---
title: 'Photo-Induced SHG: Mechanisms & Applications'
url: https://www.emergentmind.com/topics/photo-induced-second-harmonic-generation-shg
type: topic
---

# Photo-Induced SHG: Mechanisms & Applications

Photo-induced second-harmonic generation (SHG) denotes second-harmonic emission at \(2\omega\) that is activated, enhanced, or dynamically reconfigured by optical pumping or by optically addressable perturbations that alter inversion symmetry, the complex nonlinear susceptibility, or the phase relation between coherent nonlinear emitters. In the systems considered here, such perturbations include static electric fields, DC currents, optically written space-charge fields, exciton-mediated interlayer coupling, quantum-confined structural asymmetry, and defect-induced inversion-symmetry breaking; the resulting SHG is often governed as much by interference and mode symmetry as by the magnitude of \(\chi^{(2)}\) itself [2507.09306], [1203.4395], [2305.17512].

## 1. Nonlinear-optical basis and induced susceptibilities

SHG is the second-order nonlinear process in which two photons at frequency \(\omega\) combine to produce one photon at \(2\omega\). In standard notation, the second-order polarization is

\[
P^{(2)}(2\omega)=\varepsilon_0 \chi^{(2)} : E(\omega)E(\omega),
\]

and, in the electric-dipole approximation, bulk SHG is forbidden in centrosymmetric media because inversion symmetry enforces \(\chi^{(2)}=0\) [2507.09306]. This symmetry rule underlies most photo-induced SHG research: the central problem is not frequency doubling per se, but how to create, modulate, or reveal an effective second-order response in systems for which it is weak, hidden, or absent in equilibrium.

A canonical induced route is EFISH, in which a static electric field mixes the third-order susceptibility into an effective second-order response,

\[
\mathbf{P}_{2\omega} = \left( \chi^{(2)} + 3\,\chi^{(3)} : \mathbf{E}_{\mathrm{DC}} \right) : \mathbf{E}_\omega \otimes \mathbf{E}_\omega,
\]

so that a centrosymmetric medium with \(\chi^{(2)}=0\) acquires \(\chi^{(2)}_{\mathrm{eff}} \propto \chi^{(3)}E_{\mathrm{DC}}\) [2507.09306]. A distinct transport-driven route is current-induced SHG, for which the induced polarization is written

\[
P_{i}^{2\omega}=\chi_{ijkl}^{(2,1)}E_{j}^{\omega}E_{k}^{\omega}J_{l},
\]

making the SH polarization quadratic in the optical field and linear in current density \(J\) [1203.4395].

Within this broader landscape, recent work separates several physically distinct regimes.

| Route | Representative system | Characteristic signature |
|---|---|---|
| EFISH / optically written \(\chi^{(2)}\) | Si\(_3\)N\(_4\) waveguides and microresonators | DC-field-induced \(\chi^{(2)}_{\mathrm{eff}}\), self-organized quasi-phase matching |
| Current-induced SHG | Multilayer graphene on oxidized Si(001) | SHG linear in current amplitude, sign reversal with current reversal |
| Exciton- or interlayer-mediated SHG | Bilayer MoS\(_2\), TMD heterobilayers | Resonant enhancement, Stark tuning, phase shifts, charge-transfer sensitization |
| Structural / defect asymmetry | Ge-rich SiGe ACQWs, NV-diamond | Engineered or defect-driven inversion-symmetry breaking |

The same review literature also distinguishes EFISH from current-induced SHG and from the quantum-confined Stark effect (QCSE). EFISH is a real-space electrostatic asymmetry generated by \(E_{\mathrm{DC}}\); current-induced SHG arises from momentum-space asymmetry under carrier flow; QCSE primarily shifts resonant electronic structure rather than simply generating \(\chi^{(2)}_{\mathrm{eff}}\) through \(\chi^{(3)}E_{\mathrm{DC}}\) [2507.09306]. That distinction is essential because closely similar SHG phenomenology can emerge from different microscopic origins.

## 2. External-field and transport-driven activation of SHG

A clear transport-driven realization was reported for multilayer graphene carrying an in-plane DC current. The sample was a 4–5 monolayer multilayer graphene film grown by CVD on Ni foil and transferred onto an oxidized Si(001) substrate with a 300 nm SiO\(_2\) layer. Without current, the graphene contribution to SHG was negligible; with current, the SHG intensity rose sharply, the current-induced change rose linearly with current amplitude, and the sign reversed when the current direction was reversed [1203.4395]. The measured contrast was defined as

\[
\rho = \frac{2(I^{2\omega}(\vec J)-I^{2\omega}(0))}{I^{2\omega}(0)},
\]

and \(\rho(\vec J)\) was linear in \(J\) within error bars. Because the SHG baseline was dominated by the Si(001) surface, the observed graphene signal emerged through interference with the substrate SH field rather than through a large standalone graphene SH intensity.

The same work made the symmetry content explicit. If \(\phi\) is the angle between the current direction and the plane of incidence, then

\[
I_{pp,sp}\propto |a\cos\phi|^2,\qquad I_{ps,ss}\propto |b\sin\phi|^2,
\]

equivalently,

\[
\rho_{pp,sp}\propto \cos\phi,\qquad \rho_{ps,ss}\propto \sin\phi.
\]

Experimentally, the \(pp\) channel was larger when the current was parallel to the plane of incidence, and the polarization dependence inverted when the current was perpendicular, matching the tensor analysis [1203.4395]. Spectrally, the zero-current SHG resembled the Si(001) surface spectrum with a maximum near 3.34 eV, while current application produced direction-dependent spectral shifts interpreted as a wavelength-dependent phase relation between the silicon SH field and the current-induced graphene SH field.

External electric and magnetic fields activate analogous mechanisms in excitonic semiconductors. In ZnO, resonant SHG was observed at the \(1s(A,B)\), \(2s(A,B)\), \(2p(A,B)\), and \(1s(C)\) excitons in the 3.2–3.5 eV range, and strong SHG signals at these exciton resonances were induced by magnetic field when \(\mathbf{k}\parallel \mathbf{z}\), a geometry where the crystallographic SHG response vanishes [1308.6219]. The microscopic theory required treatment beyond the electric-dipole approximation and identified several field-induced mechanisms: Stark mixing, spin Zeeman mixing, orbital Zeeman mixing, and the magneto-Stark effect. For the \(2s/2p(A,B)\) manifold, the dominant mechanism was the magneto-Stark effect, with an effective field

\[
\mathbf{E}_{\mathrm{eff}}=\frac{\hbar}{M_{\mathrm{exc}}}[\mathbf{k}_{\mathrm{exc}}\times\mathbf{B}],
\]

which mixes opposite-parity exciton states and activates SHG [1308.6219].

Bilayer MoS\(_2\) provides a closely related but electrically tunable case. Although ideal 2H bilayer MoS\(_2\) is inversion symmetric and off-resonant SHG is weak, tuning \(2E_L\) into the A, B, and especially interlayer exciton resonances enhances the SHG amplitude by several orders of magnitude. A vertical electric field then further increases the response by Stark-splitting the two interlayer-exciton configurations with opposite permanent dipoles,

\[
E_{\mathrm{IE}_1} = E_{\mathrm{IE}} - \mu F_z,\qquad
E_{\mathrm{IE}_2} = E_{\mathrm{IE}} + \mu F_z.
\]

At \(F_z=0.17\) MV/cm, the interlayer-exciton SHG increased by about a factor of 25, with approximately quadratic field dependence [2104.01225]. In the model presented there, the interlayer-exciton-mediated susceptibility takes the form

\[
\chi_{\rm IE} = -2\mu F_z\frac{|T|^2}{\Delta^2}
\frac{C_2 |\Phi_{\rm B:1s}(0)|^2}{\left(2\hbar\omega - E_{\rm IE} + \frac{|T|^2}{\Delta}+ i \Gamma_{\rm IE}\right)^2},
\]

which makes the resonance enhancement and field dependence explicit [2104.01225].

## 3. Excitonic, quantum-confined, and defect-mediated routes

A major experimental theme is that photoexcited excitons can modify SHG in neighboring layers or in engineered low-dimensional potentials. In van der Waals TMD heterobilayers, exciton-sensitized SHG was demonstrated by resonantly exciting a donor layer and monitoring the SHG of an acceptor layer. The interpretation was a sequence of donor exciton absorption, ultrafast interlayer charge transfer, transient modification of the acceptor electronic state, and therefore a change in the acceptor’s nonlinear susceptibility \(\chi^{(2)}_{\rm sys}\). The enhancement factor was written

\[
EF = \frac{|\chi^{(2)}_{\rm sys}|^2}{|\chi^{(2)}_{\rm sys}{}^{0}|^2},
\]

and spectral phase interferometry showed that sensitization changed not only the SHG magnitude but also the SHG phase, with

\[
\Delta \phi = \phi_{\rm SYS-SEN} - \phi_{\rm SYS}.
\]

The effect was strongly short-ranged: inserting a 2L hBN spacer, corresponding to an average gap of \(\sim 1.1\) nm, made the enhancement nearly flat around unity and strongly suppressed the phase modulation, favoring charge transfer over Förster-type energy transfer as the operative interlayer mechanism [2305.17512].

Near-field SHG in TMDs shows how excitons can also rescue an otherwise weak nanoscale signal. In scattering-type near-field imaging of WSe\(_2\), the SHG intensity decayed exponentially with tip retraction with a \(1/e\) decay length of \(23 \pm 3\) nm, implying an average near-field enhancement factor of \(18 \pm 7\). Exciton resonance improved nano-SHG efficiency by about an order of magnitude and enabled imaging with about 20 nm spatial resolution. In bilayer WSe\(_2\), the nonlinear image directly distinguished bright 3R-stacked regions from dark 2H-stacked regions, and the average nano-SHG intensities were about 620 cps for the 3R bilayer and 160 cps for the monolayer, close to a 4:1 ratio [2111.06955]. This established SHG as a local symmetry probe at the nanometer scale.

Quantum-engineered group-IV heterostructures provide another induced route. In Ge-rich SiGe asymmetric coupled quantum wells (ACQWs) on Si(001), inversion symmetry was broken not at the lattice level but through an asymmetric coupled-well potential designed to produce three nearly equidistant confined states. The dominant tensor element was

\[
\chi_{zzz}^{(2)}(2\omega) = \frac{q^{3}(N_1-N_2)}{\varepsilon_0 \hbar^2}
\frac{z_{12}z_{23}z_{31}}{(\omega-\omega_{12}-i\gamma)(2\omega-\omega_{13}-i\gamma)},
\]

and the large response required both broken inversion symmetry and double resonance. Experimentally, the doubly resonant sample A gave \(K_2 = 5.7\times 10^{-3}\,\text{W}^{-1}\), compared with \(1.4\times 10^{-3}\,\text{W}^{-1}\) for sample B, \(0.6\times 10^{-3}\,\text{W}^{-1}\) for sample C, and \(<0.1\times 10^{-3}\,\text{W}^{-1}\) for the undoped reference D. The extracted experimental susceptibility was about \(5\times 10^4\) pm/V, the theoretical peak value was around \(3\times 10^4\) pm/V, and earlier modeling discussed in the paper could reach nearly \(10^5\) pm/V [2103.12620].

Defect-mediated inversion-symmetry breaking was demonstrated in bulk diamond containing a high density of NV centers. Under \(\sim 60\) fs excitation at 1350 nm and 100 kHz repetition rate, pure diamond showed only THG near 450 nm, whereas NV-doped diamond showed both SHG near 675 nm and THG near 450 nm. The SHG scaled as \(\sim I_{\rm ex}^{2}\) and THG as \(\sim I_{\rm ex}^{3}\), and the inferred effective \(\chi^{(2)}\) was \(4.9\times10^{-11}\,\text{m/V}\) [2203.09760]. The authors further argued that the NV-diamond THG was at least partly cascaded through SHG, via \(3\omega = 2\omega + \omega\).

## 4. Interference, phase, and symmetry selection rules

A recurrent result in photo-induced SHG is that the measured intensity is often controlled by coherent field addition rather than by a single nonlinear source term. In atom-thin heterocrystals, SHG from a bilayer was shown to be the superposition of two coherent second-harmonic fields,

\[
E_{2\omega} = E_{2\omega}^{(1)} + E_{2\omega}^{(2)} e^{i\Delta\phi},
\qquad
I_{2\omega} \propto \left|E_{2\omega}^{(1)} + E_{2\omega}^{(2)} e^{i\Delta\phi}\right|^2.
\]

This “second-harmonic Young interference” was directly measured in MoS\(_2\)/WS\(_2\) heterobilayers by spectral phase interferometry. The inter-material phase difference was strongly frequency dependent: \(\phi_{\mathrm{MoW}} \approx 61^\circ\) at 800 nm pump wavelength, about \(32^\circ\) at 900 nm, and close to zero by 1000 nm. Homo-bilayer MoS\(_2\)/MoS\(_2\) behaved essentially as an in-phase emitter, while MoS\(_2\)/WS\(_2\) generated elliptically polarized SH light because the two layer-generated SH fields carried a finite phase offset [2008.02403].

The current-induced graphene experiment belongs to the same interference class. There, the SH intensity was modeled as

\[
I^{2\omega} \propto (E_{0}^{2\omega}+E^{2\omega}(\vec J))^{2}
\approx \text{const}+2E^{2\omega}_{\text{Graphene}(\vec J)}E^{2\omega}_{\text{Si}}\cos(\Delta\phi)\propto J,
\]

so the measurable current dependence resided mainly in the interference cross term because the graphene-generated SH field was much weaker than the Si-surface SH field [1203.4395]. A plausible implication is that many “weak” induced SHG signals become experimentally visible only through phase-sensitive heterodyne-like interference with a stronger reference field.

Mode symmetry can impose equally strict selection rules. In a centrosymmetric gold two-wire transmission line, the dominant surface nonlinear polarization was taken as

\[
P^{(2)}_n = \epsilon_0 \chi^{(2)}_{nnn} E_n^2,
\]

and coupling to a guided second-harmonic mode was estimated by

\[
\eta = \left| \int_{\partial A} \mathbf{P}^{(2)}(2 \omega)\cdot \mathbf{E}(2 \omega)^*\, ds \right|^2.
\]

For either a purely symmetric or purely antisymmetric fundamental mode, \(E_n^2\) is symmetric, so only the symmetric SH mode is allowed; an antisymmetric SH mode appears only when the fundamental is a coherent superposition of symmetric and antisymmetric modes, so that the cross term \(E_sE_{as}\) contributes odd parity [1902.01720]. This replaced structural symmetry breaking by a mode-parity selection rule.

Polarization helicity can add another layer of symmetry selectivity. A multipolar treatment of SHG circular dichroism showed that two helicity-dependent SHG channels survive under \(E1\), \(E2\), and \(M1\) truncation: natural circular dichroism from the parity-odd \(E1'-E1-E1\) channel and magnetic circular dichroism from the parity-even \(E1'-E2-E1\) channel. The resulting observables were

\[
F(\mathrm{NCD}) = P_2 \langle U_{20}(1)\rangle,
\]

and

\[
F(\mathrm{MCD}) = q_0 P_2 \left\{ \langle T_{10}(2)\rangle + \sqrt{3}\,\langle T_{30}(2)\rangle \right\},
\]

identifying SHG circular dichroism as a probe of chiral and magnetic multipoles rather than only of scalar nonlinear conversion efficiency [1908.10640].

## 5. Integrated implementations, quasi-phase matching, and resonant enhancement

Integrated photonics has turned photo-induced SHG from a symmetry question into a device-engineering problem. In stoichiometric LPCVD Si\(_3\)N\(_4\) waveguides, SHG was observed only after an initialization delay of minutes to several hours under 1064 nm, 6.2 ps, 20 MHz pumping. The SH then grew with rates between \(10^{-4}\) and \(10^{-2}\,\mathrm{s^{-1}}\), with a threshold pump power \(P_0 = 1.22\) mW, and afterwards appeared instantaneously on subsequent pump turn-ons. The interpretation was the coherent photogalvanic effect: pump and seed SH fields create a rectified DC field and a semi-permanent space-charge grating, which induces an effective \(\chi^{(2)}\) grating with the quasi-phase-matching period

\[
\Lambda = \frac{2\pi}{\Delta k},\qquad \Delta k = 2k(\omega)-k(2\omega).
\]

The maximum measured conversion efficiency was 0.4%, corresponding to 53 \(\mu\)W at 532 nm with 13 mW of coupled IR power, and the extracted \(\chi^{(2)}_{\mathrm{eff}}\) was \(3.7 \pm 0.2\) pm/V [1710.03010].

Embedding the same optically written \(\chi^{(2)}\) concept in a high-\(Q\) microresonator greatly increased performance. In a Si\(_3\)N\(_4\) microring with separate telecom and visible coupling waveguides, a weak seed SHG initiated a photogalvanic DC field, yielding \(\chi^{(2)}_{\mathrm{eff}} = \chi^{(3)}E_{\mathrm{DC}}\). With exact azimuthal mode matching \(m_{2\omega}=2m_\omega\), thermal tuning, and resonant enhancement, the normalized SHG efficiency reached \((2{,}500 \pm 100)\,\%/\text{W}\), the on-chip SH output reached \((2.2 \pm 0.1)\) mW, and the absolute conversion efficiency reached \((22 \pm 1)\%\). The inferred \(\chi^{(2)}_{\mathrm{eff}}\) was \((0.20 \pm 0.04)\) pm/V and the underlying DC field was \((0.6 \pm 0.1)\) MV/cm [2003.12176].

Subsequent broadband pump-probe mapping of Si\(_3\)N\(_4\) microresonators showed that photo-induced SHG can persist far beyond the preferred doubly resonant point. A reconfigurable SHG bandwidth of about 0.4 nm, roughly 50 GHz, was measured, and the paper states that with reduced perturbation it could reach about 0.6 nm or 76 GHz. At the edge of the pump thermal triangle, the SH detuning reached 4.8 GHz while the measured conversion efficiency was still about \(0.1\%/\text{W}\), despite pump and SH linewidths of only 0.30 GHz and 0.65 GHz, respectively. This directly evidenced detuning-dependent all-optical poling and nonlinear multimode interaction [2510.05636].

Other integrated platforms reach high SHG by engineering modal dispersion and nonlinear overlap rather than by optically writing \(\chi^{(2)}\). In a reverse-polarization dual-layer LiNbO\(_3\) thin-film waveguide, modal phase matching between a 1574.6 nm pump mode and a 787.3 nm SH mode yielded a normalized conversion efficiency of \(5{,}540\%/\text{W}/\text{cm}/\text{cm}\) [2108.11047]. In a photonic crystal vertical cavity with an x-cut LiNbO\(_3\) membrane between two 1D photonic crystals, resonant confinement in the nonlinear layer produced an absolute SHG conversion efficiency of 0.28%, with \(Q \approx 200\) for one resonance and \(Q \approx 144\) for another [2307.15901]. In all-dielectric topological photonic crystals, SHG via double valley-Hall kink modes achieved tunable, bi-directional phase matching and an SHG directional dichroism approaching 0.97 [2007.04875].

## 6. Geometric design principles, applications, and interpretive boundaries

Recent band-geometric work has reformulated giant SHG as a transition-geometry problem. In noncentrosymmetric quantized-formal-polarization crystals, the Brillouin-zone average of the shift vector satisfies

\[
-e\int[d\mathbf{k}]\,\mathbf{R}_{mn}
= \mathbf{Q}W_{mn}+\mathbf{P}_m-\mathbf{P}_n,
\]

so, in the absence of optical zeros, the averaged shift vector is essentially controlled by the Berry-phase polarization difference between initial and final bands [2606.16108]. First-principles screening identified InPS\(_3\) and InNbBr\(_6\) as giant-SHG candidates because optically allowed transitions connect occupied and low-lying unoccupied states in different fractional Wannier-center sectors. In InPS\(_3\), the static SHG susceptibility was 0.23 nm\(^2\)/V, with intraband and interband contributions of 0.24 and \(-0.01\) nm\(^2\)/V; in InNbBr\(_6\), the total static SHG susceptibility was 0.70 nm\(^2\)/V, with intraband and interband contributions of 1.44 and \(-0.74\) nm\(^2\)/V [2606.16108]. This suggests a symmetry-guided design rule for large SHG: maximize optically active displacement between Wannier-center sectors while retaining strong dipole matrix elements.

The same body of work positions photo-induced SHG as a metrological tool. In multilayer graphene, current-controlled SHG suggests distant, nondestructive probing of current density [1203.4395]. In TMD nano-imaging, SHG functions as a local symmetry meter for stacking order and excitonic light–matter interaction [2111.06955]. The EFISH review identifies tunable SHG sources, on-chip frequency doubling, quasi-phase matching in silicon photonics, optical parametric amplification, frequency-comb self-referencing, carrier-transport imaging, ultrafast photocarrier probing, and THz field and polarization dynamics sensing as major application directions [2507.09306].

At the same time, photo-induced SHG is interpretively delicate because superficially similar observations may arise from different mechanisms. Silicon nitride and silicon oxynitride thin films on Si showed laser-induced SHG changes under 1028 nm femtosecond irradiation that initially resembled all-optical poling: for pure Si\(_3\)N\(_4\), the enhancement was about 3-fold, cumulative, and scaled approximately as \(I^6\). However, low-oxygen SiOxNy films showed laser-induced SHG reduction rather than enhancement; the response depended strongly on repetition rate when comparing 100 kHz and 10 kHz at the same number of pulses; and ex-situ annealing at 400 \(^\circ\)C for 30 min reproduced the trend, increasing SHG in Si\(_3\)N\(_4\) by about 20% but decreasing it in SiOxNy (0.5 sccm) by about 30%. The authors therefore attributed the observed SHG changes primarily to local layer heating caused by multiphoton absorption, together with bulk restructuring and stress modification, rather than to coherent photogalvanic optical poling alone [2212.09411]. This objective distinction echoes the broader EFISH review: induced SHG can reflect electrostatic fields, currents, resonance shifts, defect populations, or thermal modification, and mechanism assignment requires symmetry analysis, detuning dependence, spacer dependence, power dependence, and, when possible, direct phase-sensitive measurements [2507.09306].

Taken together, these studies show that photo-induced SHG is not a single effect but a family of nonlinear phenomena unified by controllable symmetry breaking and coherent wave mixing. Its present research frontier lies at the intersection of nonequilibrium transport, exciton physics, geometric band theory, and integrated nanophotonics, where the operative quantity is often not merely \(|\chi^{(2)}|\) but the coupled evolution of phase, mode symmetry, resonance condition, and optically written or field-induced asymmetry.

Source: https://www.emergentmind.com/topics/photo-induced-second-harmonic-generation-shg