Unruh-DeWitt detector and the interpretation of the horizon temperature in spherically symmetric dynamical space-times Giovanni Acquaviva(a), Roberto Di Criscienzo(a), Luciano Vanzo(a) and Sergio Zerbini(a) ∗ † ‡ § 1 1 0 2 (a) Dipartimento di Fisica, Universit`a di Trento n and Istituto Nazionale di Fisica Nucleare - Gruppo Collegato di Trento a J Via Sommarive 14, 38123 Povo, Italia 7 2 ] Abstract h t - Inthe paper,the temperature associatedwitha dynamicalsphericallysymmetricblackhole p or with a cosmological horizon is investigated from the point of view of a point-like detector. e h First, we briefly review the Hamilton-Jacobi tunneling method for a generic dynamical spher- [ ically symmetric space-time, and present two applications of the tunneling method. Then, we apply a well-known relativistic quantum theoretical technique, namely the Unruh-DeWitt de- 1 v tector formalism for a conformally coupled scalar field in a generic FRW space-time. As an 4 application, for the generic static black hole case and the FRW de Sitter case, making use of 5 peculiar Kodama observer trajectories, the tunneling semiclassical results are fully recovered, 2 automaticallycorrectedbyTolmanfactors. Someremarksonthe temperatureofFRWuniverse 5 are presented. For more generalspaces interpolating de Sitter space with the Einstein-de Sitter . 1 universe a second set of poles is present, whose exact role remains to be clarified, plus an extra 0 fluctuatingtermdescribingthewayequilibriumisreached,similarlytodeSitterspace. Thesim- 1 1 plethermalinterpretationfoundfordeSitterspaceislostandforces,atasametime,adifferent : quantum interpretation of the horizon surface gravity for the cosmologicalFRW models. v i X PACS numbers: 04.70.-s, 04.70.Dy r a 1 Introduction It is well known that Hawking radiation [1] is considered one of the most important prediction of Quantum Field Theory in curved space-time. Several derivations of this effect have been proposed [2, 3, 4, 5, 6] and recently the search for “experimental” verification making use of analogue models has been pursued by many investigators (see for example [7, 8]). ∗[email protected] †[email protected] ‡[email protected] §[email protected] 1 In 2000 Parikh and Wilczek [9](see also [10]) introduced the so-called tunneling approach for investigating Hawking radiation. Here, we shall firstly review a variant of their method, called Hamilton-Jacobi tunneling method [11, 12, 13] and also [14]. This method is covariant and can be extended to the dynamical case [15, 16, 17, 18], and to the study of decay of massive particles and particle creation by naked singularities [19]. In their approach, Parikh & Wilczek made a clever use of the Painlev´e stationary gauge for 4-dimensional Schwarzschild black hole 2mG 2Gm ds2 = 1 dT2+2 drdT +dr2+r2dΩ2, (1.1) − − r r (cid:18) (cid:19) r which is regular on the trapping horizon r = 2mG 1. This is one of the key points since the use H of singular gauges, as the Schwarzschild gauge, leads, in general, to ambiguities and it is useless in the dynamical case which we are interested in. The second merit we can address to the Parikh & Wilczek work was a treatment of the back- reaction on the metric, based on energy conservation. In the following, we shall limit to leading term results and neglect the back-reaction. However, it may be worth to recall that in the limit where the number of emitted quanta is reasonably large back-reaction effects can be accounted for by assuming that the mass parameter m is a continuous function of time T. Penrose’s diagrams for this more general case have been determined too, e.g. in [20, 21]. The Hamilton-Jacobi method is reasonably simple, even though subtleties are present (see, for example [22]). It is based on the computation of the classical action I along a trajectory starting slightlybehindthetrappinghorizonbutendinginthebulk,andtheassociatedWKBapproximation (c = 1) Amplitude eiI~ . (1.2) ∝ For an evaporating black hole such a trajectory would be classically forbiddensince at the trapping horizon photons are only momentarily at rest, whence dr/dT < 0 inside the horizon2. The related semiclassical emission rate reads Γ Amplitude2 e−2ℑ~I . (1.3) ∝ | | ∝ with standing for the imaginary part. In the tunneling across the horizon, the imaginary part of ℑ theclassical actionI stemsfromtheinterpretation ofaformalhorizondivergence, andinevaluating it, one has to make use of Feynman’s prescription related to a simple pole in integration such as a f(x) a f(x) dx dx = iπf(0)+real stuff. (1.4) x → x i0 Z0 Z0 − − − This corresponds to the choice of suitable boundary conditions in quantum field theory approach. We may anticipate that in the WKB approximation of the tunneling probability, one asymp- totically gets a Boltzmann factor, in which an energy ω appears, i.e. Γ e−β~ω. (1.5) ∝ 1The global event horizon is r0 < rH for an evaporating black hole and r0 > rH for an accreting black hole, as can beseen from theequation of radial nullrays, r˙0 =1−prH/r0. 2It means thephoton must go back in timeT toescape thehorizon. 2 It is crucial in our approach that the argument of the exponent be a coordinate scalar (invari- ant quantity), since otherwise no physical meaning can be addressed to Γ. In particular, in the Boltzmann factor, β and the energy ω have to be separately coordinate scalars, otherwise again no invariant meaning could be given to the quantity β. In the static case, we interpret T = ~ as the ~ β horizon temperature. In the cosmological case on the other hand, one can still have a trapping horizon despite the ab- sence of collapsed matter simply as a result of the expansion of the universe. As for an excreting black hole, this too is represented by a time-like hypersurface. Similarly, an approximate notion of temperature can be associated to such horizons based on the existence of a surface gravity and again the tunneling method gives a non vanishing amplitude having the Boltzmann form. However we will see that a comoving monopole detector seems to react to the expansion in a different, non “Boltzmannian” way, while reaching thermal equilibrium (or better, detailed balance conditions) only asymptotically for large times. The paper is organized in the following way: in Section 2, the Kodama-Hayward formalism and theHamilton-Jacobi method aresummarized; then, in Section 3, we consider the tunnelingmethod for a generic static black hole in Kruskal-like gauge. In Section 4, the same formalism is applied to Friedmann-Robertson-Walker (FRW) spatially flat space-times. A formula for the de Sitter temperature which is valid in any coordinate patch is provided. Section 5, contains a discussion on theUnruh-DeWittdetector whichisintroducedinordertoconfirmthroughquantumfieldtheoretic techniques the results of previous sections. In particular, a quite general formula for the response function per unit proper time is obtained and two relevant applications are presented. Section 6 contains a further example, consisting of a (Ω ,Ω ) cosmological model, where only approximate m Λ results can be obtained. Some toy models are presented and discussed. Concluding remarks follow at the end of the paper. We use the metric signature ( ,+,+,+); Greek indices run over 0 to 3 while mid-Latin as i,j − only over 0 and 1. We use Planck units in which c= ~ = G= k = 1. B 2 The Kodama-Hayward formalism and Hamilton-Jacobi tunnel- ing method In order to treat the spherically symmetric dynamical case, the use of invariant quantities plays a crucial role [17, 18, 23, 24]. Here we review the general formalism. To begin with, let us recall that any spherically symmetric metric can locally be expressed in the form ds2 = γ (xi)dxidxj +R2(xi)dΩ2, i,j 0,1 , (2.1) ij ∈ { } where the two-dimensional metric dγ2 = γ (xi)dxidxj (2.2) ij is referred to as the normal metric, xi are associated coordinates and R(xi) is the areal radius, { } considered as a scalar field in the two-dimensional normal space. A relevant scalar quantity in the reduced normal space is χ(x) = γij(x)∂ R(x)∂ R(x), (2.3) i j 3 since the dynamical trapping horizon, if it exists, is located in correspondence of χ(x) = 0, (2.4) H (cid:12) provided t(cid:12)hat ∂ χ = 0. The Misner-Sharp gravitational energy is defined by (cid:12) i |H 6 1 E (x) = R(x)[1 χ(x)] . (2.5) MS 2 − Thisisaninvariantquantity onthenormalspace. Note alsothat, onthehorizon,E = 1R MS|H 2 H ≡ M. Furthermore, one can introduce the Hayward surface gravity associated with this dynamical horizon, given by the normal-space scalar 1 κ = 2 R . (2.6) H γ 2 H (cid:12) Recall that, in th(cid:12)e spherical symmetric dynamical case, it is possible to introduce the Kodama (cid:12) vector field , with ( αG );β = 0 that can be taken as its defining property. Given the metric αβ K K (2.1), the Kodama vector components are 1 i(x) = εij∂ R, θ = 0 = ϕ . (2.7) j K √ γ K K − We may introduce the Kodama trajectories, and related Kodama observer, by means of integral lines of Kodama vector dxi 1 = i = εij∂ R. (2.8) j dλ K √ γ − As a result, dR(x(λ)) dxi 1 = ∂ R = εij∂ R∂ R = 0, (2.9) i j i dλ dλ √ γ − so, we have proved the following Lemma: In generic spherically symmetric space-times, the areal radius R is conserved along Kodama trajectories. In a generic spherically symmetric space-time a geometric dynamical identity holds true in general. This can be derived as follows. Let us introduce the normal space invariant (2) = γijT . (2.10) ij T which we shall term reduced stress energy tensor trace. Then, making use of Einstein equations, it is possible to show that, on the dynamical horizon, (see, for example, [18]) 1 (2) κ = +2πR . (2.11) H 2R HTH H Introducing the horizon area and the (formal) three-volume enclosed by the horizon, with their respective differentials = 4πR2 , d = 8πR dR , (2.12) AH H AH H H 4 4 V = πR3 , dV = 4πR2 dR , (2.13) H 3 H H H H one gets (2) κ R Hd = d H + TH dV . (2.14) H H 8π A 2 2 (cid:18) (cid:19) This equation can be re-cast in the form of a geometrical identity, once we introduce the Misner- Sharp energy at the horizon [24]: (2) κ dM = Hd AH TH dV . (2.15) H 2π 4 − 2 (cid:18) (cid:19) We conclude summarizing the main ingredients of the Hamilton-Jacobi tunneling method. The Kodama vector, introduced above, gives a preferred flow of time and in this sense it generalizes the flow of time given by the Killing vector in the static case. As a consequence, we may introduce the invariant energy associated with a particle of mass m by means of the scalar quantity on the normal space, the Kodama, or generalized Killing energy, ω = i∂ I, (2.16) i −K where I is the particle action which we assume to satisfy the reduced Hamilton–Jacobi equation γij∂ I∂ I +m2 = 0. (2.17) i j As we allow for non-minimal gravitational coupling, the substitution m2 m2 +ξ is in order → R whenever ξ = 0, being the Ricci curvature scalar and ξ a dimensionless coupling constant. 6 R If we are interested in Hawking effect, we may neglect the mass. Onecan reconstruct the action for particles coming out of the horizon by I = dt∂ I + dr∂ I (2.18) t r Z Z upon solving the Hamilton-Jacobi equation (2.17) with zero mass. Then, one applies the near horizon approximation and the null horizon expansion which forces us to make use of regular gauges across the horizon. These assumptions allow one to know the classical action and compute its imaginary part making useof Feynman’s prescription. The semiclassical resultfor the tunneling probability is then [15, 17, 18] Γ e 2 I e βHωH . (2.19) − ℑ − ≃ ≃ In the static case, there is no doubt that T = 1 can be interpreted as the Hawking temperature; H βH but what about the dynamical spherically symmetric case? From Einstein’s equations and in presence of a dynamical horizon the geometrical dynamical Law holds as we have already seen. If also the Area Law for the black hole entropy is true, namely that H S = A , (2.20) H 4~ then, we can rewrite the geometrical identity (2.15) as a First Law of Thermodynamics for black holes, (2) T dM = T dS H dV , (2.21) H H H − 2 5 with κ T = ~ H . (2.22) H 2π Here we have re-introduced ~ in order to stress the quantum nature of entropy and temperature. Thus, it seems suggestive to interpret T as a dynamical temperature of a slowly changing dy- H namical black hole, interpretation which is also supported by the fact that the Hayward dynamical surface gravity is an invariant for a generic spherically symmetric space-time. This interpretation has been put forward in [17, 18, 15, 25]. Later in the paper, we shall try to present quantum theoretical arguments in order to substantiate this interpretation. 3 Generic static black hole space-time As a first application of the formalism, let us consider a generic static black hole space-time. The starting point is a black hole metric in the Schwarzschild static gauge, dr2 ds2 = V(r)dt2+ +r2dΩ2 , (3.1) − W(r) where, for sake of simplicity, we shall assume W(r) = V(r), with V(r) having just simple poles in ordertodescribewhatwemightcallaniceblackhole. Letr bethe(greatest)solutionofV(r)= 0, H the general formalism tells us that the horizon is located at r = r ; the Kodama vector coincides H with the usual Killing vector (1,0,0,0); and the Hayward surface gravity is the Killing surface gravity, namely κ = κ = VH′ . This gauge is singular on the horizon, and it is not appropriate for H 2 the Hamilton-Jacobi tunneling method. For this reason we now introduce the Kruskal-like gauge associated with this static black hole solution. The first step consists in introducing the tortoise coordinate dr dr = . (3.2) ∗ V(r) Then one has < r < and ∗ −∞ ∞ dr2 ds2 = V(r)dt2+ +r2dΩ2 = V(r )( dt2+(dr )2)+r2(r )dΩ2 . (3.3) ∗ ∗ ∗ − W(r) − Introduce Kruskal-like coordinates, according to 1 1 R = eκr∗cosh(κt), T = eκr∗sinh(κt), (3.4) κ κ one has 1 T2+R2 = e2κr∗, (3.5) − κ2 and ds2 = V(r∗)e−2κr∗( dT2+dR2)+r2(T,R)dΩ2 − eΨ(r )( dT2+dR2)+r2(T,R)dΩ2 , (3.6) ∗ ≡ − 6 where now the coordinates are T and R, r∗ = r∗(T,R), eΨ(r∗) = V(r∗)e−2κr∗, and the normal metric turns out to be conformally flat. In this gauge, the metric is a spherically symmetric but time dependent one. The general formalism tells us that the horizon corresponds to (∂ r) = (∂ r) , (3.7) T H R H ± and this is equivalent to T = R and r . The Killing-Kodama vector is ∗ ± → −∞ = e−Ψ(r∗)(∂Rr, ∂T r) . (3.8) K − Making use of the general formula, a direct, but tedious calculation confirms that the Hayward’s surface gravity is still the Killing one. In fact, we have e ΨH V κ = − ∂2 r+∂2 r = H′ . (3.9) H 2 − T R H 2 (cid:0) (cid:1) Let us apply the Hamilton-Jacobi tunneling method, working in this time-dependent gauge. The Kodama energy is ω = e−Ψ(r∗)(∂Rr∂T I ∂T r∂RI) (3.10) − whereI istheclassicalaction. Inthisdiagonalconformallyflatgauge,theHamilton-Jacobiequation is simply ∂ I = ∂ I and the null horizon expansion condition gives dT = dR. Making the + T R ± ± choice (outgoing particle), in the horizon approximation, one has I 2 dR∂ I. (3.11) R ≃ Z Eq. (3.10) gives eΨω ∂ I = . (3.12) R ∂ r ∂ r R T − The near horizon expansion gives ∂ r ∂ r (∂2r ∂2r) (R R ). (3.13) R − T ≃ R − T H − H so that the action (3.11) becomes, ω 1 I = dR , (3.14) κ (R R i0) H H Z − − Making use of Feynman’s iǫ– prescription for the simple pole, we finally get πω 2πω H H I = = . (3.15) ℑ κ V H H′ This imaginary part can be interpreted as arising because of a non-vanishing tunneling probability rate of (massless) particles across the event horizon, Γ exp( 2 I) e−V′4(rπH)ωH . (3.16) ∼ − ℑ ∼ The well-known result T = VH′ is recovered. H 4π 7 4 Cosmological horizons As a second application of the formalism, we consider a generic FRW space-time with constant curvature spatial sections. Its line element can be written as dr2 ds2 = dt2+a2(t) +[a(t)r]2dΩ2 . (4.1) − 1 kˆr2 − Here kˆ := k, where l is such that a(t)l is the curvature radius of the constant curvature spatial l2 sections at time t and, as usual, k = 0, 1,+1 labels flat, open and closed three–geometries, − respectively. In this gauge, the normal reduced metric is diagonal and kˆ χ(t,r) = 1 [a(t)r]2 H2(t)+ . (4.2) − a2(t) " # The dynamical trapping horizon is implicitly given by χ = 0, namely H 1 a˙(t) R := a(t)r = , with H(t) = , (4.3) H H H2(t)+ kˆ a(t) a2(t) q provided the space-time energy density ρ(t) is positive. It coincides with the Hubble radius as defined by astronomers for vanishing curvature, but we shall call it Hubble radius in any case. The dynamical surface gravity is given by equation (2.6) and reads 1 kˆ κ = H2(t)+ H˙(t)+ R (t), (4.4) H − 2 2a2(t) H ! and the minus sign refers to the fact the Hubble horizon is, in Hayward’s terminology, of the inner type. According to (2.7), the Kodama vector is = 1 kˆr2(∂ rH(t)∂ ) (4.5) t r K − − q so that the invariant Kodama energy of a particle is equal to ω = 1 kˆr2( ∂ I +rH(t)∂ I) 1 kˆr2ω˜ (4.6) t r − − ≡ − q q Notice that is space-like for ra > (H2 +kˆ/a2) 1/2, i.e. beyond the horizon. It follows that we − K can only ask for particles to be emitted in the inner region r < r . H The next ingredient is the reduced Hamilton-Jacobi equation for a relativistic particle with mass parameter m, (1 kˆr2) (∂ I)2+ − (∂ I)2+m2 = 0. (4.7) − t a2(t) r Making use of (4.6), one can solve for ∂ I, namely r aHω˜(ar) a ω2 m2+m2 H2+ kˆ (ar)2 ± − a2 ∂ I = r , (4.8) r (cid:16) (cid:17) − 1 H2+ kˆ (ar)2 − a2 (cid:16) (cid:17) 8 with the signs chosen according to which direction we think the particle is propagating. The effective mass here defines two important and complementary energy scales: if one is interested in the horizon tunneling then only the pole matters (since the denominator vanishes), and we may neglect to all the extents the mass parameter setting m = 0 (since its coefficient vanishes on the horizon). On the opposite, in investigating other effects in the bulk away from the horizon, such as the decay rate of composite particles, the role of the effective mass becomes relevant as the energy of the particle can besmaller than the energy scale settled by m, and the squareroot can possiblyacquire a branch cut singularity. 4.1 The FRW space-time Asanapplicationofthelastformulawemayderive, following[18](seealso[25]), thecosmichorizon tunneling rate. To this aim, as we have anticipated, the energy scale is such that near the horizon, we may neglect the particle’s mass, and note that radially moving massless particles follow a null direction. Then along a null radial direction from the horizon to the inner region we have a(t) ∆t = ∆r. (4.9) − 1 kˆr2 − The outgoingpparticle action, that is the action for particles coming out of the horizon towards the inner region, is then I = dt∂ I + dr∂ I (4.10) t r Z Z = 2 dr∂ I (4.11) r Z upon solving the Hamilton-Jacobi equation (4.7) with zero mass and using (4.9). For ∂ I we use r now Eq. (4.8), which exhibits a pole at the vanishing of the function F(r,t) := 1 (a2H2 +kˆ)r2, − defining the horizon position. Expanding F(r,t) again along a null direction, one gets F(r,t) +4κ a(t)(r r )+... , (4.12) H H ≈ − where κ is the Hayward dynamical surface gravity. In order to deal with the simple pole in the H integrand, we implement Feynman’s iǫ – prescription. In the final result, beside a real (irrelevant) contribution, we obtain the following imaginary part [18] πω H I = . (4.13) ℑ − κ H Thisimaginary partis again interpreted as arising becauseof a non-vanishingtunnelingprobability rate of (massless) particles across the cosmological horizon, Γ ∼ exp(−2ℑI)∼ e−(−2κπH)ωH. (4.14) Notice that, since κ < 0 and ω > 0 for physical particles, (4.13) is positive definite. As showed H H in [18], this result is invariant since the quantities appearing in the imaginary part are manifestly invariant. Furthermore T = κ /2π satisfies a First Law. As a consequence, at least, in some H − 9 asymptotic regime and for slowly changes in the geometry, we may interpret T = κ /2π as the H − dynamical temperature associated with FRW space-times. Inparticular,thisgives naturallyapositivetemperaturefordeSitterspace-time, alongdebated questionyears ago, usuallyresolved bychangingthesign ofthehorizon’s energy. Itshouldbenoted that in literature, the dynamical temperature is usually given in the form T = H (exceptions are 2π the papers [26, 27]) with H2 = Λ/3 H2. Of course this is the expected result for de Sitter space ≡ 0 in inflationary coordinates, but it ceases to be correct in any other coordinate system since, for example, H = H tanhH t in global coordinate system with positive spatial curvature. In this 0 0 regard, the H˙ and kˆ terms are crucial in order to get an invariant temperature. Since this fact seems not so widely known, for sake of completeness, we shall try to show it in greater detail. de Sitter space in the global patch is described by the metric dr2 ds2 = dt2+cosh2(H t) +dΩ2 (4.15) − 0 1 kˆr2 (cid:20) − (cid:21) with a(t) = cosh(H t), and kˆ = H2. The Hubble parameter is time dependent 0 0 H(t) = H tanh(H t), (4.16) 0 0 and satisfies the identity kˆ H˙(t) = H2 H2(t) = . (4.17) 0 − a2(t) Making use of (4.3), the horizon radius is 1 1 R := a(t)r = = (4.18) H H H2(t)+ kˆ H0 a2(t) q as it should be, and the Hayward’s surface gravity 1 kˆ κ = H2(t)+ H˙(t)+ R (t) = H , (4.19) H − 2 2a2(t) H 0 ! as it should, since it is an invariant quantity. Hence, we see that the H˙ and kˆ terms have to be present in a generic FRW space-time. The important spatially flat case straightforwardly follows, H˙(t) κ = H(t)+ . (4.20) H − 2H(t) ! Note thatthis isindependentonposition, suggestingthat κ really is an intrinsicpropertyofFRW H spacelinked tothebulk. Thehorizon’s temperatureandtheensuingheatingof matter was foreseen several years ago in the interesting paper [28]. We are going to challenge this interpretation. 10