Duality-Invariant Higher-Derivative Corrections to Charged Stringy Black Holes

Dualit y-In v arian t Higher-Deriv ativ e Corrections to Charged Stringy Blac k Holes Upaman yu Moitra ∗ Institute for The or etic al Physics, Institute of Physics, Universiteit van A mster dam, Scienc e Park 904, 1089 XH A mster dam, The Netherlands W e study dualit y-inv ariant higher-deriv ativ e corrections to the charged black hole geometry in t wo-dimensional heterotic string theory . W e illustrate ho w the conv en tional p erturbativ e approac h to determine the corrected geometry breaks down. Using a non-p erturbativ e (in α ′ ) parametrization of the solution, we find the corrected c harge-to-mass ratio for extremal black holes. W e remark on the results in relation to the weak gravit y conjecture. W e also consider the entrop y of the extremal blac k hole within the attractor mechanism and find that the t wo-deriv ative en tropy is not renormalized to an y order. W e make comments on in terpretations of the results and their extension to near-extremal blac k holes. INTR ODUCTION Blac k holes (BHs) are essential ob jects in our efforts to understand v arious features of quantum gra vity (QG). String theory , a leading candidate for a unified theory of all known interactions, has elucidated man y aspects of QG. While general questions in v olving BHs in string theory can b e intractably difficult to answer, a considera- tion of low er spacetime dimensions often simplifies com- putations, while offering insigh ts ab out general asp ects of gravit y . Ev en in lo w dimensions, ho w ever, it is ex- tremely imp ortan t to take in to accoun t the straitjac k- ets that mak e string theory comp elling as a theory of QG. Among such constraints, target space duality or T - dualit y [ 1 ], describing the equiv alence of physics on small and large circles, is particularly imp ortan t. It is kno wn that at tree-level, strings compactified on a d -dimensional torus hav e O( d, d ; R ) as an exact duality symmetry group [ 2 ]. In the recen t years, there has b een significan t progress in understanding dualit y-symmetric theories with dou- ble field theory-based metho ds [ 3 ]. Inspired by the early w ork [ 4 ], there now exists a complete classification of p er- turbativ e higher-deriv ativ e (HD) corrections for cosmo- logical and 1 + 1 -dimensional static backgrounds [ 5 ] in the massless sector of closed string theory . In the recent w ork [ 6 ], we extended this classification program to in- clude heterotic strings and the asso ciated gauge fields, truncated to the Cartan of the gauge group. F or a single Maxw ell field, the duality group is O(1 , 2; R ) . The ex- tension of the formalism in [ 6 ] allows for a detailed study of tw o-dimensional charged BHs in the heterotic theory , first discov ered in [ 7 ], and v arious in teresting asso ciated questions. W e also show ed in [ 6 ] how a non-p erturbativ e parametrization [ 8 , 9 ] of the solution in the presence of arbitrary HD terms could be extended to the heterotic case. The goal of the presen t work is to study some phys- ically interesting asp ects of charged BHs within this framew ork. Charged BHs with degenerate horizons are kno wn as extr emal . Extremal BHs hav e pla yed a crucial role in illuminating v arious aspects of QG, from micro- scopic state-counting [ 10 ] to the ongoing efforts of map- ping the landscap e of consistent QG theories [ 11 ], known as the Swampland program. One statement of the we ak gr avity c onje ctur e (W GC) [ 12 ], a central part of this pro- gram, is that extremal BHs should be able to decay by emitting particles of sufficiently high charge-to-mass ra- tio. It w as also suggested [ 13 ] that BHs themselves could meet the criterion of such particles: consistent HD correc- tions in the low-energy effective field theory (EFT) were argued to incr e ase the extremal c harge-to-mass ratio. One motiv ation of this work is addressing such ques- tions in the tw o-dimensional framework. There are many kno wn subtleties related to the Sw ampland program in d = 2 [ 14 ]. F or instance, the no-glob al-symmetries conjec- ture is b eliev ed to be generally v alid only for d ≥ 4 , when gra vity is dynamical. In d = 2 , b oth gravit y and electro- magnetism are non-dynamical and the electrostatic p o- ten tial for a p oin t charge gro ws linearly with distance. A general statemen t of WGC cannot b e easily extended to d = 2 , 3 ; see [ 15 ]. Therefore, it is not a priori clear what a Sw ampland-like statement would resem ble and whether one can mak e any suc h universal statemen t at all. On the other hand, the BH w e consider is a gen uine string bac kground arising from a w orld-sheet description. It has b een argued [ 16 ] that contin uous global symmetries on w orld-sheet alwa ys translate to gauge symmetries in spacetime. Given this, one do es exp ect an absence of global symmetries in the background and consequently , something akin to WGC. W e hop e that this work is the first step tow ards addressing suc h subtle questions. In what follo ws, we explore v arious physical conse- quences of duality-in v ariant HD corrections. First, we calculate the corrected extremal c harge-to-mass ratio. While part of the motiv ation was outlined in the previ- ous paragraph, this calculation, not previously attempted in the literature, is of broader interest in BH ph ysics. W e first perform the computation in p erturbation the- ory with only the four-deriv ative terms and encounter a surprising difficult y . T o circumv en t this, we use a non- p erturbativ e formalism [ 6 , 8 , 9 ]. Pleasingly , w e are able to write do wn an explicit form ula for the ratio, with a remarkable feature, whic h we comment up on. 2 W e next turn our attention to the calculation of W ald en tropy [ 17 ] of the extremal BH with the HD terms within the attractor mechanism [ 18 ]. T o this end, w e apply the entrop y function formalism [ 19 ] prop osed b y Sen. W e find yet another striking result concerning the non-renormalization of the t wo-deriv ative en tropy . W e conclude the pap er with some comments on interpreta- tions of the foregoing results, BHs slightly aw a y from extremalit y and p ossible future directions. W e hav e rel- egated some formulas to the End Matter on pp. 6 . CORRECTED EXTREMAL CHARGE-TO-MASS RA TIO The t w o-deriv ative target space action describing the massless sector of heterotic strings inv olving the metric g µν , string dilaton ϕ and Maxwell field A µ is I (2) = Z d 2 x √ − g e − 2 ϕ  ξ 2 + R + 4( ∇ ϕ ) 2 − F µν F µν 4  , (1) where ξ 2 ≡ 8 /α ′ , √ α ′ b eing the string length. W e hav e absorb ed the Newton constant in ϕ . The full action is augmen ted with HD terms, inv olving combinations of R , F µν and ∇ µ ϕ and their deriv ativ es. While the general HD action is complicated in form, enormous simplifica- tions o ccur under the assumption of duality in a time- indep enden t background describ ed b y d s 2 = − m ( x ) 2 d t 2 + n ( x ) 2 d x 2 , ϕ = ϕ ( x ) , A µ = V ( x ) δ t µ . (2) Defining the dualit y-inv ariant dilaton Φ( x ) = 2 ϕ ( x ) − log | m ( x ) | , (3) under appropriate field redefinitions [ 5 , 6 ] the full ac- tion, dimensionally reduced, can be shown to assume a remarkably simple O(1 , 2; R ) -inv ariant form, I full = Z d x n ( x ) e − Φ  ξ 2 + Φ ′ ( x ) 2 n ( x ) 2 + F ( ρ )  , (4) where F ( ρ ) is a function of the single v ariable ρ , a dualit y-inv arian t combination of the curv ature and gauge field-strength, F ( ρ ) = − 1 2 ρ 2 + ∞ X n =2 c 2 n ρ 2 n , ρ 2 ≡ 2 m ′ ( x ) 2 − V ′ ( x ) 2 n ( x ) 2 m ( x ) 2 . (5) W e can generalize the discussion to include r ab elian gauge fields, leading to the symmetry group O(1 , 1+ r ; R ) . Remarkably , duality is so constricting that there is only one Wilson coefficient at eac h deriv ativ e order [ 6 ]. F or definiteness, w e consider a single Maxwell field. The t w o-deriv ativ e action admits the well-kno wn c harged BH solution [ 7 ] m ( x ) 2 = 1 − 2 µe − ξx + Q 2 e − 2 ξx = n ( x ) − 2 , ϕ ( x ) = − 1 2 ξ x, V ( x ) = − √ 2 Qe − ξx . (6) The physical mass and charge are 2 ξ µ and √ 2 ξ Q resp ec- tiv ely but w e will k eep using the parameters µ and Q . The BH is non-extremal for µ > | Q | and extremal for µ = | Q | . Henceforth, w e consider a p ositiv e Q . The BH has horizons at x ± = ξ − 1 log  µ ± p µ 2 − Q 2  . HD terms are known to mo dify such a linear mass- c harge relationship, characteristic of asymptotically flat extremal BHs in higher dimensions. It w as suggested [ 13 ] that the mo dification o ccurs in such a wa y that ( Q/µ ) ext ≥ 1 . Finding the nature of HD-corrected so- lutions in 2D, besides their relev ance for a possible for- m ulation of W GC, is a worth while goal in itself, a result that seems to b e missing from the literature. F ailure of P erturbation Theory W e emplo y p erturbation theory to calculate the cor- rected ( Q/µ ) ext due to the duality-in v ariant HD terms. In the function F ( ρ ) in ( 5 ), we truncate the series at the four-deriv ative term with c 4 = ϵc ( ϵ ≪ 1 ) and w ork to O ( ϵ ) . The background fields ( 6 ) receive O ( ϵ ) corrections, m = m + ϵ δ m, n = n + ϵ δ n, V = V + ϵ δ V . (7) W e choose a gauge where the string dilaton is fixed to ϕ = ϕ . The dilaton actually defines the spatial co ordinate x : leaving it fixed is analogous to considering the size of the transverse space fixed in higher dimensions. W e can solve for the equations of motion arising from ( 4 ) and imp osing suitable b oundary conditions at asymptotic infinit y , x → ∞ , to ensure that the corrections are sub- leading. The solution is giv en in eq. ( 25 ). Rather surprisingly , perturbation theory breaks down near the horizon. Near the unp erturbed even t horizon x = x + , the curv ature and electric field strength b eha v e as, see eq. ( 26 ), R ∼ ϵcξ 2 ( x − x + ) 2 , F 2 ∼ ϵcξ 3 x − x + , (8) whic h indicates a failure of the p erturbativ e approac h near the non-extr emal horizon. Such a dramatic break- do wn of p erturbation theory is disconcerting and sur- prising from an EFT p oin t-of-view. Th us, any hope of systematically studying the effects of HD terms on BHs w ould app ear to b e lost. 3 Non-P erturbative P arametrization There does exist a framework to obtain a solution in presence of a full tow er of HD terms. In [ 6 ], a non- p erturbativ e parametrization for c harged BHs was de- riv ed, building on the w orks [ 8 , 9 ]. The solution pro ceeds from the assumption that the deriv ative f ( ρ ) = F ′ ( ρ ) of the function app earing in ( 5 ) has a single-v alued in- v erse ρ = ρ ( f ) . In this f -parametrization, the duality- in v arian t dilaton is simply Φ = log ( f /ξ ) + Φ 0 and in the gauge n ( x ) = 1 , the BH exterior is describ ed by [ 6 ] x = x 0 + Z ∞ f d f ′ 1 f ′ p ξ 2 − G ( f ′ ) , m ( f ) 2 = 1 Q 2 csc h 2 " Z f f m d f ′ ρ ( f ′ ) √ 2 f ′ p ξ 2 − G ( f ′ ) # , V ( f ) = Q Z f 0 d f ′ ρ ( f ′ ) m ( f ′ ) 2 f ′ p ξ 2 − G ( f ′ ) , (9) where x 0 , Q and f m < 0 are constan ts and G ( f ) = R f 0 d f ′ ρ ( f ′ ) . F or the tw o-deriv ative theory , ρ ( f ) = − f . F or an HD extension, one considers [ 9 ] a p erturbation of the form, ρ ( f ) = − f  1 + h ( f 2 )  , (10) where h ( z ) is a function satisfying h (0) = 0 and deca ying sufficien tly fast at infinity . The function is b ounded in a w ay so that ξ 2 − G ( f ) > 0 betw een f = 0 (spatial infinity) and f = ∞ (even t horizon). In terms of the function β ( y ) ≡ Z y 0 d y ′ h ( y ′ ) , (11) w e hav e G ( f ) = − 1 2  f 2 + β ( f 2 )  . (12) Th us, h ( f 2 ) = β ′ ( f 2 ) . F or a BH solution, β ( ∞ ) ≡ β 0 m ust be finite [ 9 ]. W e can now see wh y the example of the ρ 4 term in the previous subsection failed to work. In this case, for large f , h ( f 2 ) ∼ 1 and hence | β 0 | = ∞ and there is no BH. A related example was also studied in [ 9 ]. W e can now readily mak e use of the general solu- tion ( 9 ) to find the extremal c harge-to-mass ratio. It is conv enien t to express the solution in Sc hw arzsc hild- Droste–lik e gauge ( 6 ). T o b e precise, we shall use the v arious transformation form ulas to use the string dilaton 2 ϕ = Φ + log m as a co ordinate, whereb y we obtain a line elemen t of the form d s 2 = − m ( ϕ ) 2 d t 2 + n ( ϕ ) 2 4 d ϕ 2 ξ 2 . (13) W e first p erform an expansion ab out asymptotic infin- it y . The solution is given b y ( 28 ); up to the order shown, it is identical in form to the tw o-deriv ative result ( 6 ). The HD terms change only the terms sub-leading to ( 6 ). W e th us identify Q = csch γ and the charge-to-mass ratio, Q µ = sec h γ , (14) where γ ≡ Z 0 f m d f ′ 1 + h ( f ′ 2 ) p 2 ξ 2 + f ′ 2 + β ( f ′ 2 ) . (15) The extremality condition, on the other hand, m ust b e deduced from a near-horizon ( f = ∞ ) expansion, eq. ( 29 ). In this case, assuming β ( f 2 ) admits an ex- pansion of the form P ∞ n =0 β n f − 2 n , (neglecting possible exp onen tially small corrections) we find from the near- horizon metric ( 30 ) the extremality condition to b e given b y ∆ = 0 , where ∆ ≡ 4 ξ 2 − e 2( γ + χ )  β 0 + 2 ξ 2  , (16) with χ ≡ Z ξ 0 d f ′ 1 + h ( f ′ 2 ) p 2 ξ 2 + f ′ 2 + β ( f ′ 2 ) + Z ∞ ξ d f ′ 1 + h ( f ′ 2 ) p 2 ξ 2 + f ′ 2 + β ( f ′ 2 ) − 1 f ′ ! . (17) W e ha v e to p erform a suitable rescaling of the co or- dinates to transform the metric components ( 30 ) to a more familiar form. Note that, in general, the co ordi- nate system breaks do wn in the limit ∆ → 0 . This is not unexpected from the discussion in [ 6 ]. (How ev er, as clarified in the End Matter, the limit is smooth for ex- p onen tially small corrections to β ( ∞ ) : when β n = 0 for n > 0 .) Since our goal is to find precisely this limiting v alue, w e need not worry ab out this p ossibilit y . Setting ∆ = 0 gives the corrected extremal charge-to-mass ratio,  Q µ  ext = 4 ξ p β 0 + 2 ξ 2 4 e − χ ξ 2 + e χ ( β 0 + 2 ξ 2 ) . (18) This is one of the main results in this pap er, whic h ex- presses the extremal charge-to-mass ratio in terms of the parameters of the theory . Notice that this relation is state-indep enden t. By the AM-GM inequalit y , w e im- mediately see that irresp ectiv e of the form of the correc- tions, ( Q/µ ) ext ≤ 1 . This bound is completely general in a dualit y-in v arian t theory and cannot be c hanged by tuning the Wilson co efficien ts. The tw o-deriv ative theory saturates the b ound but is not unique in this regard. W e see that the bound is opp osite of the usual W GC b ound in higher dimensions. This is not to o surprising in the presen t tw o-dimensional scenario, for reasons men- tioned in the introduction. It w ould be interesting to further explore related asp ects in this set-up. 4 A TTRACTOR MECHANISM AND ENTROPY Extremal BHs hav e also been crucial in our under- standing of BH thermo dynamics and statistical mechan- ics, esp ecially vis-à-vis state-coun ting [ 10 ]. Extremal BHs often exhibit the attractor mec hanism [ 18 ], in which the near-horizon prop erties of the BH are determined only by the charges carried b y it and not by asymp- totic moduli. In the near-horizon region, the BHs ha ve a tw o-dimensional Anti-de Sitter ( AdS 2 ) geometry and a co v arian tly constant electric field. The entrop y function formalism [ 19 ] is an elegan t metho d to calculate the W ald en tropy [ 17 ] due to the HD terms in the action. W e no w turn our attention to this problem. The tw o-dimensional heterotic BH entrop y was also considered in [ 20 ] in a spe- cific, limited con text with the HD corrections in volving only the Ricci scalar. In this formalism, one starts with a t w o-dimensional Lagrangian densit y (p ossibly after in tegrating o ver the transv erse directions, which does not apply here). In- serting in this Lagrangian the following ansätze for the metric, gauge field and scalar, d s 2 = v  − x 2 d t 2 + d x 2 x 2  , V ′ ( x ) = e , ϕ ( x ) = ϕ 0 , (19) one obtains the function f ( ϕ 0 , v , e ) . One then defines the Legendre transform E ( ϕ 0 , v , e , q ) = 2 π ( eq − f ( ϕ 0 , v , e )) (20) and sets eac h of ∂ E / ∂ v , ∂ E / ∂ ϕ 0 and ∂ E / ∂ e to zero (kno wn as the attr actor e quations ). Inserting the solution in ( 20 ) one obtains the entrop y . It is worth explaining one technical detail at length. The Lagrangian density we use m ust b e supplemen ted with an additional total-deriv ativ e term in comparison with ( 4 ), L = ne − Φ  ξ 2 + Φ ′ 2 n 2 + F ( ρ )  − 2 d d x  e − 2 ϕ m ′ n  . (21) While the inclusion of this term do es not affect the equa- tions of motion, it do es matter in the en tropy computa- tion. A t the tw o-deriv ative level, the generally cov ariant Lagrangian ( 1 ) reduces precisely to the form ( 21 ). In ob- taining the dualit y-inv arian t form ( 4 ), w e discard the to- tal deriv ativ e [ 6 ]. In the deriv ation of the HD action ( 4 ), one p erforms rep eated integration by parts [ 5 , 6 ]. These t wo pro cedures are, how ever, not on equal fo oting. In the second case, w e use the tw o-deriv ative equations of mo- tion to perform field redefinitions. These are consisten t with cov ariant field redefinitions; see, for example, [ 4 ]. It is kno wn that the entrop y ev aluated in this formalism is in v arian t under such field redefinitions [ 19 ]. Inserting the ansätze ( 19 ) in ( 21 ), the suitable combi- nations of the first tw o attractor equations yield e 2 − 2 v 2 v 2 + ∞ X n =2 c 2 n  2 v − e 2 v 2  n − 1 − v ξ 2 v = 0 , ( e 2 − v ) " 1 − 2 ∞ X n =2 nc 2 n  2 v − e 2 v 2  n − 1 # − v = 0 . (22) It is not hard to see that the equations admit the exact, ph ysically unique solution e 2 = 2 /ξ 2 = 2 v ,. F rom the third attractor equation, w e also find e − 2 ϕ 0 = | q | / ( √ 2 ξ ) , whereb y we find the entrop y , S = 2 √ 2 π ξ − 1 | q | , (23) whic h exhibits no dep endence on the Wilson co efficien ts. The HD terms therefore do not renormalize the leading order tw o-deriv ative en trop y . This non-renormalization is another key result of this pap er. It is w orth p oin ting out some in teresting features of the en tropy . Note that the A dS 2 radius is of order the string length, L AdS = p α ′ / 8 . This necessitates that w e take in to accoun t all HD terms in the effective ac- tion. W e implicitly assumed in ( 5 ) that the coefficients c 2 n deca y sufficiently fast so that the infinite sum con- v erges at least within some finite radius; see [ 9 ]. Note that the dualit y-in v ariant combination ρ 2 ensures that non-renormalization p ersists even if we truncate the sum in ( 5 ) to any finite num b er of terms. The constant dila- ton scenario here is v ery different from the linear dilaton case b efore in that it is p ossible to find a consisten t BH solution even with a finite num ber of HD terms. F urther- more, unlike in many typical examples, the near-horizon electric field e is not a free parameter but fixed b y the attractor equations. It would b e interesting to examine if the quan tization of c harge plays an y role in this non- renormalization. W e emphasize that the non-renormalization is note- w orthy for the reason that we work ed in an EFT frame- w ork with no input apart from duality inv ariance. One is comp elled to w onder whether this EFT “knows” physics b ey ond this input. A relation betw een the attractor mec hanism and top ological strings has b een conjectured in [ 21 ]. It w ould b e in teresting to examine if the entrop y is protected b ecause it describ es some top ological sector of the theory . DISCUSSION The different results in the previous sections are in- triguing from multiple p erspectives. Up on the introduc- tion of a single duality-in v ariant HD term, p erturbation theory breaks down rather dramatically , against EFT exp ectations, and a singularit y seems to develop in the vicinit y of the BH horizon, originally present in the tw o- deriv ative theory . It would b e interesting to examine the 5 formation and dynamical b eha vior of suc h a singularit y [ 22 ]. The situation is ameliorated by the introduction of an infinite num ber of HD terms, satisfying certain conditions outlined previously . Using the non-p erturbativ e form of the solutions, w e calculated an explicit form of the HD correction to the extremal charge-to-mass ratio. Remark- ably , w e found a sp ecific b ound for this ratio, alb eit one that go es in the opposite direction compared to higher dimensions [ 13 ]. The existence of the b ound itself is in- dicativ e of univ ersalit y of some kind: it would b e inter- esting to find similar properties of the t w o-dimensional string landscap e in the spirit of the Swampland program. In relation to the extremal c harge-to-mass ratio, the non-renormalization of the attractor entrop y might seem a little puzzling at first. It w as argued in [ 23 ] that the dif- ference in entrop y due to HD terms is non-negative — a result that m ust hold in general — and also prop ortional to the correction to the charge-to-mass ratio. While we ha ve ∆ S = 0 , the second part of the argument do es not extend to our case. There is no con tradiction because the higher-dimensional deriv ation relied on p erturbation theory , which we hav e found to break down in 2D. The results can b e easily generalized to r Maxwell fields [ 6 ]. F or instance, the form of ( 18 ) would b e the same with the replacement Q →     Q    . In the entrop y calculation, one finds similarly  e 2 = 2 /ξ 2 = 2 v . This further suggests a role of charge quantization in the non-renormalization of the en tropy since only one out of r Maxwell fields seems to b e relev ant to the story . One would hav e in this case, similar to ( 23 ), S = 2 √ 2 π ξ − 1 |  q | . The non-renormalization of the en tropy in vites a more refined understanding of the underlying mechanism. It w as p oin ted out in [ 24 ] that the tw o-dimensional extremal BH can b e obtained as an orbifold of an extremal AdS 3 BH. Non-renormalization of the micr osc opic en tropy in A dS 3 has been argued for in [ 25 ]. It w ould be in terest- ing to see ho w our lo w-energy EFT captures suc h mi- croscopic details irresp ectiv e of the v alues of the Wilson co efficien ts. It is w orth emphasizing that without in- sisting on duality inv ariance, such a non-renormalization w ould not hold true: generic HD corrections would mo d- ify ( 23 ). W e ha v e only considered the string tree-level action [ 26 ]. A consideration of the quantum entrop y [ 27 ] w ould shed more ligh t on the microscopic asp ects — the dualit y group w ould also b e broken do wn to a discrete subgroup. In this resp ect, it might b e w orth consider- ing the torus partition function of strings on orbifolds of A dS 3 [ 28 ]. Considering a small departure from extremality at the attractor point would b e an interesting direction to ex- plore. While the Jackiw-T eitelb oim (JT) mo del [ 29 ], suit- able for describing BHs near extremalit y [ 30 ], is typi- cally obtained via dimensional reduction, w e can directly obtain the JT model and the corresponding Sch w arzian from the tw o-deriv ativ e action in the near-horizon region without an y reduction. One can combine the results of [ 30 ] and [ 31 ], and directly obtain the T 3 / 2 correction to the BH partition function. The relative simplicit y of tw o- dimensional string theory would p erhaps facilitate com- parison betw een the effective and microscopic theories. W e leav e suc h promising questions for future in vestiga- tions. A ckno wledgments My research is supp orted by the Europ ean Union’s Horizon 2020 Research and Inno v ation Programme (Gran t Agreement No. 101115511). ∗ u.moitra@uv a.nl [1] A. Giveon, M. Porrati, and E. Rabinovici, Phys. Rept. 244 , 77 (1994) , arXiv:hep-th/9401139 . [2] K. A. Meissner and G. V eneziano, Ph ys. Lett. B 267 , 33 (1991) ; A. Sen, Phys. Lett. B 271 , 295 (1991) ; M. Gasp erini and G. V eneziano, Phys. Lett. B 277 , 256 (1992) , arXiv:hep-th/9112044 ; J. Maharana and J. H. Sch w arz, Nucl. Phys. B 390 , 3 (1993) , arXiv:hep- th/9207016 . [3] W. Siegel, Ph ys. Rev. D 48 , 2826 (1993) , arXiv:hep- th/9305073 ; C. Hull and B. Zwiebac h, JHEP 09 , 099 (2009) , arXiv:0904.4664 [hep-th] . [4] K. A. Meissner, Ph ys. Lett. B 392 , 298 (1997) , arXiv:hep- th/9610131 . [5] O. Hohm and B. Zwiebach, Phys. Rev. D 100 , 126011 (2019) , arXiv:1905.06963 [hep-th] ; T. Co dina, O. Hohm, and B. Zwiebach, Phys. Rev. D 108 , 026014 (2023) , arXiv:2304.06763 [hep-th] . [6] U. Moitra, (2025), arXiv:2511.21687 [hep-th] . [7] M. D. McGuigan, C. R. Nappi, and S. A. Y ost, Nucl. Ph ys. B 375 , 421 (1992) , arXiv:hep-th/9111038 . [8] M. Gasp erini and G. V eneziano, JHEP 07 , 144 (2023) , arXiv:2305.00222 [hep-th] . [9] T. Co dina, O. Hohm, and B. Zwiebach, Ph ys. Rev. D 108 , 126006 (2023) , arXiv:2308.09743 [hep-th] . [10] A. Strominger and C. V afa, Phys. Lett. B 379 , 99 (1996) , arXiv:hep-th/9601029 . [11] C. V afa, (2005), arXiv:hep-th/0509212 . [12] N. Arkani-Hamed, L. Motl, A. Nicolis, and C. V afa, JHEP 06 , 060 (2007) , arXiv:hep-th/0601001 . [13] Y. Kats, L. Motl, and M. Padi, JHEP 12 , 068 (2007) , arXiv:hep-th/0606100 . [14] J. McNamara and C. V afa, (2020), [hep-th] . [15] U. Moitra, Ph ys. Rev. D 109 , L041903 (2024) , arXiv:2305.08907 [hep-th] . [16] T. Banks and L. J. Dixon, Nucl. Phys. B 307 , 93 (1988) . [17] R. M. W ald, Phys. Rev. D 48 , R3427 (1993) , arXiv:gr- qc/9307038 . [18] S. F errara, R. Kallosh, and A. Strominger, Ph ys. Rev. D 52 , R5412 (1995) , arXiv:hep-th/9508072 . 6 [19] A. Sen, JHEP 09 , 038 (2005) , arXiv:hep-th/0506177 ; Gen. Rel. Grav. 40 , 2249 (2008) , arXiv:0708.1270 [hep- th] . [20] S. Hyun, W. Kim, J. J. Oh, and E. J. Son, JHEP 04 , 057 (2007) , arXiv:hep-th/0702170 ; J. Sadeghi, M. R. Setare, and B. Pourhassan, Acta Phys. Polon. B 40 , 251 (2009), arXiv:0707.0420 [hep-th] . [21] H. Ooguri, A. Strominger, and C. V afa, Ph ys. Rev. D 70 , 106007 (2004) , arXiv:hep-th/0405146 . [22] U. Moitra, JHEP 06 , 194 (2023) , arXiv:2211.01394 [hep- th] . [23] C. Cheung, J. Liu, and G. N. Remmen, JHEP 10 , 004 (2018) , arXiv:1801.08546 [hep-th] . [24] A. Giveon and A. Sever, JHEP 02 , 065 (2005) , arXiv:hep- th/0412294 . [25] A. Dabholkar, A. Sen, and S. P . T rivedi, JHEP 01 , 096 (2007) , arXiv:hep-th/0611143 . [26] Since the HD action is constructed perturbatively , one can en tertain the possibility that the non-renormalization is non-p erturbativ ely violated. [27] A. Sen, Int. J. Mo d. Phys. A 24 , 4225 (2009) , arXiv:0809.3304 [hep-th] . [28] A. Dabholkar and U. Moitra, JHEP 06 , 053 (2024) , arXiv:2312.14253 [hep-th] ; A. Dabholkar, E. Harris, and U. Moitra, JHEP 02 , 136 (2026) , arXiv:2507.15939 [hep- th] . [29] C. T eitelb oim, Phys. Lett. B 126 , 41 (1983) ; R. Jackiw, Nucl. Phys. B 252 , 343 (1985) . [30] U. Moitra, S. K. Sak e, S. P . T rivedi, and V. Vishal, JHEP 11 , 047 (2019) , arXiv:1905.10378 [hep-th] ; U. Moitra, S. P . T rivedi, and V. Vishal, JHEP 07 , 055 (2019) , arXiv:1808.08239 [hep-th] ; U. Moitra, A sp e cts of Quan- tum Gr avity, Holo graphy and Entanglement , Ph.D. the- sis, T ata Institute of F undamental Research. [31] D. Stanford and E. Witten, JHEP 10 , 008 (2017) , arXiv:1703.04612 [hep-th] . END MA TTER P erturbative and Non-Perturbativ e Solutions to the Equations of Motion F or the p erturbativ e analysis, as describ ed in the main text, we consider the HD action of the form ( 4 ) with a single four-deriv ative term, I (4) = Z d x n ( x ) e − Φ( x ) " ξ 2 + Φ ′ ( x ) 2 n ( x ) 2 − 2 m ′ ( x ) 2 − V ′ ( x ) 2 2 n ( x ) 2 m ( x ) 2 + ϵc  2 m ′ ( x ) 2 − V ′ ( x ) 2 n ( x ) 2 m ( x ) 2  2 # , (24) with ϵ ≪ 1 a p erturbativ e parameter and the duality-in v ariant dilaton Φ related to the string dilaton ϕ b y ( 3 ). With the O  ϵ 0  solution given by ( 6 ), we consider p erturbativ e ansätze of the form ( 7 ). Solving the resulting equations of motion and suitably choosing constants so as not to change the asymptotically measured mass and charge, we find δ m ( x ) = cξ 2 e − 2 ξx  e ξx µ − Q 2  m ( x ) " (2 Q 2 − 3 µ 2 ) e − 2 ξx + 2 e − ξx µ − 1 m ( x ) 2 + e ξx − µ p µ 2 − Q 2 coth − 1 e ξx − µ p µ 2 − Q 2 # , δ n ( x ) = 2 cξ 2 e − ξx  µ 2 − Q 2  m ( x ) 3 " 9 e − 2 ξx  e ξx − µ  2 m ( x ) 2 − 4 e − ξx − 1 2 p µ 2 − Q 2 coth − 1 e ξx − µ p µ 2 − Q 2 # , δ V ( x ) = √ 2 cξ 2 Qe − ξx " (2 Q 2 − 3 µ 2 ) e − 2 ξx + 2 e − ξx µ − 1 m ( x ) 2 + e ξx − µ p µ 2 − Q 2 coth − 1 e ξx − µ p µ 2 − Q 2 # . (25) The scalar inv ariants up to O ( ϵ ) are given by R = 2 e − 2 ξx ξ 2  e ξx µ − 2 Q 2  − 2 ϵcξ 4 e − ξx p µ 2 − Q 2 coth − 1 e ξx − µ p µ 2 − Q 2 + 2 ϵcξ 4 e − 2 ξx  µ 2 − Q 2  1 + 21 e − ξx µ − e − 2 ξx (41 Q 2 + 36 µ 2 ) + 87 e − 3 ξx µQ 2 − 32 e − 4 ξx Q 4 m ( x ) 4 , F µν F µν = − 4 ξ 2 Q 2 e − 2 ξx − ϵcξ 4 e − 4 ξx 64 Q 2  µ 2 − Q 2  m ( x ) 2 . (26) Since the red-shift factor m ( x ) 2 , with the unp erturb ed horizon, is present in the denominator in the scalar inv ariants, w e conclude that p erturbation theory breaks down in the vicinit y of the horizon. W e next turn to the non-p erturbativ e solution parametrized as ( 9 ). In order to determine the mass and c harge at 7 asymptotic infinit y , we expand v arious quantities ab out f = 0 , which yields d x d f = − 1 ξ f + f 4 ξ 3 − 3 − 4 ξ 2 β ′′ (0) 32 ξ 5 f 3 + · · · , m ( f ) 2 = csc h 2 γ Q 2  1 − √ 2 f ξ coth γ + 2 + cosh 2 γ 2 csc h 2 γ f 2 ξ 2 + · · ·  , 2 ϕ = log f ξ + Φ 0 + log csc h γ Q | {z } ≡ 2 ϕ 1 − f coth γ √ 2 ξ + f 2 csc h 2 γ 4 ξ 2 + · · · , (27) where γ is defined in eq. ( 15 ). In verting the last expansion, w e obtain the expansion of the metric and gauge field comp onen ts in the string dilaton co ordinate, m ( ϕ ) 2 = csc h 2 γ Q 2  1 − √ 2 coth γ e 2( ϕ − ϕ 1 ) + e 4( ϕ − ϕ 1 ) 2 csc h 2 γ + · · ·  , n ( ϕ ) 2 = 1 + √ 2 coth γ e 2( ϕ − ϕ 1 ) + 1 + 2 cosh 2 γ 2 csc h 2 γ e 4( ϕ − ϕ 1 ) + · · · , V ( ϕ ) = − csc h 2 γ Q e 2( ϕ − ϕ 1 ) + · · · . (28) In a similar vein, the determination of the extremalit y condition requires a knowledge of the metric close to the BH horizon ( f = ∞ ). As mentioned in the main text, we assume the ansatz β ( f 2 ) = P ∞ n =0 β n f − 2 n and expand v arious quan tities ab out f = ∞ , d x d f = − √ 2 f 2 + 2 ξ 2 + β 0 √ 2 f 4 − 3  2 ξ 2 + β 0  2 − 4 β 1 4 √ 2 f 6 + · · · , m ( f ) 2 = 4 ξ 2 e − 2( γ + χ ) Q 2 f 2 + 2 ξ 2 e − 4( γ + χ ) ∆ Q 2 f 4 + e − 6( γ + χ ) ξ 2  ∆(5∆ − 8 ξ 2 ) − 12 e 4( γ + χ ) β 1  4 Q 2 f 6 + · · · , 2 ϕ = Φ 0 + log 2 e − ( γ + χ ) Q | {z } ≡ 2 ϕ h + ∆ e − 2( γ + χ ) 4 f 2 − 12 β 1 − e − 4( γ + χ ) ∆(3∆ − 8 ξ 2 ) 32 f 4 + · · · , (29) where ∆ and χ are defined in ( 16 ) and ( 17 ) resp ectiv ely . Then, inv erting the last expression and using the definition of the metric ( 13 ), we find the near-horizon expansion of the metric comp onen ts to b e Q 2 m ( ϕ ) 2 = 16 ξ 2 ∆ ( ϕ − ϕ h ) + 8 ξ 2 ∆(∆ + 8 ξ 2 ) + 12 e 4( γ + χ ) β 1 ∆ 3 ( ϕ − ϕ h ) 2 + · · · , ξ 2 n ( ϕ ) 2 = ∆( ϕ − ϕ h ) 2 e 2( γ + χ ) + ∆(∆ + 8 ξ 2 ) − 36 e 4( γ + χ ) β 1 4∆ e 2( γ + χ ) ( ϕ − ϕ h ) 2 + · · · . (30) By making a suitable rescaling of the temp oral co ordinate, we can bring m ( ϕ ) 2 to the more familiar form in whic h m ( ϕ ) 2 ∼ ∆( ϕ − ϕ h ) . W e can thus find the extremality condition to b e ∆ = 0 . Note that as ∆ → 0 , the expansions b ecome ill-defined. How ev er, in case the large- f corrections to β ( f 2 ) are exponentially small, the limiting b eha vior as ∆ → 0 is smo oth.