D3-Branes and Two Low-Energy Limits

The main idea

The canonical AdS/CFT duality begins with one physical system in type IIB string theory: a stack of $N$ coincident D3-branes. The same system has two complementary low-energy descriptions.

From the open-string viewpoint, D3-branes are hypersurfaces where open strings can end. At energies much smaller than the string scale, the massive open-string oscillator modes decouple, and the massless open-string modes on the branes become four-dimensional $\mathcal N=4$ super-Yang-Mills theory.

From the closed-string viewpoint, the same D3-branes are heavy Ramond-Ramond charged objects. They curve spacetime and source a self-dual five-form flux $F_5$. At low energies, the near-horizon region of the corresponding black D3-brane geometry decouples from the asymptotically flat exterior and becomes type IIB string theory on

$$\mathrm{AdS}_5\times S^5.$$

Both viewpoints also contain the same decoupled free closed-string sector in asymptotically flat ten-dimensional space. Removing that common spectator sector gives the canonical correspondence:

$$\boxed{ \mathcal N=4\; SU(N)\;\text{super-Yang-Mills in four dimensions} \quad\Longleftrightarrow\quad \text{type IIB string theory on }\mathrm{AdS}_5\times S^5 }$$

with $N$ units of five-form flux through the $S^5$.

The same stack of $N$ coincident D3-branes has two low-energy descriptions. The open-string description gives $\mathcal N=4$ super-Yang-Mills theory plus a decoupled free closed-string sector. The closed-string geometry description gives type IIB string theory on the near-horizon throat $\mathrm{AdS}_5\times S^5$ plus the same decoupled free sector. The interacting sectors are identified.

This page explains the decoupling argument. The next pages describe the two sides in detail: first $\mathcal N=4$ SYM, then type IIB on $\mathrm{AdS}_5\times S^5$, then the parameter map and first tests.

The object: $N$ coincident D3-branes

A D3-brane is a dynamical $3+1$ dimensional object in type IIB string theory. Let its worldvolume coordinates be

$$x^\mu=(x^0,x^1,x^2,x^3),$$

and let the six transverse coordinates be $y^i$, $i=1,\ldots,6$. The radial distance away from the branes is

$$r^2=\sum_{i=1}^6 (y^i)^2.$$

Open strings ending on a D3-brane obey Neumann boundary conditions along $x^\mu$ and Dirichlet boundary conditions along $y^i$. Their massless modes produce a four-dimensional gauge multiplet:

Worldvolume field	Brane interpretation	CFT interpretation
$A_\mu$	gauge field on the brane	spin-1 gauge field
$X^i$	transverse fluctuations of the brane	six adjoint scalars
fermions	supersymmetric partners	four adjoint Weyl fermions

The six scalars are not arbitrary matter fields. They are the collective coordinates for moving the brane in the transverse $\mathbb R^6$.

For $N$ coincident D3-branes, open strings carry Chan-Paton labels $a,b=1,\ldots,N$ at their endpoints. A string can begin on brane $a$ and end on brane $b$, so the light fields are $N\times N$ matrices:

$$A_\mu=A_\mu{}^a{}_b, \qquad X^i=X^i{}^a{}_b.$$

The gauge symmetry is enhanced to $U(N)$. Separating the branes makes the scalar matrices approximately diagonal and breaks

$$U(N)\longrightarrow U(1)^N.$$

The off-diagonal fields then come from strings stretched between different branes. If two branes are separated by a distance $|y_a-y_b|$, the stretched string has mass

$$M_{ab}\sim \frac{|y_a-y_b|}{2\pi\alpha'}.$$

Thus nonabelian gauge symmetry appears because strings can begin and end on different branes. When the branes coincide, these stretched strings become massless.

The overall $U(1)$ factor describes the center-of-mass motion of the whole brane stack. It is a free multiplet. The interacting part of the theory is the $SU(N)$ sector. In large-$N$ discussions one often says $U(N)$ or $SU(N)$ somewhat interchangeably, but the precise interacting holographic CFT is the nonabelian sector after removing the decoupled center-of-mass multiplet.

First low-energy limit: the open-string description

The open-string spectrum has a tower of massive oscillator modes with masses of order

$$M_s=\frac{1}{\sqrt{\alpha'}}.$$

To isolate the low-energy worldvolume theory, take

$$\alpha'\to0$$

while keeping the energies of interest fixed. The massive open-string states become infinitely heavy and decouple. The massless open-string fields remain and organize into the $\mathcal N=4$ vector multiplet.

The same limit also simplifies the ambient closed-string sector. The ten-dimensional Newton constant scales as

$$G_{10}\sim g_s^2\alpha'^4.$$

At fixed $g_s$, gravitational interactions between finite-energy brane excitations and asymptotically flat bulk gravitons vanish as $\alpha'\to0$. The open-string low-energy description therefore splits into two decoupled sectors:

$$\boxed{ \text{open-string low-energy limit} = \mathcal N=4\;U(N)\;\mathrm{SYM} \;\oplus\; \text{free type IIB closed strings in flat space} }$$

The free flat-space closed-string sector is a spectator. It is not the gravitational dual of the gauge theory. The dual geometry will come from a different part of the same brane system: the near-horizon throat.

Gauge coupling from the D3-brane action

The bosonic D3-brane action contains the Dirac-Born-Infeld term

$$S_{\mathrm{DBI}} = -T_3\int d^4x\,e^{-\Phi}\, \mathrm{Tr}\sqrt{-\det\left(\eta_{\mu\nu}+2\pi\alpha' F_{\mu\nu}+\cdots\right)}.$$

Expanding the determinant for weak fields gives the Yang-Mills kinetic term,

$$S_{\mathrm{DBI}} \supset - \frac{T_3e^{-\Phi}(2\pi\alpha')^2}{4} \int d^4x\,\mathrm{Tr}\,F_{\mu\nu}F^{\mu\nu}.$$

Using

$$T_3=\frac{1}{(2\pi)^3\alpha'^2}, \qquad e^\Phi=g_s,$$

one obtains a dimensionless four-dimensional Yang-Mills coupling proportional to $g_s$. In the convention used throughout the canonical AdS/CFT dictionary,

$$\boxed{g_{\mathrm{YM}}^2=4\pi g_s.}$$

Some references instead write $g_{\mathrm{YM}}^2=2\pi g_s$. This is usually a trace-normalization convention. The convention-independent physics is captured by the relation

$$\lambda=g_{\mathrm{YM}}^2N=4\pi g_sN,$$

and by the radius formula

$$\boxed{\frac{L^4}{\alpha'^2}=4\pi g_sN=\lambda.}$$

This formula will become the most important practical bridge between the gauge theory and the geometry.

Second low-energy limit: the closed-string geometry

D3-branes are not merely boundary conditions for open strings. They carry Ramond-Ramond charge and tension, so they source the closed-string fields of type IIB supergravity.

The extremal solution for $N$ coincident D3-branes is

$$ds_{10}^2 = H(r)^{-1/2}\eta_{\mu\nu}dx^\mu dx^\nu + H(r)^{1/2}\left(dr^2+r^2d\Omega_5^2\right),$$

where

$$H(r)=1+\frac{L^4}{r^4}, \qquad L^4=4\pi g_sN\alpha'^2.$$

The dilaton is constant,

$$e^\Phi=g_s,$$

and the background carries $N$ units of self-dual five-form flux through the transverse sphere:

$$\frac{1}{(2\pi)^4\alpha'^2} \int_{S^5}F_5=N,$$

up to conventional normalization of $F_5$. The integer $N$ is therefore the number of colors in the gauge theory and the number of flux units in the bulk.

This geometry has two regions. At large radius,

$$r\gg L, \qquad H(r)\simeq1,$$

so the spacetime is approximately ten-dimensional flat space. Near the branes,

$$r\ll L, \qquad H(r)\simeq\frac{L^4}{r^4},$$

and the metric becomes

$$ds_{10}^2 \simeq \frac{r^2}{L^2}\eta_{\mu\nu}dx^\mu dx^\nu + \frac{L^2}{r^2}dr^2 + L^2d\Omega_5^2.$$

Introducing the Poincaré coordinate

$$z=\frac{L^2}{r},$$

the first two terms become

$$\frac{L^2}{z^2}\left(dz^2+\eta_{\mu\nu}dx^\mu dx^\nu\right).$$

Therefore the near-horizon metric is

$$\boxed{ ds_{10}^2 = L^2\left[ \frac{dz^2+\eta_{\mu\nu}dx^\mu dx^\nu}{z^2} +d\Omega_5^2 \right], }$$

which is $\mathrm{AdS}_5\times S^5$ with common radius $L$.

Why the throat decouples

The near-horizon region is not selected merely because it looks mathematically pretty. It becomes a separate low-energy sector because of gravitational redshift.

For a static excitation at radius $r$, the energy measured at infinity is related to the local proper energy by

$$E_\infty=\sqrt{-g_{tt}(r)}\,E_{\mathrm{local}} =H(r)^{-1/4}E_{\mathrm{local}}.$$

Near the horizon,

$$H(r)^{-1/4}\simeq\frac{r}{L},$$

so

$$E_\infty\simeq\frac{r}{L}E_{\mathrm{local}}.$$

A finite local excitation deep in the throat has very small energy as seen from infinity. Low-energy observers therefore see excitations localized in the throat even when those excitations are not low-energy in local string units.

A useful scaling variable is

$$U=\frac{r}{\alpha'}.$$

The D3-brane decoupling limit may be described as

$$\alpha'\to0, \qquad U=\frac{r}{\alpha'}\;\text{fixed}, \qquad g_s\;\text{fixed}, \qquad N\;\text{fixed}.$$

In this limit the asymptotically flat exterior and the throat stop interacting. The closed-string low-energy description also splits into two decoupled sectors:

$$\boxed{ \text{closed-string low-energy limit} = \text{type IIB strings on }\mathrm{AdS}_5\times S^5 \;\oplus\; \text{free type IIB closed strings in flat space} }$$

This is the closed-string counterpart of the open-string splitting found above.

Equating the interacting sectors

The two descriptions came from the same underlying D3-brane system. They must therefore describe the same low-energy physics. We have

$$\begin{aligned} \text{open-string description:} &\qquad \mathcal N=4\;U(N)\;\mathrm{SYM} \oplus \text{free flat-space IIB}, \\ \text{closed-string description:} &\qquad \text{IIB strings on }\mathrm{AdS}_5\times S^5 \oplus \text{free flat-space IIB}. \end{aligned}$$

The free flat-space sector is common. Removing it leaves the interacting dual pair:

$$\boxed{ \mathcal N=4\;SU(N)\;\mathrm{SYM} \quad\longleftrightarrow\quad \text{type IIB string theory on }\mathrm{AdS}_5\times S^5 \text{ with }N\text{ units of }F_5\text{ flux}. }$$

This is the canonical $\mathrm{AdS}_5/\mathrm{CFT}_4$ correspondence.

The brane argument is not a formal proof in the mathematical sense. It is a dynamical derivation of why two apparently different theories should be two descriptions of the same decoupled sector of string theory. The full conjecture is stronger than the low-energy supergravity approximation: it asserts equality between the complete quantum CFT and the complete quantum type IIB string theory on $\mathrm{AdS}_5\times S^5$, including all $\alpha'$ corrections, string-loop corrections, D-brane states, and nonperturbative effects.

The first dictionary entries

The D3-brane construction already gives the first entries of the canonical dictionary.

Gauge-theory quantity	String/gravity quantity	Meaning
rank $N$	five-form flux $\int_{S^5}F_5$	number of D3-branes becomes flux
$g_{\mathrm{YM}}^2$	$4\pi g_s$	gauge coupling becomes string coupling
$\lambda=g_{\mathrm{YM}}^2N$	$L^4/\alpha'^2$	’t Hooft coupling controls curvature in string units
large $N$	small bulk loop expansion	suppresses quantum gravity/string loops
large $\lambda$	$L\gg \ell_s$	suppresses stringy curvature corrections
$SO(6)_R$	isometry of $S^5$	R-symmetry becomes internal geometry
conformal group $SO(2,4)$	isometry of $\mathrm{AdS}_5$	spacetime conformal symmetry becomes bulk isometry
scalar vevs	brane positions in $\mathbb R^6$	Coulomb branch geometry
stretched strings	W-bosons	off-diagonal fields after Higgsing

The symmetry match is especially striking:

$$\mathrm{Isom}(\mathrm{AdS}_5)=SO(2,4), \qquad \mathrm{Isom}(S^5)=SO(6)\simeq SU(4)_R.$$

Together with supersymmetry, the full symmetry is the superconformal group

$$PSU(2,2|4).$$

Symmetry matching alone would not prove the duality, but the D3-brane construction explains why this particular symmetry match occurs.

Parameters and approximations

The exact duality is a statement about two complete quantum theories. Practical calculations usually require approximations. The D3-brane dictionary tells us which approximation on one side corresponds to which approximation on the other.

From

$$\frac{L^4}{\alpha'^2}=\lambda,$$

we get

$$\frac{L^2}{\alpha'}=\sqrt{\lambda}.$$

Thus the curvature radius is large in string units when

$$\lambda\gg1.$$

This suppresses stringy $\alpha'$ corrections.

The string coupling is

$$g_s=\frac{\lambda}{4\pi N}.$$

Thus closed-string loops are suppressed when $g_s\ll1$. In the five-dimensional effective gravity description, the same statement is often written as

$$\frac{L^3}{G_5}\sim N^2,$$

so bulk quantum loops are suppressed by powers of $1/N^2$.

The classical type IIB supergravity regime is therefore

$$N\gg1, \qquad \lambda\gg1, \qquad g_s=\frac{\lambda}{4\pi N}\ll1.$$

The slogan “strongly coupled gauge theory equals classical gravity” refers only to this corner. The actual duality is broader:

$$\begin{array}{ccl} \text{exact CFT} &\leftrightarrow& \text{full string theory}, \\ N\to\infty &\leftrightarrow& \text{classical string genus expansion}, \\ \lambda\to\infty &\leftrightarrow& \text{small curvature in string units}, \\ N\gg1,\;\lambda\gg1 &\leftrightarrow& \text{classical supergravity}. \end{array}$$

At large $N$ but small $\lambda$, the CFT may be perturbative, but the bulk is highly stringy. At finite $N$, the bulk is quantum gravitational. At finite $\lambda$, massive string modes cannot be ignored.

Including the theta angle

The canonical dictionary also identifies the complexified gauge coupling with the type IIB axio-dilaton. The gauge-theory coupling is

$$\tau_{\mathrm{YM}} = \frac{\theta}{2\pi} + \frac{4\pi i}{g_{\mathrm{YM}}^2},$$

while the type IIB axio-dilaton is

$$\tau_{\mathrm{IIB}}=C_0+i e^{-\Phi}.$$

With the normalization used above,

$$\boxed{\tau_{\mathrm{YM}}=\tau_{\mathrm{IIB}}.}$$

For most of this course we set $\theta=0$ and work with constant real $g_s$. But this identification is conceptually important: the $SL(2,\mathbb Z)$ duality of type IIB string theory is reflected in the Montonen-Olive duality of $\mathcal N=4$ SYM.

Why D3-branes are special

D$p$-branes exist for many $p$, and their worldvolume theories are maximally supersymmetric Yang-Mills theories in $p+1$ dimensions. The D3-brane case is special because the Yang-Mills coupling is dimensionless.

For D$p$-branes,

$$g_{\mathrm{YM},p+1}^2 \sim g_s(\alpha')^{(p-3)/2},$$

so in mass units

$$[g_{\mathrm{YM},p+1}^2]=3-p.$$

Only for $p=3$ is the coupling dimensionless. Correspondingly, only for D3-branes does the near-horizon geometry become an ordinary AdS spacetime times a compact sphere with constant dilaton:

$$\text{D3-branes}\quad\longrightarrow\quad\mathrm{AdS}_5\times S^5.$$

For $p\neq3$, the near-horizon limits are still extremely useful, but they describe generalized holographic duals of nonconformal gauge theories rather than ordinary AdS/CFT in the strict conformal sense.

What decouples, and what does not

The decoupling limit removes several sectors from the interacting dual pair:

• massive open-string oscillator modes, because $M_s\sim1/\sqrt{\alpha'}$;
• interactions with asymptotically flat closed strings, because $G_{10}\sim g_s^2\alpha'^4$;
• the center-of-mass $U(1)$ multiplet of the D3-brane stack;
• the asymptotically flat exterior, which becomes a separate free closed-string sector.

But the limit does not remove the full string theory in the throat. At finite $\lambda$, massive string modes in $\mathrm{AdS}_5\times S^5$ are part of the dual. At finite $N$, string loops and quantum gravity effects are part of the dual. Nonperturbative objects such as D-branes wrapping cycles are also part of the full theory when their dual CFT states are included.

This is why the exact statement is not

$$\text{classical gravity}=\text{strongly coupled gauge theory}.$$

The exact statement is

$$\text{full type IIB string theory on }\mathrm{AdS}_5\times S^5 = \mathcal N=4\;SU(N)\;\mathrm{SYM}.$$

Classical gravity is the simplest calculational corner.

Common mistakes

Mistake 1: “The D3-branes sit at the AdS boundary”

In the original asymptotically flat D3-brane geometry, the branes are at $r=0$. In the near-horizon coordinate $z=L^2/r$, this is the Poincaré horizon $z\to\infty$, not the AdS boundary. The AdS boundary corresponds to $z\to0$, or $r\to\infty$ within the throat region.

Mistake 2: “The near-horizon limit is just setting $r=0$”

The near-horizon limit is a scaling limit, not evaluation at a point. One zooms into the throat while keeping the energy variable $U=r/\alpha'$ fixed.

Mistake 3: “Large $N$ automatically gives Einstein gravity”

Large $N$ suppresses bulk loops, but it does not suppress stringy curvature corrections. For classical two-derivative gravity one also needs $\lambda\gg1$.

Mistake 4: “The $S^5$ is optional decoration”

The $S^5$ carries the five-form flux, controls the Kaluza-Klein spectrum, and geometrizes the $SO(6)_R$ symmetry of $\mathcal N=4$ SYM. It is part of the duality, not an add-on.

Mistake 5: “The brane argument proves only supergravity”

The brane argument motivates the full string duality. Supergravity is a later approximation obtained when $N$ and $\lambda$ are large in the appropriate sense.

Exercises

Exercise 1: Derive the near-horizon metric

Start from

$$ds^2 = H(r)^{-1/2}dx_{1,3}^2 + H(r)^{1/2}\left(dr^2+r^2d\Omega_5^2\right), \qquad H(r)=1+\frac{L^4}{r^4}.$$

Show that for $r\ll L$ this becomes $\mathrm{AdS}_5\times S^5$ with common radius $L$.

Solution

For $r\ll L$,

$$H(r)\simeq\frac{L^4}{r^4}.$$

Hence

$$H(r)^{-1/2}\simeq\frac{r^2}{L^2}, \qquad H(r)^{1/2}\simeq\frac{L^2}{r^2}.$$

Substituting gives

$$ds^2 \simeq \frac{r^2}{L^2}dx_{1,3}^2 + \frac{L^2}{r^2}dr^2 +L^2d\Omega_5^2.$$

With $z=L^2/r$,

$$\frac{r^2}{L^2}dx_{1,3}^2 + \frac{L^2}{r^2}dr^2 = \frac{L^2}{z^2}\left(dx_{1,3}^2+dz^2\right).$$

Thus

$$ds^2 = \frac{L^2}{z^2}\left(dx_{1,3}^2+dz^2\right) +L^2d\Omega_5^2,$$

which is $\mathrm{AdS}_5\times S^5$.

Exercise 2: Redshift and the survival of throat excitations

For a static excitation in the D3-brane geometry, show that

$$E_\infty=H(r)^{-1/4}E_{\mathrm{local}}.$$

Then show that near the horizon

$$E_\infty\simeq\frac{r}{L}E_{\mathrm{local}}.$$

Solution

For a static metric, energy redshifts as

$$E_\infty=\sqrt{-g_{tt}}\,E_{\mathrm{local}}.$$

The D3-brane metric has

$$g_{tt}=-H(r)^{-1/2},$$

so

$$\sqrt{-g_{tt}}=H(r)^{-1/4}.$$

Near the horizon,

$$H(r)\simeq\frac{L^4}{r^4},$$

and therefore

$$H(r)^{-1/4}\simeq\frac{r}{L}.$$

A finite local excitation near $r=0$ has arbitrarily small energy as measured at infinity. This is why throat excitations remain part of the low-energy physics.

Exercise 3: The string-scale curvature criterion

Using

$$\frac{L^4}{\alpha'^2}=\lambda,$$

show that stringy curvature corrections are suppressed when $\lambda\gg1$.

Solution

Stringy curvature corrections are suppressed when the curvature radius is large compared with the string length,

$$L\gg \ell_s, \qquad \ell_s=\sqrt{\alpha'}.$$

Equivalently,

$$\frac{L^2}{\alpha'}\gg1.$$

But

$$\frac{L^2}{\alpha'}=\sqrt{\lambda}.$$

Thus the condition is

$$\sqrt{\lambda}\gg1,$$

or simply

$$\lambda\gg1.$$

Exercise 4: The bulk loop expansion

Using

$$g_s=\frac{\lambda}{4\pi N},$$

explain why large $N$ suppresses closed-string loops at fixed $\lambda$.

Solution

Closed-string perturbation theory is an expansion in powers of $g_s$. Since

$$g_s=\frac{\lambda}{4\pi N},$$

taking $N\to\infty$ at fixed $\lambda$ gives $g_s\to0$. Hence closed-string loops are suppressed. In the gauge theory this is the planar large-$N$ expansion, where nonplanar corrections are suppressed by powers of $1/N^2$.

Exercise 5: Why D3-branes give a CFT

For a D$p$-brane,

$$g_{\mathrm{YM},p+1}^2\sim g_s(\alpha')^{(p-3)/2}.$$

Show that $p=3$ is the case in which the Yang-Mills coupling is dimensionless, and explain why this matters for AdS/CFT.

Solution

Since $\alpha'$ has dimensions of length squared,

$$(\alpha')^{(p-3)/2}$$

has dimensions of length$^{p-3}$, or mass$^{3-p}$. Thus

$$[g_{\mathrm{YM},p+1}^2]=\mathrm{mass}^{3-p}.$$

Only for $p=3$ is this dimensionless. A dimensionful coupling introduces a scale, whereas a dimensionless coupling is compatible with conformal invariance. This is why D3-branes lead to the clean conformal duality $\mathrm{AdS}_5/\mathrm{CFT}_4$.

Further reading

• J. M. Maldacena, “The Large $N$ Limit of Superconformal Field Theories and Supergravity”. The original brane decoupling argument and statement of the AdS/CFT correspondence.
• J. Polchinski, “Dirichlet-Branes and Ramond-Ramond Charges”. The foundational paper identifying D-branes as dynamical Ramond-Ramond charged objects.
• O. Aharony, S. S. Gubser, J. Maldacena, H. Ooguri, and Y. Oz, “Large $N$ Field Theories, String Theory and Gravity”. The classic review; see especially the discussion of D3-branes and the canonical example.
• N. Itzhaki, J. M. Maldacena, J. Sonnenschein, and S. Yankielowicz, “Supergravity and the Large $N$ Limit of Theories With Sixteen Supercharges”. The general D$p$-brane decoupling-limit perspective.
• I. R. Klebanov, “TASI Lectures: Introduction to the AdS/CFT Correspondence”. A useful pedagogical derivation of the canonical example and related brane constructions.
• C. V. Johnson, D-Branes. A detailed textbook treatment of D-brane dynamics and the open/closed string viewpoint.