Second-Harmonic Generation

Definition

Second-harmonic generation (SHG) is a nonlinear optical process in which two photons of frequency ω are converted into a single photon of frequency 2ω inside a χ⁽²⁾ crystal. The new photon carries exactly the combined energy of its two parents, so its frequency is doubled and its wavelength halved.

photon(ω) + photon(ω)  →  photon(2ω)

energy:    ℏω + ℏω = ℏ(2ω)      ✓ conserved
momentum:  ℏk(ω) + ℏk(ω) = ℏk(2ω)   ⟹  phase matching

Because energy is conserved automatically (two equal energies add to twice the energy), the only real engineering challenge is conserving momentum — and that is exactly what the phase-matching condition Δk = 0 enforces. Get phase matching wrong and you still satisfy energy conservation, but the green light never builds up.

How it works — the χ⁽²⁾ nonlinearity

When light passes through any transparent medium, its electric field pushes the bound electrons back and forth, and that oscillating charge re-radiates light. In a weak field the response is linear — the polarization P tracks the field E proportionally. But crank the field high enough (and a focused laser gets you there) and the response stops being a straight line. Expanded as a power series:

P = ε₀ ( χ⁽¹⁾·E  +  χ⁽²⁾·E²  +  χ⁽³⁾·E³  + … )
         └ linear ┘  └ second harmonic ┘  └ third harmonic, Kerr ┘

Now feed in a field oscillating at one frequency, E = E₀·cos(ωt), and look at the second-order term:

χ⁽²⁾·E² = χ⁽²⁾·E₀²·cos²(ωt)
        = ½·χ⁽²⁾·E₀² · [ 1 + cos(2ωt) ]
          └ DC term ┘   └ oscillates at 2ω! ┘

Squaring a cosine produces a constant (a static field, called optical rectification) plus a brand-new oscillation at twice the frequency. That 2ω term is a charge wiggling at double speed, and a charge wiggling at 2ω radiates light at 2ω. That radiated light is the second harmonic. The whole effect lives entirely in the χ⁽²⁾E² term — no χ⁽²⁾, no doubling.

There is a deep symmetry constraint hiding here. In a material with inversion symmetry (a centrosymmetric crystal, a gas, a liquid, ordinary glass), flipping the field direction must flip the polarization exactly: P(−E) = −P(E). But the χ⁽²⁾E² term is even — it gives the same sign whether E is positive or negative — so it can only survive if χ⁽²⁾ = 0. SHG is forbidden in any centrosymmetric medium. This is why frequency-doubling crystals are always non-centrosymmetric: KTP, BBO, LBO, KDP, lithium niobate. It's also why the very first thing materials scientists test for a candidate nonlinear crystal is whether it produces a second-harmonic signal at all.

Worked example — the green laser pointer

The green laser pointer is the most familiar SHG device on Earth, and it is a beautiful Russian doll of physics. Trace the light from the back:

1. An infrared diode laser emits around 808 nm and acts purely as a pump.
2. It optically pumps a Nd:YVO₄ or Nd:YAG crystal, which lases at 1064 nm — deep infrared, completely invisible to the eye.
3. That 1064 nm beam passes through a KTP frequency-doubling crystal, where SHG converts it to 532 nm green: 1064 / 2 = 532 nm exactly, because doubling the frequency halves the wavelength.
4. An IR-blocking filter (in a properly built unit) removes the leftover 1064 nm so only green emerges.

The numbers are exact and worth stating plainly:

fundamental:   λ₁ = 1064 nm   (ω, infrared, invisible)
second harm.:  λ₂ = λ₁ / 2 = 532 nm   (2ω, green, visible)

check:  f = c/λ
        f₁ = 3.0×10⁸ / 1064×10⁻⁹ = 2.82×10¹⁴ Hz
        f₂ = 3.0×10⁸ /  532×10⁻⁹ = 5.64×10¹⁴ Hz = 2·f₁   ✓

Here is the unsettling consequence: the green you see is the weakly-converted second harmonic. The original 1064 nm beam inside the device is typically far more powerful than the green output, and it is invisible. Cheap green pointers that skip the IR-blocking filter can leak hundreds of milliwatts of invisible, focusable infrared — which is precisely why poorly-made green pointers are an eye-safety concern in a way that the visible green light alone would not suggest.

Phase matching — the real difficulty

If doubling were as easy as "have a χ⁽²⁾ crystal," every clear non-centrosymmetric crystal would glow green in sunlight. It doesn't, because of dispersion: the refractive index n depends on wavelength, so the fundamental at ω and its second harmonic at 2ω travel through the crystal at different speeds.

Each thin slice of crystal radiates a little second harmonic. For those contributions to add constructively into a strong beam, the 2ω wave generated at the front of the crystal must stay in step with the 2ω wave generated at the back. The bookkeeping is captured by the phase mismatch:

Δk = k(2ω) − 2·k(ω) = (2ω/c)·[ n(2ω) − n(ω) ]

phase matching  ⟺  Δk = 0  ⟺  n(2ω) = n(ω)

In a normally-dispersive material n(2ω) > n(ω) always, so Δk ≠ 0 and the second harmonic dephases. The distance over which it builds before flowing back into the fundamental is the coherence length:

L_c = π / |Δk|     (typically a few µm without phase matching)

The clever fix is birefringent phase matching: in an anisotropic crystal the index depends on polarization direction, so you can tilt the crystal until the (slow) extraordinary index at one frequency equals the (fast) ordinary index at the other, forcing n(2ω) = n(ω). The magic tilt is the phase-matching angle. An alternative, quasi-phase-matching, periodically flips the sign of χ⁽²⁾ (periodically-poled lithium niobate, PPLN) every coherence length so the conversion always adds up even though Δk ≠ 0.

Performance — why output scales as I²

The single most important quantitative fact about SHG: second-harmonic power scales as the square of the input intensity. The chain of reasoning is short:

P(2ω) ∝ χ⁽²⁾² · L² · I(ω)² · sinc²(Δk·L / 2)

• polarization at 2ω  ∝ E²          (the χ⁽²⁾ term)
• E  ∝ √I                            (intensity = field squared)
• radiated SH field   ∝ E²  ∝ I
• SH intensity        ∝ (field)²  ∝ I²

⟹  double the input  ⟹  4× the green

Three levers fall straight out of that formula:

• Intensity I². Tight focusing and high peak power win quadratically. This is why a pulsed laser, whose peak intensity may be a million times its average, can convert 50% to the second harmonic while a continuous-wave beam of the same average power converts a small fraction of a percent.
• Crystal length L². With perfect phase matching (Δk = 0), the sinc² term equals 1 and conversion grows as the square of crystal length — longer crystals double harder, until the beam diffracts or walks off.
• Phase mismatch sinc²(ΔkL/2). The whole conversion is gated by this factor. At Δk = 0 it is 1; move off phase matching and it collapses, oscillating with a period set by the coherence length. The angular and temperature tolerances of a real crystal (its "acceptance bandwidth") are exactly how sharply this sinc² peaks.

Variants and related nonlinear processes

SHG is the simplest member of a whole family of three-wave (χ⁽²⁾) and four-wave (χ⁽³⁾) mixing processes. Knowing where it sits clarifies what it is and is not:

Process	Photon bookkeeping	Order	Typical use
Second-harmonic generation (SHG)	ω + ω → 2ω	χ⁽²⁾	Green laser pointers, frequency doubling
Sum-frequency generation (SFG)	ω₁ + ω₂ → ω₁+ω₂	χ⁽²⁾	Surface spectroscopy, UV generation
Difference-frequency generation (DFG)	ω₁ − ω₂ → ω₁−ω₂	χ⁽²⁾	Mid-IR and THz sources
Optical parametric oscillation (OPO)	ω → ω₁ + ω₂	χ⁽²⁾	Tunable laser sources (SHG run in reverse)
Third-harmonic generation (THG)	ω + ω + ω → 3ω	χ⁽³⁾	Microscopy contrast, UV; works in glass too
Optical Kerr / self-focusing	ω → ω (n depends on I)	χ⁽³⁾	Mode-locking, supercontinuum, soliton pulses

A nice unifying picture: OPO and DFG are SHG running backwards, splitting one high-frequency photon into two lower-frequency ones. THG is doubling's cousin but lives in the χ⁽³⁾ term, so unlike SHG it can occur even in centrosymmetric media like glass.

Common pitfalls and misconceptions

• "It's just absorption then re-emission at a new color." No. SHG is a coherent, near-instantaneous nonlinear scattering process — no electrons get excited to real energy levels and decay. The crystal is transparent at both wavelengths; it never stores the energy.
• "Any transparent crystal can do it." Only non-centrosymmetric crystals have a nonzero χ⁽²⁾. Glass, water, and centrosymmetric crystals produce no SHG (in the bulk) no matter how hard you push them. This selection rule is so reliable it's used as a structural probe.
• "More power always means more green." True only up to a point and only with phase matching. Without Δk = 0, the green oscillates within one coherence length and stays microscopic regardless of input power. And at very high conversion the fundamental gets depleted, so growth saturates rather than continuing as I² forever.
• "The green pointer is fundamentally a green laser." It's an infrared (1064 nm) laser wearing a doubling crystal. The lasing happens in the IR; the green is a converted afterthought — and the leftover IR can be the most dangerous part.
• "Phase matching means the two beams are in phase at the entrance." It means their wave-vectors match (Δk = 0) so they stay in step throughout the crystal — a momentum condition, not a one-time alignment at the face.
• "Energy isn't conserved — you doubled the frequency." Two photons went in for every one that came out. Frequency doubled, photon count halved, total energy unchanged.

Where SHG shows up

• Green and blue laser sources. 1064 → 532 nm green pointers and laboratory lasers; 946 → 473 nm blue. Doubling gives visible coherent light from mature, efficient infrared laser technology.
• Ultraviolet generation. Doubling a doubled beam (532 → 266 nm) or doubling a green pump produces deep-UV light for lithography, micromachining, and DNA spectroscopy where no good direct laser exists.
• SHG microscopy. Collagen, muscle myosin, and microtubules are non-centrosymmetric and emit a second harmonic under a femtosecond laser — giving label-free, high-contrast biological imaging with intrinsic optical sectioning.
• Surface and interface science. Because bulk centrosymmetric materials give no SHG, a second-harmonic signal comes almost entirely from the symmetry-broken surface — making SHG an exquisitely surface-sensitive probe of monolayers and buried interfaces.
• Laser pulse diagnostics. Autocorrelators measure femtosecond pulse durations by overlapping a pulse with a delayed copy of itself in an SHG crystal — the I² nonlinearity makes the doubled signal peak only when the pulses overlap in time.
• Materials characterization. SHG is a fast, contactless test for whether a new crystal lacks inversion symmetry, a key property for ferroelectrics and nonlinear photonics.

Derivation sketch — the coupled-wave picture

Under the slowly-varying-envelope approximation, the second-harmonic field amplitude A₂ grows along the crystal (coordinate z) driven by the square of the fundamental A₁:

dA₂/dz  ∝  i · χ⁽²⁾ · A₁² · exp(i·Δk·z)

If the fundamental is undepleted (A₁ roughly constant for low conversion), integrate from 0 to L:

A₂(L) ∝ χ⁽²⁾·A₁² · [ exp(i·Δk·L) − 1 ] / (i·Δk)

|A₂(L)|² ∝ χ⁽²⁾²·|A₁|⁴ · L² · sinc²(Δk·L / 2)
         └────────────────┘    └ length²┘  └ phase-match factor ┘
              ∝ I(ω)²

Two results drop out immediately. First, |A₁|⁴ means the second-harmonic intensity goes as the square of the fundamental intensity — the I² law. Second, the bracket reduces to L² only when Δk → 0; otherwise the sinc² caps the useful interaction at one coherence length L_c = π/|Δk|. Set Δk = 0 and you recover the headline scaling P(2ω) ∝ I² · L², the design equation for every frequency doubler.