignano, “Inhomogeneous chiral from neutron-star masses and radii,” Astrophys. J. 398, condensates,” Prog. Part. Nucl. Phys. 81, 39-96 (2015). 569-573 (1992). [283] D. H. Rischke, D. T. Son, and M. A. Stephanov, [293] C. A. Raithel, F. Ozel,¨ and D. Psaltis, “From neutron “Asymptotic deconfinement in high-density QCD ” star observables to the equation of state: II, Bayesian in- Phys. Rev. Letters 87, 062001 (2001). ference of equation of state pressures, arXiv:1704.00737 [284] S. A. Bludman and M. A. Ruderman, “Possibility of [astro-ph] (2017). the speed of sound exceeding the speed of light in ul- [294] A. Bauswein, H.-T. Janka, and R. Oechslin, “Testing tradense matter,” Phys. Rev. 170, 1176-1184 (1968); approximations of thermal effects in neutron star merger “Noncausality and instability in ultradense matter,” simulations,” Phys. Rev. D 82, 084043:1-10 (2010). Phys. Rev. D 1, 3243-3246 (1970). [295] K. Masuda, T. Hatsuda, and T. Takatsuka, “Hadron- [285] M. A. Ruderman, “Causes of sound faster than light in quark crossover and hot neutron stars at birth,” Prog. classical models of ultradense matter,” Phys. Rev. 172, Theor. Exp. 2016 Phys. 021D01.