---
title: Aperiodically Poled KTP (apKTP)
url: https://www.emergentmind.com/topics/aperiodically-poled-potassium-titanyl-phosphate-apktp
type: topic
---

# Aperiodically Poled KTP (apKTP)

Searching arXiv for recent and foundational papers on apKTP and related quasi-/aperiodic poling in KTP.
Aperiodically poled potassium titanyl phosphate (apKTP) denotes KTP crystals whose ferroelectric domain inversion pattern is deliberately engineered to be non-periodic. In such structures, the spatial modulation of the second-order nonlinear coefficient \(d(z)\) is not a single-period grating; instead, it is designed to synthesize multiple Fourier components in the grating spectrum. Quasi-periodically poled KTP (QPPKTP) is a specific, well-controlled form of apKTP in which the domain sequence has long-range order but no strict periodicity, enabling more than one strong reciprocal-lattice vector in the grating spectrum and thereby supporting multiple quasi-phase-matched \(\chi^{(2)}\) processes in a single crystal. Reported apKTP implementations span simultaneous SHG and SFG for cascaded THG and orbital-angular-momentum (OAM) spectral manipulation, broadband adiabatic SHG with pulse shaping, and broadband parametric down-conversion in high-gain SU(1,1) interferometry with undetected mid-infrared photons [2204.04175] [2204.04173] [1804.02639] [2509.03349].

## 1. Domain engineering and the distinction from periodic poling

Standard periodically poled KTP (PPKTP) uses a single spatial period \(\Lambda\) to create one dominant grating vector \(G = 2\pi/\Lambda\). This typically phase-matches a single process at a given wavelength and temperature, providing one narrow quasi-phase-matching peak. By contrast, QPPKTP/apKTP provides multiple grating vectors \(\{G_m\}\) with appreciable Fourier weights, thereby enabling concurrent phase matching of coupled processes such as SHG of the fundamental and SFG between the fundamental and its second harmonic to realize cascaded THG, all within a single crystal [2204.04175].

A central point is that “aperiodic” in this context does not imply uncontrolled disorder. Quasi-periodic poling is explicitly described as a specific, well-controlled form of aperiodic poling where the domain sequence has long-range order but no strict periodicity. In one QPPKTP implementation for THG of structured light, the crystal is designed by the projection method, a standard route to 1D nonlinear quasicrystals that produces a dense set of reciprocal vectors for quasi-phase matching. The structure is built from two types of building blocks, \(A\) and \(B\), each comprising a pair of antiparallel domains, with widths set by \(l_A\), \(l_B\), a projection angle \(\theta\), and \(\gamma = \tan \theta\); the overall crystal length is \(5.532\ \mathrm{mm}\) [2204.04173].

The reciprocal vectors of projection-designed 1D nonlinear quasicrystals can be written in the general form
$$
G_{p,q} = 2\pi\,\frac{p + q\,\gamma}{\Lambda},
$$
with \(p,q \in \mathbb{Z}\). In another apKTP regime, the grating is not designed to realize several discrete QPM channels simultaneously, but to vary continuously along the crystal length through a spatially dependent local wavevector \(K(z) = 2\pi/\Lambda(z)\). This chirped or aperiodic design lets different frequencies encounter their local phase matching at different positions in the crystal and is the basis of adiabatic broadband SHG and broadband PDC bandwidth engineering [1804.02639] [2509.03349].

## 2. Quasi-phase matching, tensor access, and cascaded interactions

The apKTP/QPPKTP implementations discussed in the cited works are based on type-0 interactions in KTP, with all waves co-polarized so as to access the largest tensor element \(d_{33}\). For the 1560 nm QPPKTP devices used for OAM manipulation and THG of structured light, all interacting waves \((\omega, 2\omega, 3\omega)\) are polarized along the same principal axis, denoted type-0 \((eee)\) in one report and \(|V\rangle \to |V\rangle \to |V\rangle\) in another. Using \(d_{33}\) ensures high nonlinear coupling for both SHG \((\omega+\omega\to2\omega)\) and SFG \((\omega+2\omega\to3\omega)\), while co-polarization minimizes group-velocity and spatial walk-off effects relative to mixed-polarization schemes [2204.04175] [2204.04173].

For cascaded THG, the phase mismatches are defined as
$$
\Delta k_{\mathrm{SHG}} = k_{2\omega} - 2k_\omega,
\qquad
\Delta k_{\mathrm{SFG}} = k_{3\omega} - k_{2\omega} - k_\omega.
$$
In the cited reports, the QPM condition is written either as
$$
\Delta k - G_m = 0
$$
or as
$$
\Delta k_{\mathrm{SHG}} + G_{m1} = 0,\qquad
\Delta k_{\mathrm{SFG}} + G_{m2} = 0.
$$
The key physical point is unchanged: the aperiodic grating supplies distinct reciprocal vectors so that both SHG and SFG are phase matched concurrently within one crystal [2204.04175] [2204.04173].

For quasi-periodic designs with a discrete Fourier spectrum, if \(c_m\) denotes the \(m\)-th Fourier coefficient of the normalized domain pattern, then the \(m\)-th QPM channel has effective coupling
$$
d_{\mathrm{eff},m} \propto d_{33} c_m.
$$
The design rationale is to allocate sufficient weight to two target components, \(G_a\) and \(G_b\), such that
$$
\Delta k_{\mathrm{SHG}} - G_a = 0,
\qquad
\Delta k_{\mathrm{SFG}} - G_b = 0,
$$
thus enabling concurrent phase matching for both steps of the cascade [2204.04175].

In apKTP used for broadband PDC, the poling period varies continuously and the local QPM condition is
$$
\Delta k(\omega_s,\omega_i,z) = k_p(\omega_p)-k_s(\omega_s)-k_i(\omega_i)-\frac{2\pi}{\Lambda(z)}.
$$
The corresponding phasematching function is
$$
\Phi(\omega_s,\omega_i) \propto \int_0^L dz\, \chi^{(2)}(z)\, e^{i\Delta k(\omega_s,\omega_i,z) z}.
$$
By tailoring \(\Lambda(z)\), the crystal design shapes both the amplitude and the phase of \(\Phi\), suppresses spectral sidelobes, and flattens gain [2509.03349].

## 3. OAM spectral manipulation and third-harmonic generation of structured light

In the OAM-manipulation experiments at a fundamental wavelength \(\lambda_\omega = 1560\ \mathrm{nm}\), the QPPKTP crystal was operated in a type-0 \((eee)\) configuration to simultaneously phase-match SHG \((1560 \to 780\ \mathrm{nm})\) and SFG \((1560 + 780 \to 520\ \mathrm{nm})\). Under rotational symmetry, OAM is conserved in three-wave mixing. For an input beam with a single OAM mode \(\ell\),
$$
\ell_{2\omega} = 2\ell,
\qquad
\ell_{3\omega} = 3\ell.
$$
For a general azimuthal field
$$
E_\omega(\phi) = \sum_\ell c_\ell e^{i\ell\phi},
$$
the output OAM spectra satisfy
$$
c^{(2\omega)}_m \propto \sum_\ell c_\ell c_{m-\ell},
\qquad
c^{(3\omega)}_n \propto \sum_{\ell,m} c_\ell c_m c_{n-\ell-m}.
$$
Thus SHG generates the convolution of the input OAM spectrum with itself, and THG produces the triple convolution. In the same framework, the experiment identified phase doubling and phase tripling through
$$
E_{2\omega} \propto E_\omega^2 = A_\omega^2 e^{i2\psi_\omega(\phi)},
\qquad
E_{3\omega} \propto E_{2\omega}E_\omega = A_\omega^3 e^{i3\psi_\omega(\phi)}.
$$
The measurements confirmed the theoretical predictions for integer and fractional topological charges, including the special case where fractional input charges yield whole-doughnut intensity distributions in the far field when the effective charge becomes integer after doubling or tripling [2204.04175].

The same work used an 80 MHz repetition-rate femtosecond laser with 200 fs pulses at 1560 nm. The beam was azimuthally phase-modulated by SLM1, reduced by a \(1{:}15\) telescope before entering the QPPKTP, and the SH and TH OAM spectra were measured by projection via SLM2/SLM3, a single-mode fiber, and a power meter. For fundamental-wave superpositions with \(W_{\omega 0}:W_{\omega 1}=1:1\) and \(W_{\omega -1}:W_{\omega 0}:W_{\omega 1}=1:1:1\), the measured SH and TH OAM spectra agreed with theory, with fidelities \(0.98 \pm 0.02\) and \(0.98 \pm 0.02\) in the two-mode case, and \(0.99 \pm 0.02\) and \(0.97 \pm 0.03\) in the three-mode case. The same study further inversely designed fundamental-wave phase functions to produce target OAM spectra directly at SH or TH; the measured fidelities to the targets ranged from \(0.87 \pm 0.02\) to \(0.95 \pm 0.02\), and from \(0.94 \pm 0.02\) to \(0.99 \pm 0.02\) when referenced to the inversely designed spectra. The fidelity metric was
$$
F = \sum_m \sqrt{W^E_m W^T_m}.
$$
A stated practical limitation was imperfect realization of the inverse phase function on the SLM, which slightly reduced the spectral fidelity at SH and TH relative to targets [2204.04175].

A related QPPKTP implementation demonstrated THG of scalar and vector structured light in a nonlinear Sagnac interferometer. Because the QPPKTP only supports \(|V\rangle \to |V\rangle \to |V\rangle\) nonlinear interactions, the interferometer routes each fundamental-wave polarization component so that it reaches the crystal as \(|V\rangle\), and a dual-wavelength HWP at \(45^\circ\) flips \(|H\rangle \leftrightarrow |V\rangle\) at \(\omega\) and \(3\omega\). For a general input vector state
$$
|\phi\rangle = \alpha\,|H\rangle\,|{+}m\rangle + \beta\,|V\rangle\,|{-}m\rangle,\qquad \alpha^2+\beta^2=1,
$$
the output TH state is
$$
|\phi'\rangle = \frac{\alpha^3\,|V\rangle\,|{-}3m\rangle + \beta^3\,|H\rangle\,|{+}3m\rangle}{\sqrt{\alpha^6+\beta^6}}.
$$
The work verified \(\ell_{3\omega}=3\ell_\omega\) with measured tilt-lens interferograms and reported, for a Gaussian fundamental with average input power \(\approx 2\ \mathrm{W}\) and beam radius \(\approx 100\ \mu\mathrm{m}\) in the crystal, single-pass efficiencies of \(17.6\%\) for SHG and \(8.0\%\) for THG. For structured-light experiments at \(91.2\ \mathrm{mW}\), reported THG efficiencies included \(\eta_{\mathrm{THG}} \approx 1.2\times10^{-4}\) for input \(|V\rangle|-1\rangle\), \(\eta_{\mathrm{THG}} \approx 3.0\times10^{-5}\) for input \(|L\rangle|-2\rangle\), and \(\eta_{\mathrm{THG}} \approx 3.1\times10^{-5}\) for a vector input \(|\phi_1\rangle\) [2204.04173].

## 4. Adiabatic apKTP for broadband second-harmonic generation and pulse shaping

In broadband SHG, apKTP is used not to synthesize a few discrete reciprocal vectors, but to implement a spatially varying QPM wavevector \(K(z)=2\pi/\Lambda(z)\) so that different frequencies encounter their local phase matching at different positions. If this spatial variation is slow compared to the nonlinear coupling, the conversion becomes adiabatic and efficiently spans a wide bandwidth [1804.02639].

One reported adiabatic apKTP crystal for a tunable mode-locked Ti:sapphire oscillator had length \(10\ \mathrm{mm}\), nominal \(50\%\) duty cycle, and a local grating wavevector
$$
K(z)=118.2\,z^3-45.2\,z^2-997.9\,z+7957.1\ \mathrm{cm}^{-1},
\qquad z\in[-5\ \mathrm{mm},+5\ \mathrm{mm}],
$$
corresponding to a monotonic chirp of \(\Lambda(z)\) from approximately \(7.46\ \mu\mathrm{m}\) to \(8.43\ \mu\mathrm{m}\). The interaction was type-0 \(eee\), using \(d_{33}\), and the oscillator operated at 80 MHz with a fundamental spectral FWHM of \(17.5\ \mathrm{nm}\), pulse duration of about \(70\ \mathrm{fs}\), and pulse energy of \(10\)–\(30\ \mathrm{nJ}\). The Gaussian beam had a measured \(40\ \mu\mathrm{m}\) FWHM at the crystal, and diffraction was negligible over the crystal length [1804.02639].

The reported performance was an efficient flat conversion region of more than \(35\ \mathrm{nm}\) of fundamental bandwidth, measured acceptance bandwidth bigger than \(40\ \mathrm{nm}\), modeled bandwidth up to \(\sim 80\ \mathrm{nm}\), and energy conversion efficiency above \(50\%\) with \(\sim 70\ \mathrm{fs}\) pulses. Modeling included two-photon absorption with \(\beta_{\mathrm{TPA}}\approx 4\ \mathrm{cm/GW}\), which was found important at the oscillator’s average powers and explained deviations from loss-free predictions [1804.02639].

Because the apKTP transfer function is broad and nearly flat across the band, spectral phase imposed on the fundamental can be translated into controlled broadband SH structure. The work assigned an absolute-value spectral phase,
$$
\phi(\omega)=\alpha|\omega-\omega_0|,
$$
which yields two pulses separated by \(\Delta t_{\mathrm{pp}}=2\alpha\), and a \(\pi\)-step spectral phase,
$$
\phi(\omega)=\pi\,\Theta(\omega-\omega_s),
$$
which produced a tunable deep notch in the SH spectrum, with possible complete destructive interference at specific SH wavelengths. The paper identifies these results as a proof of concept for pulse-shaping applications in nonlinear spectroscopy and imaging [1804.02639].

## 5. Broadband PDC, SU(1,1) interferometry, and undetected mid-infrared photons

A different apKTP modality appears in high-gain SU(1,1) interferometry, where the crystal is the broadband PDC gain medium. The reported system used several apKTP samples fabricated by electric-field domain inversion to realize prescribed aperiodic poling profiles \(\Lambda(z)\), operated in type-0 phase matching with pump, signal, and idler all Z-polarized and propagating along the X-axis of KTP. A 15 ps pulsed laser at 532 nm and 1 kHz repetition rate pumped the apKTP in a collinear geometry. Two experimental designs were implemented: a linear profile with \(\Lambda_1 = 12.3\)–\(14.0\ \mu\mathrm{m}\), producing signal centered at approximately \(635\ \mathrm{nm}\), idler centered at approximately \(3.3\ \mu\mathrm{m}\), and measured signal FWHM of approximately \(10\ \mathrm{nm}\); and a logarithmic profile with \(\Lambda_2 = 10.9\)–\(14.0\ \mu\mathrm{m}\), signal centered at approximately \(670\ \mathrm{nm}\), idler centered at approximately \(2.6\ \mu\mathrm{m}\), and measured signal FWHM of approximately \(22\ \mathrm{nm}\) [2509.03349].

The interferometer used a single apKTP crystal twice. On the first pass, the pump generated broadband signal-idler pairs via high-gain PDC; the idler traversed a sample, and on the second pass signal and idler recombined with the pump in the same crystal, undergoing optical parametric amplification. Only the signal was detected on an optical spectrum analyzer with \(0.6\ \mathrm{nm}\) resolution, enabling low-coherence interferometry with undetected mid-IR photons. The interferogram was modeled as
$$
S(\omega_s)=S_0(\omega_s)\left[1+\mathcal{V}_0\cos\!\left(2z(\omega_p-\omega_s)/c+\rho(\omega_s)\right)\right].
$$
The chirped phase \(\rho(\omega_s)\), set by group-velocity dispersion and the poling profile, was compensated numerically by quadratic phase removal after Hilbert transformation; physical dispersion compensation by a glass slab was also suggested [2509.03349].

The reported system achieved a signal-to-noise ratio as high as \(40\ \mathrm{dB}\) and axial resolution of \(30\ \mu\mathrm{m}\), with a \(3\ \mu\mathrm{m}\)-centered idler beam. By increasing the poling-period range, the axial resolution was improved to \(17\ \mu\mathrm{m}\). The measured axial resolution was \((29.4 \pm 0.2)\ \mu\mathrm{m}\) for Design 1 and \(17\ \mu\mathrm{m}\) for Design 2; the \(10\ \mathrm{dB}\) roll-off depth was approximately \(270\)–\(280\ \mu\mathrm{m}\). With \(1\ \mathrm{mW}\) pump, the detected signal power was \((9.0 \pm 0.5)\ \mathrm{nW}\) and the SNR was approximately \(40\ \mathrm{dB}\). The ratio \(P_{\mathrm{signal},2}/P_{\mathrm{idler},1}\) reached 208, whereas the low-gain expectation under the same \(\omega_s,\omega_i\) was \(\sim 10.3\). Demonstrations included reconstruction of a \(100\ \mu\mathrm{m}\) aluminum step measured as \(100 \pm 3\ \mu\mathrm{m}\), and measurement of a \(50\ \mu\mathrm{m}\) silicon layer behind a \(1\ \mathrm{mm}\) BBAR-coated Ge window, yielding \(52.2 \pm 0.9\ \mu\mathrm{m}\) after refractive-index correction [2509.03349].

## 6. Practical constraints, common misconceptions, and reported directions

A recurring practical requirement across apKTP implementations is high-fidelity domain engineering. QPPKTP/apKTP requires precise electric-field poling to impose a quasi-periodic domain-length set, a prescribed duty cycle, or a spatially varying \(\Lambda(z)\). Duty-cycle errors, domain-length variations, and profile deviations perturb \(\chi^{(2)}(z)\), yielding spectral ripple and phase errors. This is especially consequential when the design goal is to allocate appreciable Fourier weight to several reciprocal vectors or to maintain a broad and flat chirped phase-matching response [2204.04175] [2509.03349].

A common misconception is to treat quasi-periodic and aperiodic poling as synonymous with random domain inversion. The cited reports instead describe quasi-periodic poling as a specific, controlled subset of apKTP with long-range order but no strict periodicity. Another misconception is that simultaneous phase matching of several processes comes without trade-offs. In fact, the acceptance bandwidth for each QPM channel narrows as \(c_m\) is spread among multiple Fourier components, and sharing the grating spectrum between two processes reduces the peak \(d_{\mathrm{eff}}\) available to each. The reported benefit is integration: one crystal, co-linear beams, and single alignment, which is particularly advantageous for mode-sensitive operations such as OAM spectral manipulation [2204.04175].

A further point of clarification concerns THG. In the structured-light QPPKTP experiments, direct THG via \(\chi^{(3)}\)-like QPM is not used; the work explicitly relies on cascaded \(\chi^{(2)}\) processes enabled by multiple grating vectors. Practical limitations reported in these experiments included weak focusing to preserve high-fidelity spatial and polarization structure, SLM damage thresholds that capped average fundamental power at \(91.2\ \mathrm{mW}\), and the need for triple-wavelength optics for simultaneous extraction and control of SH at \(780\ \mathrm{nm}\) together with TH at \(520\ \mathrm{nm}\) [2204.04173].

Across the cited works, KTP is associated with favorable thermo-optic properties, high damage threshold, and lower photorefractive sensitivity compared to materials such as LiNbO\(_3\). No active temperature tuning was reported in the apKTP studies summarized here; room-temperature operation was sufficient in the cited experiments, and no photorefractive damage was observed with the employed femtosecond pulses in the OAM-manipulation work [2204.04175] [2509.03349].

Reported future directions include optimized inverse-phase algorithms for higher-fidelity OAM spectral shaping, apKTP designs with three or more strong grating vectors to concurrently phase-match additional processes, integration with entangled-photon sources for high-dimensional quantum state engineering, broader poling ranges and optimized chirp/apodization for bandwidth and flatness, pump-pulse shaping, longer crystals with engineered \(\Lambda(z)\) to support wider bands while managing nonlinear loss, and focusing the idler to enable OCT lateral resolution [2204.04175] [2509.03349]. Taken together, these results suggest that apKTP is not a single device class but a crystal-engineering framework in which the longitudinal domain pattern is used as a direct control variable for nonlinear coupling topology, bandwidth, spectral phase, and the fidelity of structured-light transformations.

Source: https://www.emergentmind.com/topics/aperiodically-poled-potassium-titanyl-phosphate-apktp