Skip to content

Chiral Lagrangians and Low-Energy QCD

At momenta and pion masses small compared with the chiral scale, QCD is represented by the most general local action of its Goldstone field consistent with chiral symmetry, discrete symmetries, and a power counting. At leading order this construction fixes the pion kinetic and mass terms and the complete tree-level ππ\pi\pi amplitude; loops and higher-order low-energy constants then improve the prediction while canceling each other’s renormalization-scale dependence.

Required background. Chiral Order Parameters, Current Algebra, and Pions fixes the breaking pattern and decay-constant convention; Chiral Effective Theory and Nonlinear Symmetry supplies effective-field-theory matching and truncation logic.

Helpful background. Cosets and Nonlinear Realizations supplies the general coset construction.

For two light flavors, encode the three Goldstone fields in

U(x)=exp ⁣(iτaπa(x)F)SU(2),ULUR.U(x)=\exp\!\left(\frac{i\tau^a\pi^a(x)}{F}\right)\in SU(2), \qquad U\mapsto LUR^\dagger.

The vacuum is U=1U=\mathbf1 after aligning the mass matrix. A different coordinate choice on the same coset changes off-shell vertices but not on-shell amplitudes.

To derive currents and impose local Ward identities, couple QCD to left, right, scalar, and pseudoscalar sources lμ,rμ,s,pl_\mu,r_\mu,s,p. Their covariant combinations are

DμU=μUilμU+iUrμ,χ=2B(s+ip),D_\mu U=\partial_\mu U-il_\mu U+iUr_\mu, \qquad \chi=2B(s+ip),

with transformations

DμUL(DμU)R,χLχR.D_\mu U\mapsto L(D_\mu U)R^\dagger, \qquad \chi\mapsto L\chi R^\dagger.

The physical limit is lμ=rμ=p=0l_\mu=r_\mu=p=0 and s=Ms=M. Functional derivatives with respect to these sources generate the corresponding QCD currents and densities, so every effective operator automatically obeys the symmetry’s Ward identities.

Count a derivative or external momentum as O(p)O(p) and the quark mass, hence χ\chi, as O(p2)O(p^2). The unique even-parity two-derivative action without external field strengths is

L2=F24tr(DμUDμU)+F24tr(χU+Uχ).\mathcal L_2= \frac{F^2}{4}\operatorname{tr}(D_\mu U D^\mu U^\dagger) +\frac{F^2}{4}\operatorname{tr}(\chi U^\dagger+U\chi^\dagger).

FF and BB are low-energy constants in the two-flavor chiral limit. Matching the scalar source gives Σ=F2B\Sigma=F^2B at leading order in the convention of the preceding page.

For a connected mesonic graph with LL loops and vertices from operators of chiral dimension did_i, dimensional counting gives

ν=2+2L+i(di2),\nu=2+2L+\sum_i(d_i-2),

so each loop raises the order by two powers of pp. The expansion parameters are schematically p2/(4πF)2p^2/(4\pi F)^2 and mπ2/(4πF)2m_\pi^2/(4\pi F)^2. The theory ceases to be predictive at a stated truncation when these are not small or when omitted resonances become dynamical.

The diagram summarizes how that expansion inherits QCD information. The upper chain is fixed by symmetry and vacuum realization; external sources and matched low-energy constants turn it into amplitudes, while the singlet-anomaly and theta branches require additional topological information.

QCD chiral currents and vacuum breaking generate Goldstone pion fields and chiral EFT, whose external sources and low-energy constants produce amplitudes, alongside separate anomaly, topology, and theta branches.

Low-energy QCD map. Symmetry realization fixes the pion field space and operator structure, but low-energy constants must be matched and predictions require p,mπΛχp,m_\pi\ll\Lambda_\chi with an explicit truncation order. The U(1)AU(1)_A, topology, and theta sectors are related but not identical. The diagram is schematic.

Set M=m^1M=\widehat m\mathbf1 and mπ2=2Bm^=B(mu+md)m_\pi^2=2B\widehat m=B(m_u+m_d) at leading order. The exponential field expands as

U=1+iτaπaFπaπa2F2+O(π3).U=\mathbf1+\frac{i\tau^a\pi^a}{F} -\frac{\pi^a\pi^a}{2F^2}+O(\pi^3).

Using tr(τaτb)=2δab\operatorname{tr}(\tau^a\tau^b)=2\delta^{ab} gives

L2=12μπaμπa12mπ2πaπa+L2(4π)+O(π6),\mathcal L_2= \frac12\partial_\mu\pi^a\partial^\mu\pi^a -\frac12m_\pi^2\pi^a\pi^a +\mathcal L_2^{(4\pi)}+O(\pi^6),

where, in this exponential parametrization,

L2(4π)=16F2[(πaμπa)2(πaπa)(μπbμπb)]+mπ224F2(πaπa)2.\mathcal L_2^{(4\pi)}= \frac{1}{6F^2} \left[(\pi^a\partial_\mu\pi^a)^2 -(\pi^a\pi^a)(\partial_\mu\pi^b\partial^\mu\pi^b)\right] +\frac{m_\pi^2}{24F^2}(\pi^a\pi^a)^2.

The quadratic terms verify canonical normalization and the leading mass relation. Together with Σ=F2B\Sigma=F^2B, the latter reproduces

F2mπ2=(mu+md)ΣF^2m_\pi^2=(m_u+m_d)\Sigma

at leading order. The distinction between FF in the chiral limit and the measured FπF_\pi begins at higher chiral order.

Let πa(p1)πb(p2)πc(p3)πd(p4)\pi^a(p_1)\pi^b(p_2)\to\pi^c(p_3)\pi^d(p_4), with all external pions on shell and

s=(p1+p2)2,t=(p1p3)2,u=(p1p4)2,s+t+u=4mπ2.s=(p_1+p_2)^2, \qquad t=(p_1-p_3)^2, \qquad u=(p_1-p_4)^2, \qquad s+t+u=4m_\pi^2.

Inserting the four-pion interactions and using the on-shell relation reduces the tree amplitude to

Mab;cd(s,t,u)=δabδcdA(s,t,u)+δacδbdA(t,s,u)+δadδbcA(u,t,s),\begin{aligned} \mathcal M^{ab;cd}(s,t,u) ={}&\delta^{ab}\delta^{cd}A(s,t,u) +\delta^{ac}\delta^{bd}A(t,s,u)\\ &+\delta^{ad}\delta^{bc}A(u,t,s), \end{aligned}

with

A(s,t,u)=smπ2Fπ2+O(p4).A(s,t,u)=\frac{s-m_\pi^2}{F_\pi^2}+O(p^4).

Replacing FF by FπF_\pi changes only the indicated higher-order remainder. The result is Weinberg’s low-energy theorem Weinberg 1966, pp. 616–619.

Three checks expose most normalization or crossing errors:

  1. Crossing: exchanging an incoming and outgoing pion permutes (s,t,u)(s,t,u) and the corresponding Kronecker tensors; the displayed decomposition is closed under every such permutation.
  2. Adler zero: at the analytically continued point s=t=u=mπ2s=t=u=m_\pi^2, every scalar function AA vanishes. This point is explicitly unphysical for on-shell equal-mass 222\to2 scattering because it violates s+t+u=4mπ2s+t+u=4m_\pi^2; it is a soft-current constraint on the analytic amplitude.
  3. Isospin projection: using the on-shell sum gives
T0=2smπ2Fπ2,T1=tuFπ2,T2=2mπ2sFπ2+O(p4),T^0=\frac{2s-m_\pi^2}{F_\pi^2}, \qquad T^1=\frac{t-u}{F_\pi^2}, \qquad T^2=\frac{2m_\pi^2-s}{F_\pi^2} +O(p^4),

which has the required tut\leftrightarrow u parity in each channel.

A reproducible calculation implements the same quadratic expansion, four-pion terms, tensor decomposition, crossing permutations, and Adler-point test. A mismatch with these equations is a convention or algebra error, not a new prediction.

Loops, low-energy constants, and remainders

Section titled “Loops, low-energy constants, and remainders”

At O(p4)O(p^4), one-loop graphs built from L2\mathcal L_2 generate nonanalytic logarithms and ultraviolet poles. The most general L4=iiOi\mathcal L_4=\sum_i\ell_i O_i contains local counterterms with renormalized coefficients ir(μ)\ell_i^r(\mu). In a conventional subtraction scheme,

ir(μ2)=ir(μ1)+γi16π2lnμ1μ2,\ell_i^r(\mu_2)=\ell_i^r(\mu_1) +\frac{\gamma_i}{16\pi^2}\ln\frac{\mu_1}{\mu_2},

with coefficients γi\gamma_i fixed by the divergence basis. For any observable amplitude,

μddμ(Aloop(4)+ALEC(4))=0,\mu\frac{d}{d\mu} \left(\mathcal A_{\rm loop}^{(4)} +\mathcal A_{\rm LEC}^{(4)}\right)=0,

up to O(p6)O(p^6). A loop logarithm by itself is not a prediction; its scale dependence cancels only after the appropriate low-energy constants are included. The systematic one-loop construction is given by Gasser and Leutwyler 1984, §§3–7.

A complete result therefore states:

  • the operator basis and subtraction convention;
  • the scale μ\mu at which renormalized constants are quoted;
  • whether F,mπF,m_\pi or physical Fπ,mπF_\pi,m_\pi parameterize lower-order terms;
  • the chiral order retained and an O(pν+2)O(p^{\nu+2}) remainder;
  • the kinematic domain below inelastic thresholds and omitted resonances.

Mixing decay-constant conventions. If the axial generator is rescaled, FπF_\pi rescales inversely. Keep Ta=τa/2T^a=\tau^a/2 from the current definition through the scattering amplitude.

Calling the Adler point physical. For massive on-shell pions, s=t=u=mπ2s=t=u=m_\pi^2 conflicts with s+t+u=4mπ2s+t+u=4m_\pi^2. It is an analytically continued soft point used to check the symmetry structure.

Keeping loops but dropping counterterms. The result then depends on the subtraction scale and is not an observable. Include every local operator required at the same chiral order.

Expand the leading mass term for M=m^1M=\widehat m\mathbf1 through fourth order in πa\pi^a and verify its contribution to L2(4π)\mathcal L_2^{(4\pi)}.

Solution

For SU(2)SU(2),

U+U=2cos ⁣(πF)1,χ=mπ21.U+U^\dagger=2\cos\!\left(\frac{|\boldsymbol\pi|}{F}\right)\mathbf1, \qquad \chi=m_\pi^2\mathbf1.

Thus

F24tr[χ(U+U)]=F2mπ212mπ2π2+mπ224F2(π2)2+O(π6),\frac{F^2}{4}\operatorname{tr}[\chi(U+U^\dagger)] =F^2m_\pi^2 -\frac12m_\pi^2\pi^2 +\frac{m_\pi^2}{24F^2}(\pi^2)^2+O(\pi^6),

which gives the displayed mass and four-pion interaction after the irrelevant vacuum constant is dropped.

  • Burgess, C. P. Introduction to Effective Field Theory: Thinking Effectively about Hierarchies of Scale. Cambridge: Cambridge University Press, 2020, §§8.1–8.2. DOI.
  • Gasser, Jürg, and Heinrich Leutwyler. “Chiral Perturbation Theory to One Loop.” Annals of Physics 158 (1984): 142–210. DOI.
  • Weinberg, Steven. “Pion Scattering Lengths.” Physical Review Letters 17 (1966): 616–621. DOI.
  • Weinberg, Steven. The Quantum Theory of Fields. Volume II: Modern Applications. Cambridge: Cambridge University Press, 1996, §§19.4–19.5. DOI.