Near-extremal black holes Bhramar Chatterjee∗, Amit Ghosh† Saha Institute of Nuclear Physics, 1/AF Bidhannagar, Kolkata-700 064, India January 20, 2012 2 1 0 2 Abstract n We present a new formulation of deriving Hawking temperature for a J near-extremal black holes using distributions. In this paper the near- extremal Reissner-Nordstr¨om and Kerr black holes are discussed. It is 9 shown that the extremal solution as a limit of non-extremal metric is 1 well-defined. The pureextremal case is also discussed separately. ] c q 1 Introduction - r g Classicallyablackholehorizonisaone-waymembranethatabsorbseverything. [ Since the discovery of Hawking effect [1], it became clear that quantum effects 1 causetheblackholetoemitthermalradiation. TheoriginalderivationofHawk- v ing radiation did not involve a full quantum theory of gravity. Only the free 7 fields propagating in a curved spacetime were quantized while the background 1 spacetimeremainedclassical. SeveralothermethodsofderivingHawkingradia- 0 4 tionexistintheliterature. RecentlyasemiclassicalmethodofderivingHawking . radiationthroughtunnelling [2]-[16],[32]hasbeen verypopular. The tunnelling 1 methodinvolvescalculatingthe imaginarypartofthe actionforthe (classically 0 2 forbidden) process of s-wave emission across the horizon, which in turn is re- 1 lated to the Boltzmann factor for emission at the Hawking temperature. Using v: the WKBapproximationthetunnelling probabilityforthe classicallyforbidden i trajectory of the s-wave coming from inside to outside the horizon is given by: X r Γ=e−2ImS. (1) a where S is the classical action of the trajectory to leading order. This is equal to the Boltzmann factor e−βE where β is the inverse Hawking temperature. Equating the Boltzmann distribution with (1) implies, E T = . (2) H 2ImS Usually the results are obtained considering a scalar field in the black hole background, but one can also consider fermionic field [17, 18, 19]. Though the calculations are a bit complicated it is essentially the same. ∗[email protected] †[email protected] 1 Like all other formulations the tunnelling picture has its merits and demer- its. Onthepositivesides,this methodhasbeensuccessfullyappliedtodifferent typesofhorizonsincludinganti-deSitter(AdS)[24],deSitter(dS)[20,21,22,23], BTZ[25, 26],higher dimensional black holes and some exotic spacetimes[27, 29, 30, 31] apartfromthe more conventionaltypes of black holes, in eachcase pro- ducing the corresponding Hawking temperature correctly. Also, as the deriva- tion involves only the local geometry, the tunnelling method can be applied to any local horizon, in particular to cosmological and weakly isolated horizons [28,29]. Ithasalsobeenappliedtopasthorizonsandwhiteholes,inwhichcase a clear notion of temperature emerges in complete analogy to black holes [13]. Ontheotherhand,thetunnellingformulationgivesalessdetailedpictureof the radiation process since it is mainly related to a semiclassical emission rate. In this paper we propose a new formulation of the Hawking effect which does not require the WKB approximation to compute the emission rate. Instead we will directly construct the modes from the field equation and calculate the emissionrate. WeshalldiscusstheReissner-Nordstro¨m(RN)andtheKerrblack holes. The idea is to construct single particle states only outside the horizon and somehow continue these states to ‘inside’ keeping in mind that the horizon or the surface from where particle escapes to infinity is not a null horizon, but something like Hayward’s timelike trapping horizon [10, 36]. Since this is essentiallyahorizoncrossingphenomenon,agoodsetofcoordinatesisrequired which is regular across the horizon, and we will use Kruskal coordinates. Near the horizon the metric is flat and we shall construct the field modes using this metric only. In that sense this is alsoa localcalculationofHawking effect since we need not be bothered about the global structure of the spacetime. The outgoing modes have a logarithmic singularity at the horizon but we have to keep in mind that the field modes are essentially distributions and it is shown that the distributions are quite well behavedat the horizon. This approachhas been previously suggested by Damour and Ruffini [33]. After constructing the modes, we have calculated the probability current coming out of the horizon using the standard field theoretic formula. The conditional probability that a particle emits when it is incident on the surface from the other side is then equated to the Boltzmann factor P P = (emission∩incident) =e−βE (3) (emission|incident) P (incident) which gives the Hawking temperature associated with the horizon. ThemainreasonforchoosingKerrand/orRNspacetimeisthatbothexhibit extremal limits. Both metrics have two horizons: the outer (r = r ) and the + inner (r = r ) having different properties. They do not even have the same − temperature, the temperature of the inner horizon is higher than the outer horizon. The extremal limit is achieved as r r . In this paper we have + − → calculated the probability flux coming out of both the horizon separately and also what happens in the extremal case. It is found that if one considers the extremal solution as a limiting case of the non-extremal one, then the results agree with the standard picture, i.e, the temperature of an extremal black hole is zero. On the other hand if one takes an extremal metric from the beginning, then the findings are to be interpreted carefully. This paper is organized as follows: in section 2 we shall discuss Reissner- Nordstro¨msolution,constructingtheKruskalcoordinatedforboththehorizons 2 and then computing the scalar modes and the probability current for both the cases. Kerrblackhole will be consideredinsection3,emphasizing againonthe different sets of Kruskal coordinates for the outer and the inner horizons, the field modes using distributions and finally the probability flux coming out of both the horizons. We will show how the extremal solution as a limiting case of the non-extremal one produces the standard results at the end of section 3. In section 4 we shall consider anextremal metric andusing the aforementioned formulation calculate the probability flux across the horizon. 2 The Reissner-Nordstro¨m black hole 2.1 The metric and Kruskal coordinates The Reissner-Nordstro¨mmetric is dr2 ds2 = f(r)dt2+ +r2dΩ2. (4) − f(r) 2M Q2 f(r) = 1 + . (5) − r r2 (cid:18) (cid:19) Here,M isthemassandQistheelectricchargeoftheblackhole. Thehorizons are situated at, f(r)=0, i.e, r =M M2 Q2. (6) ± ± − The outer horizon r is the event hoprizon and the inner horizon r is the + − apparent horizon for the Reissner-Nordstro¨m spacetime. The extremal limit is obtained from Q M, or r r . + − → → Thecoordinatest,raresingularattheouterhorizon(r =r )andonecanin- + troduce Kruskal-likecoordinatesto extend the metric acrossthis surface. How- ever, these coordinates fail to be regular at the inner horizon and so another coordinate system is needed to extend the metric beyond the inner horizon. So the coordinatesare specific to a given horizonand eventwo coordinate patches fail to cover the entire Reissner-Nordstro¨m manifold. Let us consider the two horizons separately. 2.1.1 The outer horizon As, r r , + → f(r) 2κ (r r ). (7) + + ≈ − where, 1 r r κ f′(r )= +− −. (8) + ≡ 2 + 2r2 + is the surface gravity at the outer horizon. It follows that near r =r , + dr 1 r = ln κ (r r ) . (9) ∗ ≡ f(r) ∼ 2κ | + − + | + Z Introducing the null coordinates u = t r and v = t+r , the surface r = r ∗ ∗ + − appears at v u= and we define the Kruskal-like coordinates U and V + + − −∞ by, U = e−κ+u, V =eκ+v. (10) + + ∓ 3 Here the upper sign refers to r > r and the lower sign refers to r < r . The + + future outer horizon is defined as U =0,V >0. The metric is regular at the + + outer horizon as seen from the near horizon form, 2 ds2 dU dV +r2dΩ2. (11) ≃−κ2 + + + + But r at the inner horizon which is located at v u= or U V = ∗ + + →∞ − ∞ ∞ and the Kruskal coordinates are singular there. This coordinate patch can be used for r < r < , where, r > r . Thus, we need another set of Kruskal 1 1 − ∞ coordinates to extend the spacetime beyond r =r . − 2.1.2 The inner horizon The new set of Kruskal coordinates for the inner horizon can be constructed in a similar manner. As r r , the function f(r) becomes, − → f(r) 2κ (r r ). (12) − − ≈− − where, 1 r r κ f′(r ) = +− −. (13) − ≡ 2| − | 2r2 − Near r =r , − dr 1 r = ln κ (r r ) . (14) ∗ ≡ f(r) ∼−2κ | − − − | Z − With u=t r and v =t+r , the surface r =r appears at v u=+ and ∗ ∗ − − − ∞ we define the new Kruskal coordinates by, U = eκ−u, V = e−κ−v. (15) − − ∓ − Here, the upper sign refers to r > r and the lower sign refers to r < r . The − − future inner horizon is defined as U =0,V <0. − − Then f 2U V and the metric becomes − − ≃− 2 ds2 dU dV +r2dΩ2. (16) ≃−κ2 − − − − which is regular at r =r . − 2.2 Scalar modes and the probability current A scalar field Φ satisfies the covariantKlein-Gordonequation and in this back- ground the modes can be separated as, 1 Φ (r ,t) ω ∗ Φ = Y (θ,φ). (17) ωlm lm √4πω r We shall consider only the positive frequency (ω >0) modes which satisfies i∂ Φ =ωΦ . (18) t ω ω In Kruskal coordinates (for the outer horizon), the Killing vector ∂ becomes, t ∂ = κ U ∂ +κ V ∂ . (19) t − + + U+ + + V+ 4 So, the U and V modes are decoupled, + + Φ =[f (U )+g (V )]. (20) ω ω + ω + And the solutions are, iω fω(U+) = Nω U+ κ+. (21) | | gω(V+) = Nω(V+)−κiω+. (22) whereN isthenormalizationconstant. TheV modesareingoingintheouter ω + horizon and is well behaved across the horizon. The U modes are outgoing and are not well behaved close to the horizon + because they oscillate infinitely rapidly. However, to calculate the emission probability we shall need these modes only. Theprobabilitycurrentis positivedefinite forpositivefrequencymodes and associated with the U modes it is + jout = i κ U ∂ Φ Φ +κ U Φ ∂ Φ . (23) − − + + U+ ω ω + + ω U+ ω The U modes are defi(cid:2)ned insid(cid:0)e and o(cid:1)utside the horizo(cid:0)n,but a(cid:1)s(cid:3)it approaches + the horizon at U = 0 the modes pick up a logarithmic singularity and is not + differentiable. So jout cannot be calculated naively. But actually, these modes aredistributionvaluedasmentionedearlierandnottobeinterpretedasordinary functions. As distributions, they arewell defined andinfinitely differentiable at the horizon. The distributions are of the form [34], iω fω =ǫl→im0Nω|U++iǫ|κiω+ =( NNωω(UU++)κiωκ++e−κπ+ω ffoorr UU++ ><00,. (24) | | and the complex conjugate distribution is, fω =ǫl→im0Nω∗|U+−iǫ|−κiω+ =( NNωω∗∗|(UU++|)−−κiκiω+ω+e−κπ+ω ffoorr UU++ <>00., (25) ThesedistributionsareuniquelyassociatedwiththeU -modesifweimposethe + additional condition that these are well behaved for large frequencies, ω . →∞ The derivatives of the distributions are also uniquely determined, ∂U+fω = Nω(cid:18)κiω+(cid:19)ǫli→m0|U++iǫ|κiω+−1. (26) ∂U+fω = Nω∗(cid:18)−κiω+(cid:19)ǫl→im0|U+−iǫ|−κiω+−1. (27) So, the probability current associated with the outgoing modes is, jout =ωU+ Nω 2lim 1 + 1 (U+ iǫ)−κiω+ (U++iǫ)κiω+ . (28) | | ǫ→0(cid:20)U+−iǫ U++iǫ(cid:21) − Now, U (U iǫ)−1 gives the identity distribution, because (U iǫ)−1 = + + + ± ± PV(1/U ) iπδ(U ) and U δ(U )=0. Finally, + + + + ∓ lim(U+ iǫ)∓κiω+ = lime∓κiω+ ln(U+∓iǫ) =e∓κiω+(ln|U+|∓iπθ(−U+)). (29) ǫ→0 ∓ ǫ→0 5 As a result, we get the outgoing probability current, N 2 for U >0, jout =( |Nωω|2e−2κπ+ω for U++ <0. (30) | | In a similar manner, using the Kruskal coordinates on the inner horizon one can calculate the outgoing probability current inside and outside of the inner horizon. In this case, N˜ 2 for U >0, jout =( |N˜ωω|2e−2κπ−ω for U−− <0. (31) | | 3 Kerr black hole 3.1 The metric and Kruskal coordinates The Kerr metric is given by, 2Mr 4Marsin2θ Σ ρ2 ds2 = 1 dt2 dtdφ+ sin2θdφ2+ dr2+ρ2dθ2. (32) − − ρ2 − ρ2 ρ2 ∆ (cid:18) (cid:19) where, ρ2 = r2+a2cos2θ. (33) ∆ = r2+a2 2Mr. (34) − Σ = r2+a2 2 a2∆sin2θ. (35) − (cid:0) (cid:1) The horizons are situated at ∆=0, i.e at, r =M M2 a2. (36) ± ± − p ConstructionofKruskalcoordinatesfortheKerrspacetimeisverysimilartothe Reissner-Nordstro¨mspacetimethoughabitmorecomplicated. Letusdefinethe u,v coordinates for the Kerr spacetime by using the following transformations, r2+a2 r = dr. (37) ∗ ∆ Z at φ˜ = φ . (38) − 2Mr + And as usual, u = t r . (39) ∗ − v = t+r . (40) ∗ Let us now define the Kruskal like coordinates for the Kerr spacetime, U = e−κ+u, V =eκ+v. (41) + + ∓ 6 where U V = e2κ+r(r r )(r r )−κκ−+ . (42) + + + − ∓ − − 1 r r + − κ = − . (43) + 2 r2 +a2 (cid:18) + (cid:19) 1 r r + − κ = − . (44) − 2 r2 +a2 (cid:18) − (cid:19) κ is the surface gravityat the outer horizon. Here again, the upper sign is for + r >r , and the lower sign is for r <r . The future outer horizon is defined as + + U =0,V >0. + + The metric near the horizon(r r ,U 0) in the Kruskal coordinates + + → → takes the form ds2 = 4ρ2 (r r )κκ−+−1e−2κ+r+dU dV +ρ2dθ2 ∓ + +− − + + + + 4a2sin2θ(r r )2κκ−+−2 r+2 + r+2 −a2 e−4κ+r+V2dU2 +− − ρ2 r2 +a2 + + (cid:18) + + (cid:19) 2asin2θ(r r )κκ−+ 1+ r+ e−2κ+r+V dU dφ˜ ± +− − κ ρ2 + + (cid:18) + +(cid:19) r2 +a2 2 + + sin2θdφ˜2. ρ2 (cid:0) + (cid:1) So,thecoordinatesingularityisremovedandthemetricisregularatthehorizon in the U ,V coordinates. + + Just as for the RN spacetime, the coordinates (U ,V ) are singular at the + + inner horizon (r =r ), and another coordinate patch is required to extend the − Kerr metric beyond this horizon. For the inner horizon we define r2+a2 r = dr. (45) ∗ ∆ Z at ψ˜ = φ . (46) − 2Mr − With u=t r and v =t+r , the appropriate choice for Kruskal coordinates ∗ ∗ − are, U = eκ−u,V = e−κ−v. (47) − − ∓ − The upper sign refers to r > r and the lower sign refers to r < r . Near the − − horizon (r r ,U 0) the metric becomes − − → → ds2 = ∓4ρ2−(r+−r−)κκ−+−1e2κ−r−dU−dV−+ρ2−dθ2 + 4a2sin2θ(r+−r−)2κκ−+−2 ρr−22 + rr−22 +−aa22 e4κ−r−V−2dU−2 (cid:18) − − (cid:19) κ− r ± 2asin2θ(r+−r−)κ+ κ +ρ2 −1 e2κ−r−V−dU−dψ˜ (cid:18) + + (cid:19) r2 +a2 2 + − sin2θdψ˜2. ρ2 (cid:0) − (cid:1) Themetricisregularatr=r ,andthespacetimecanbeextendedbeyondthis − horizon. 7 3.2 Scalar modes and the probability current For the Kerr metric the scalar modes can be separated as [35] 1 Ψ (r ,t) Ψ = ω ∗ eimφΘ(θ). (48) ωm √4πω r The Killing vectorin this caseis (∂ +Ω ∂ ) andconsidering only the positive t H φ frequency solutions as before we get i(∂ +Ω ∂ )Ψ =ωΨ . (49) t H φ ωm ωm Here, ω =E mΩ where E =ω is the frequency at infinity. H ∞ − In Kruskal coordinates the Killing vector becomes ∂φ˜ ∂φ˜ ∂ +Ω ∂ = ∂ +∂ + ∂ +Ω ∂ (50) t H φ u v ∂t φ˜ H∂φ φ˜ = κ U ∂ +κ V ∂ . (51) − + + U+ + + V+ just the same as the RN spacetime, and hence the U and the V modes are + + decoupled, Ψ =[f (U )+g (V )]. (52) ω ω + ω + Concerning ourselves with only the outgoing U modes, we get the solutions + iω fω(U+)=Nω U+ κ+. (53) | | Againasbefore,thesemodesarewellbehavedbothinsideandoutsidethehori- zonandisnotdifferentiableatthehorizonbecauseofthelogarithmicsingularity. Andwehavetoresortbacktothedistributionstocalculatetheprobabilitycur- rent for emission. The distributions have the same form as in the case of RN spacetime, and the probability current through the outer horizon is N 2 for U >0, jout =( |Nωω|2e−2κπ+ω for U++ <0. (54) | | For the inner horizon, using the specific Kruskal coordinates, the probability current is the same as the RN inner horizon N˜ 2 for U >0, jout =( |N˜ωω|2e−2κπ−ω for U−− <0. (55) | | 3.3 Extreme limit of non-extremal solutions Anon-extremalspacetimewithanouter(r )andaninner(r )horizonbecomes + − extremalasr r . Thetwohorizonsareinequilibriumattwodifferenttem- + − → peratures and as a result, the outgoing fluxes are also different. As was shown earlier, both for the Reissner-Nordstro¨m and Kerr metrics, the temperature of the outer horizon is, T = h¯κ+ and that of the inner horizon is T = h¯κ−. out 2π in 2π Since k >k the inner horizonis in equilibriumat a higher temperature than − + that outer one. So the outgoing flux from the outer horizon, given by (30) and (52) for the R-N and Kerr spacetimes respectively, are less than the incoming 8 flux through the inner horizon given by (31) and (53) respectively. The results are consistent with expectations. In the extremal limit as r r , T 0 + − out → → as κ =0. So no thermality is observedat the outer horizonas expected. This + canbeshownmoreclearlybyconsideringaneffectivetemperaturefortheouter horizon. For both RN and Kerr black holes, since the spacetime between the two horizons for (r <r <r ) is vacuum, the fluxes have to match. This implies, − + N˜ω 2e−2κπ−ω = Nω 2. (56) | | | | So, the effective flux coming out of the outer horizon is given by, jout = Nω 2e−2κπ+ω = N˜ω 2e−2πω κ1++κ1− . (57) | | | | (cid:0) (cid:1) This gives an effective temperature of the outer horizon as the harmonic mean of κ and κ , + − 2π 1 1 β = + . (58) eff ¯h κ κ (cid:18) + −(cid:19) Far from extremality, M Q, as a result r r and κ κ . Thus, + − − + ≫ ≫ ≫ T T . eff out ≈ But in the extremal limit, as r r , T = 0. This shows that the − + eff → outgoing flux approaches zero not only on the outer horizon but on the inner horizon as well. This limit is consistent with what we expect from an extremal solution. 4 Extremality The nature of the extremal metric is quite different from other stationary so- lutions. Still we can calculate the scalar modes and the outgoing probability current through the horizon following the same procedure as that of a station- ary metric. Naturally, the mode solutions near the horizon is different and this leads to a different type of distributions for the extremal case. We found that thoughthe flux vanishes preciselyatthe horizon,leadingto a zerotemperature for the horizon, it is non-zero both outside and inside of the horizon, which is not physically acceptable. However, no such problems arise if we consider the extremal spacetime as a limiting case of a non-extremal one. 4.1 Extremal solution The extreme Reissner-Nordstro¨mmetric is M 2 dr2 ds2 = 1 dt2+ +r2dΩ2. (59) − − r 1 M 2 (cid:18) (cid:19) − r where M is the mass of the black hole. Th(cid:0)e horizo(cid:1)n is at r = M. Introducing the null coordinates u=t r , v =t+r ,with ∗ ∗ − dr M2 r =r+2Mln r M . (60) ∗ ≡ 1 M 2 | − |− r M Z − r − (cid:0) (cid:1) 9 the surface r = M appears at v u = . The Kruskal coordinates U,V are − −∞ given by the implicit relations, u = McotU. (61) − v = MtanV. (62) − The future horizon is located at U =0,V < π. 2 The Killing vector ∂ is t 1 ∂ = sin2U∂ cos2V∂ . (63) t U V M − AgaintheU andtheV modesare(cid:2)decoupledandthepo(cid:3)sitivefrequencysolutions for a scalar field are found to be Ψ (U) = N eiωMcotU. (64) ω ω Ψ (V) = N eiωMtanV. (65) ω ω The ingoing V-modes are regular across the horizon. Near the horizon the outgoing U-modes takes the form Ψω(U)=NωeiωMcotU NωeiωUM. (66) ≃ The singularity at the horizon can be removed by using distributions. In this case the appropriate distributions are found by taking logarithm of the modes, iωM lnΨ (U) = lnN +lim . (67) ω ω ǫ→0U iǫ − iωM lnΨ (U) = lnN∗ lim . (68) ω ω−ǫ→0U +iǫ The extremal solution for the Kerr spacetime is obtained by setting a = M in the Kerr metric. The line element becomes, 2Mr 4M2rsin2θ Σ ρ2 ds2 = 1 dt2 dtdφ+ sin2θdφ2+ dr2+ρ2dθ2. (69) − − ρ2 − ρ2 ρ2 ∆ (cid:18) (cid:19) with, ρ2 = r2+M2cos2θ. (70) ∆ = (r M)2. (71) − Σ = r2+M2 2 M2∆sin2θ. (72) − The horizon is situated at r =(cid:0)M. Intr(cid:1)oducing the null coordinates u = t − r ,v =t+r as before, with ∗ ∗ r2+M2 2M2 r dr =r+2Mln r M . (73) ∗ ≡ (r M)2 | − |− r M Z − − 1 φ˜ = φ t=φ Ωt (74) − 2M − the surface r = M appears at v u = . The Kruskal coordinates U,V are − −∞ the same as in the case of extremal Reissner-Nordstro¨mmetric, u = McotU. (75) − v = MtanV. (76) − 10