---
title: Entangled Two-Photon Absorption
url: https://www.emergentmind.com/topics/entangled-two-photon-absorption-etpa
type: topic
---

# Entangled Two-Photon Absorption

Entangled-two-photon absorption (ETPA) is the proposed counterpart of ordinary two-photon absorption when the excitation is driven not by intense classical light but by correlated, usually entangled, photon pairs. In the standard low-flux picture, a molecule, atom, or solid-state system absorbs the two photons of a pair to reach a final excited state, and the absorption rate can scale linearly with photon-pair flux rather than quadratically as in classical two-photon absorption. At the same time, the subject remains controversial because transmission and fluorescence observables can be contaminated by single-photon-loss mechanisms, hot-band absorption, scattering, and detector noise, so a reduced coincidence or fluorescence signal is not by itself a unique witness of ETPA [2208.11387][1910.07346].

## 1. Rate laws and physical mechanism

The standard rate decomposition used throughout the literature is
\[
R_c = \delta_r \phi^2, \qquad R_e = \sigma_e \phi, \qquad R_2 = \delta_r \phi^2 + \sigma_e \phi,
\]
where \(R_c\) is the classical two-photon contribution, \(R_e\) is the entangled-photon contribution, \(\delta_r\) is the classical TPA cross section, \(\sigma_e\) is the ETPA cross section, and \(\phi\) is the photon-pair flux density. In this formulation, the linear term is the signature of ETPA, and the crossover is estimated at
\[
\tilde{\phi} = \frac{\sigma_e}{\delta_r}.
\]
At low flux density, the entangled contribution dominates; at higher flux density the classical quadratic term eventually wins [1910.07346].

The microscopic rationale is usually split into two linked ingredients. First, photon-number correlations matter because the photons are created in pairs, so if one photon arrives, its partner is present as well. Second, frequency anticorrelations matter because two-photon absorption is sensitive to the sum frequency rather than the two photons separately. One formulation writes the ETPA probability as
\[
P = 2\,T_e\,\frac{N_{\mathrm{EPP}}}{A_0}\;\sigma^{(2)}\; n,
\]
with
\[
n = 2\int dx\, L(x)\,|K_\phi(x)|^2,
\]
where \(T_e\) is the entanglement time, \(A_0\) is the beam area at the molecule, \(\sigma^{(2)}\) is the conventional classical TPA cross section, and \(n\) is a spectral overlap factor. In this description, the strongest enhancement occurs when the joint spectral amplitude is narrow along the diagonal direction \(\omega+\omega'=\text{const}\) and broad along the antidiagonal direction \(\omega-\omega'=\text{const}\) [2103.10530].

A frequently used phenomenological relation is
\[
\sigma_e \simeq \frac{\delta_r}{A_e T_e},
\]
with \(A_e\) the entanglement area and \(T_e\) the entanglement time. A related low-flux formulation writes the rate as
\[
R^{TPA} = \sigma_e \phi + \delta_r \phi^2.
\]
More recent microscopic analyses, however, emphasize that this simple form is recovered only in a specific limit and can require a correction factor depending on sample linewidth broadening relative to the pump bandwidth [2312.00167].

In molecular discussions, ETPA is often described as closer to single-photon absorption than to ordinary TPA because the pair behaves as if it were a single entity with pump-like bandwidth \(\Delta_p\), with the condition
\[
\Delta_p \ll \Delta_f,
\]
where \(\Delta_f\) is the molecular final-state bandwidth. This emphasizes that the relevant resource is not merely two photons, but a spectrally and temporally correlated biphoton wavepacket [1910.07346].

## 2. Biphoton state, spectral structure, and two-photon filtering

A standard starting point is an SPDC-generated two-photon state. One representative form is
\[
|\Psi\rangle=\int d\nu_s\,d\nu_i\,\Phi(\nu_s,\nu_i)\, \hat a^\dagger(\nu_s+\omega_0)\hat a^\dagger(\nu_i+\omega_0)|0\rangle,
\]
with joint spectral amplitude
\[
\Phi(\nu_s,\nu_i)=E_p(\nu_s+\nu_i)\, \sinc\!\left[\frac{L(\eta_s\nu_s+\eta_i\nu_i)}{2}\right] \exp\!\left[-i\frac{L(\eta_s\nu_s+\eta_i\nu_i)}{2}\right].
\]
An equivalent notation writes
\[
\ket{\psi}=\ket{0}_1 \ket{0}_2+\xi \int d\omega_s \int d\omega_i\, f(\omega_s,\omega_i)\ket{\omega_s}\ket{\omega_i},
\]
where \(|f(\omega_s,\omega_i)|^2\) is the joint spectral intensity (JSI) [2208.11387][2512.01117].

A central conceptual move in current ETPA theory is to model the sample as a two-photon notch filter acting on the sum-frequency coordinate. One form is
\[
h_N(\omega_s,\omega_i)=1-\eta \exp\!\left[-\frac{(\omega_s+\omega_i-2\Omega_{N_0})^2}{2\sigma_N^2}\right],
\]
so that the transmitted biphoton amplitude becomes
\[
f_{\text{sam}}(\omega_s,\omega_i)=f(\omega_s,\omega_i)\,h_N(\omega_s,\omega_i).
\]
A closely related expression writes the two-photon filter as
\[
f_{\text{TP}}(\nu_s,\nu_i)=1-\exp\!\left[-\frac{(\nu_s+\nu_i)^2}{2\sigma_{\text{TP}}^2}\right].
\]
In both cases, true ETPA removes a frequency-correlated slice of the JSI along the anti-diagonal direction \(\nu_s+\nu_i\), whereas linear single-photon loss is modeled by filters acting on one frequency variable at a time,
\[
f_{s,i}(\nu_s,\nu_i)=1-\exp\!\left[-\frac{(\nu_{s,i}-\nu_{s,i}^0)^2}{2\sigma_{s,i}^2}\right].
\]
The physical distinction is that two-photon loss carves out correlations along the anti-diagonal of the joint spectrum, whereas single-photon loss removes vertical or horizontal slices [2208.11387][2512.01117].

The spectral geometry of the biphoton also matters independently of the absorber. In semiconductor quantum wells, the Schmidt number
\[
K=\frac{1}{2}\left(\frac{\tau}{T}+\frac{T}{\tau}\right)
\]
is used to quantify entanglement, but the same \(K\) can correspond to frequency-anti-correlated and frequency-correlated states with very different absorption spectra. The reported result is that exciton oscillator strengths are highly increased when photons arrive almost simultaneously in an entangled state, and narrow semiconductor quantum wells make two-photon absorption a highly sensitive probe of photon quantum correlations [1109.6895].

## 3. Measurement ambiguity and the central controversy

The core controversy is that most experiments do not observe the microscopic absorption event directly. In transmission measurements, the direct observable is a reduction in detected photon-pair coincidences after the sample. A decrease in coincidence counts can be caused by true ETPA, but it can also be mimicked by single-photon-loss mechanisms that remove one photon from the pair without a genuine correlated two-photon excitation. Two of the most discussed examples are scattering and hot-band absorption, and this is why a transmission dip alone is not enough to prove ETPA [2208.11387].

Hot-band absorption (HBA) is especially problematic because it is a one-photon process that can produce the same linear power dependence usually taken as evidence for entangled two-photon excited fluorescence. In the mixed regime, the fluorescence can be written as
\[
F = N K \sigma_{\text{HBA}} \phi + \frac{1}{2} N K \sigma_{\text{C2PA}} \phi^2.
\]
Experiments on Rhodamine 6G and LDS798 at \(\sim 1060\) nm found that attenuation of the SPDC beam after generation also gave a linear dependence, with slope \(1.01 \pm 0.03\), whereas true E2PA/E2PEF should cross over toward quadratic scaling under post-generation attenuation. That behavior was interpreted as strong evidence for a one-photon mechanism such as HBA rather than genuine entangled two-photon absorption [2111.05946].

Transmission-based delay experiments have also produced negative results. In Rhodamine B and Zinc tetraphenylporphyrin, deterministic delay between the photons was introduced with a HOM interferometer, with the expectation that at \(\tau=0\) a measurable ETPA signal would appear and at \(|\tau| \gg T_e\) it would vanish. Instead, the corrected \(g^{(2)}\)-based signal fluctuated around zero and showed no systematic suppression at large delay. The interpretation advanced there was that the photon-pair flux typically used in these experiments is not sufficient to promote the two-photon absorption process in these molecules when using transmission readout [2101.10987].

A related null result was reported for indocyanine green in a resonance-enhanced configuration. Broadband entangled near-IR photons with correlation time \(<50\) fs were used, but no measurable r-ETPA signal was observed, leading to an upper bound
\[
\sigma_{\mathrm{ETPA}} \le 6 \times 10^{-23}\ \mathrm{cm}^2/\mathrm{molecule}.
\]
In the same study, a small but statistically significant R6G signal in the SPAD configuration showed linear dependence on both pump attenuation and pairs attenuation, which again indicated one-photon processes such as hot-band absorption rather than genuine ETPA [2312.14382].

These results do not establish that ETPA is absent in principle. They establish that linear scaling, reduced transmission, or weak fluorescence alone are not unique diagnostics.

## 4. Interferometric and spectral certification strategies

Because simple transmission is ambiguous, several certification strategies have been proposed that probe how the sample modifies the biphoton state itself. One line of work uses Hong–Ou–Mandel interference as the sensing element. In this approach, the photons interact with the sample before the HOM beam splitter, and ETPA is inferred from changes in the temporal width and visibility of the HOM coincidence dip. The standard visibility is
\[
V=\frac{R_c(\delta t)_{\max}-R_c(\delta t)_{\min}}{R_c(\delta t)_{\max}+R_c(\delta t)_{\min}}.
\]
In Rhodamine B at \(800\) nm, the HOM-based model treated the sample as a notch filter in the sum-frequency coordinate,
\[
H(\omega_s,\omega_i)=1-\eta\,\exp\!\left[-\frac{(\omega_s+\omega_i-\omega_H)^2}{2\Delta\omega_H^2}\right],
\]
and reported changes in HOM dip width and visibility consistent with that model [2202.05985].

A more restrictive analysis of the same HOM concept emphasized that linear losses do not change HOM visibility in the model; they change count levels, not the normalized dip visibility. This led to the proposal of the visibility ratio \(V_{sam}/V_{sol}\) as a figure of merit. Under the experimental conditions of that work, however, no unambiguous visibility change attributable to ETPA was observed in Rhodamine B, and the model reproduced the data with \(\eta \approx 0\) and frequency-independent loss \(\eta' > 0\) [2210.16500].

A stronger interferometric proposal compares three two-photon beam-splitter geometries: a single-port configuration, a two-port or HOM configuration, and a two-photon N00N-state configuration. For a lossless 50:50 beam splitter, the single-port and two-port coincidence curves depend on the frequency difference \((\nu_i-\nu_s)\tau\), whereas the N00N-state coincidence rate depends on the frequency sum \((\nu_s+\nu_i)\tau\),
\[
R_N(\tau)=\frac14\!\int d\nu_s\,d\nu_i \left[ |\phi(\nu_s,\nu_i)|^2+|\phi(\nu_i,\nu_s)|^2+\phi(\nu_s,\nu_i)\phi^*(\nu_i,\nu_s)+\phi^*(\nu_s,\nu_i)\phi(\nu_i,\nu_s) \right]\{1+\cos[(\nu_s+\nu_i)\tau]\}.
\]
The reported central result is that “the N00N-state configuration is the only one amongst those considered insensitive to linear (single-photon) losses.” In that geometry, the no-filter curve and the single-photon-loss curve are essentially overlapped, while the ETPA curve deviates clearly [2208.11387].

A complementary spectral-certification strategy seeks an asymmetric distortion of the transmitted JSI. The idea is to engineer a source whose input JSI is already asymmetric enough that a two-photon notch filter cuts it unevenly, making the output JSI no longer symmetric under exchange \(\omega_s \leftrightarrow \omega_i\). In a HOM interferometer this lowers the dip visibility,
\[
V=\frac{R_C(\infty)-R_C(0)}{R_C(\infty)+R_C(0)}.
\]
The reported conclusion is that type-II SPDC with a broadband pump is the most promising configuration because its tilted, asymmetric JSI can convert pair-selective absorption into both a measurable spectral distortion and a measurable HOM-visibility reduction [2512.01117].

Spatial correlations provide a further route. By using an SLM to reshape the transverse second-order correlation function \(G^{(2)}\) at fixed pair flux, experiments compared cases in which the photons in a pair overlap spatially with cases in which they are separated by a distance \(\Delta x\). The diagnostic ratio \(\xi(\Delta x)/\xi_0\) was reported to be robust against linear absorption, scattering, and background fluorescence, and to become stronger when the entanglement area was reduced from
\[
A_e = 1.72 \times 10^{-3}\ \mathrm{mm}^2
\]
to
\[
A_e = 69.2 \times 10^{-6}\ \mathrm{mm}^2.
\]
This suggests that correlation-structured measurements can suppress several common artifacts that plague raw transmission or fluorescence observables [2410.06199].

## 5. Material systems and representative results

Organic dyes remain the most studied molecular platforms. In Rhodamine 6G, energy-time entangled photon pairs generated by type-0 SPDC in PPLN produced an ETPA-induced fluorescence signal with a clear linear dependence on photon-pair flux and a strong dependence on inter-photon delay. The fluorescence versus delay showed a Gaussian-shaped peak with
\[
\mathrm{FWHM} \approx 140\ \mathrm{fs},
\]
interpreted as the coherence time of the entangled two-photon wave-packet. The fluorescence rate changed by less than \(1\%\) across the polarization scan, leading to the conclusion that energy-time entanglement, rather than polarization, is the relevant degree of freedom in that geometry. The reported concentration-dependent cross sections were
\[
\sigma_e = (9.9 \pm 4.9)\times 10^{-22}\ \mathrm{cm}^2 \quad \text{at } C = 4.5\ \mathrm{mmol/l},
\]
\[
\sigma_e = (1.9 \pm 0.9)\times 10^{-21}\ \mathrm{cm}^2 \quad \text{at } C = 0.038\ \mathrm{mmol/l},
\]
and for \(C = 110\ \mathrm{mmol/l}\),
\[
Y\sigma_e = (6.4 \pm 0.5)\times 10^{-23}\ \mathrm{cm}^2.
\]
These are among the most widely cited positive molecular results [1910.07346].

Rhodamine B has yielded mixed conclusions. A HOM-based transmission study around \(800\) nm reported that ETPA could be monitored by changes in the temporal width and visibility of the HOM interferogram and extracted
\[
\sigma_e=(1.6978\times10^{-21})\pm1.32\% \; \text{cm}^2/\text{molecule}
\]
without correcting linear losses and
\[
\sigma_e=(5.6874\times10^{-21})\pm1.37\% \; \text{cm}^2/\text{molecule}
\]
after correcting linear losses. Later RhB studies using solvent variation, silica-nanoparticle scattering controls, and HOM-visibility analysis concluded that under the same general transmission conditions ETPA was not unequivocally detected and that linear optical losses could emulate the apparent signal [2202.05985][2310.07810].

Indocyanine green provided a test of resonance-enhanced ETPA in an organic dye. Despite broadband entangled photons centered at approximately \(806\) nm, no measurable r-ETPA signal was found, and the experiment placed the upper bound
\[
\sigma_{\mathrm{ETPA}} \le 6 \times 10^{-23}\ \mathrm{cm}^2/\mathrm{molecule}.
\]
For comparison, the classical resonance-enhanced TPA cross section of ICG at \(800\) nm was measured for the first time as
\[
20(\pm 13)\ \mathrm{GM}.
\]
The interpretation was that a resonant intermediate state did not significantly enhance two-photon processes in ICG under those conditions [2312.14382].

Semiconductor and atomic platforms provide cleaner level structures. In semiconductor quantum wells, the reported result is that frequency-anti-correlated, unentangled, and frequency-correlated biphotons with the same Schmidt number produce different absorption spectra, and that narrow QWs make ETPA a highly sensitive probe of photon quantum correlations [1109.6895]. In cesium atoms, a theoretical study of the \(6^2S_{1/2}\to 8^2S_{1/2}\) transition found that the far-off-resonance approximation gives a constant enhancement factor \(36\pi\), whereas the full treatment without that approximation produces an enhancement factor that oscillates with entanglement time and reaches a maximum value of \(\sim 500\). Under one realistic parameter set the estimated ETPA rate was
\[
\mathcal{R}_{ETPA} \approx 50~\text{Hz}.
\]
This makes alkali atoms a candidate benchmark system for testing ETPA theory without many molecular-system complications [2606.01318].

A distinct solid-state realization has been proposed for monolayer WSe\(_2\). In a frequency-nondegenerate ladder scheme, the Bell-state phase \(\varphi\) of a polarization-entangled photon pair controls the biexciton eigenstate distribution produced by ETPA. The prepared biexciton state is
\[
\ket{\psi_\mathrm{prep}}= \frac{1}{\sqrt{2}}\left(\ket{B_2}+e^{i\varphi}\ket{B_3}\right).
\]
The symmetric Bell state \((\varphi = 0)\) selectively drives bright eigenstates, while the antisymmetric Bell state \((\varphi = \pi)\) drives the exchange-dark eigenstate. No classical polarization source reproduces this \(\varphi\)-dependent eigenstate distribution, and the predicted phase-scan visibility exceeds \(0.97\) for broadband SPDC with \(T_e \sim 100\) fs and high source purity [2605.05633].

## 6. Detectability, bounds, and sensitivity analysis

A recurrent conclusion of the modern literature is that the mere existence of a linear-flux term does not guarantee observability. A quantum-theoretic bound on enhancement found that the entanglement-induced gain factor satisfies
\[
f_{\mathrm{EPP}} \le 2.
\]
For Rhodamine 6G under representative conditions, the estimated single-molecule absorption probability was
\[
P \approx 1.5\times 10^{-24},
\]
the total absorbance in solution was
\[
\approx 1.1\times 10^{-14}\ \text{per mmol},
\]
and the expected fluorescence-detected TPA rate in a \(100\ \mathrm{mM}\) solution was only
\[
8.8\times 10^{-6}\ \mathrm{s}^{-1}
\]
for a pulsed source at \(80\ \text{MHz}\) with \(\epsilon^2=0.1\). Even for a CW scheme with roughly \(10^{13}\ \mathrm{s}^{-1}\) pairs, the estimated absorption rate was only about \(11\ \mathrm{s}^{-1}\). The stated conclusion was that, for ordinary molecular systems and available SPDC sources, detectable ETPA is not feasible unless one simultaneously achieves very high EPP flux, very high molecular concentration, and large classical TPA cross section \(\sigma^{(2)}\) [2103.10530].

Noise-floor analysis reaches a similar conclusion from the measurement side. One spectral-certification study used an SPDC source with
\[
R^{(2)}_{\text{in}}\approx 7.99\times 10^7 \text{ pairs/s}
\]
and estimated a detector fluctuation scale
\[
\delta R_{\text{det}} = 6.322\times 10^3 \text{ pairs/s}.
\]
For a Rh6G estimate \(\sigma_E=1.2\times10^{-25}\,\text{cm}^2\text{molecule}^{-1}\), the predicted absorbed pair rate was
\[
R^{(2)}_{\text{abs}}=333.81 \text{ pairs/s},
\]
far below the noise floor. The inferred minimum detectable values under those conditions were
\[
\eta_E^{\min}=7.91\times10^{-5}, \qquad \sigma_E^{\min}=2.27\times10^{-24}\,\text{cm}^2\text{molecule}^{-1}.
\]
The same analysis concluded that many literature values of \(\sigma_E\) are still too small to be resolved above this threshold [2512.01117].

A general sensitivity framework formulates ETPA detection as a signal-background discrimination problem. If \(S\) is a measurement with maximal correlations and \(B\) a background measurement with correlations reduced while keeping the single-photon flux fixed, detectability requires
\[
S - B \geq u(S) + u(B),
\]
with Poissonian uncertainties
\[
u(S) = n_\sigma \sqrt{S}, \qquad u(B) = n_\sigma \sqrt{B}.
\]
In this model the recorded fluorescence contains four contributions,
\[
N_{rec} = N_{ent} + N_c + N_h + N_{dark},
\]
from desired ETPA, classical two-photon absorption, hot-band or single-photon absorption, and detector dark counts. The resulting sensitivity is expressed as a minimum detectable two-photon cross section in Göppert-Mayer units, allowing direct comparison between experimental designs. The reported practical conclusions were that dark counts matter enormously, higher pair flux helps only up to a point, and the separation method generally outperforms attenuation [2512.19261].

A broader review frames the same problem in terms of competing linear and quadratic channels. Under entangled-photon illumination, the total absorption rate can be written as
\[
R_{\scriptscriptstyle TE} = \sigma_{\!\scriptscriptstyle B}\phi_{\scriptscriptstyle E} + \sigma_{\!\scriptscriptstyle E}\phi_{\scriptscriptstyle E} + \sigma^{(2)}\Gamma \phi_{\scriptscriptstyle E}^2.
\]
Here singleton-induced Boltzmann-tail absorption scales linearly and can easily masquerade as ETPA, while cousin-induced or singleton-pair-induced TPA scales quadratically and becomes dominant when the entangled-photon density is high enough that pairs are no longer isolated. The reported “linear ETPA window” lies between two crossover fluxes,
\[
\phi_{\scriptscriptstyle BC} < \phi < \phi_{\scriptscriptstyle EC},
\]
and it shrinks with one-photon background and optical loss [2604.13375].

## 7. Extensions, applications, and emerging directions

Despite the detection difficulty, ETPA continues to motivate new spectroscopic and materials concepts. One proposal reframes weak delay-dependent transmission traces as a supervised-learning problem. In that approach, a neural network was trained on \(2000\) theoretical ETPA traces, with \(500\) traces per class for systems containing \(1\), \(2\), \(3\), or \(4\) intermediate levels, each trace sampled at \(500\) time points over
\[
\tau \in (-100,100)\,\text{fs}.
\]
For the baseline \(T_e = 63\ \text{fs}\), classification efficiency exceeded \(99\%\) for intermediate-level bands up to \(20\) nm wide. For the shorter entanglement time
\[
T_e = 7.16\ \text{fs},
\]
all mean classification efficiencies exceeded \(99\%\) for all band widths and spacings tested. This suggests that, even when full parameter reconstruction is unrealistic, delay-dependent ETPA traces may still function as informative fingerprints of intermediate-state structure [2501.18534].

Another extension is continuous plasma spectroscopy. A feasibility study proposed using time-frequency entangled photon pairs to drive a two-photon transition in plasma continuously with a lower-intensity continuous-wave laser, writing the rate as
\[
R = \sigma_e \phi + \sigma_c \phi^2.
\]
That proposal emphasized short entanglement time, non-collinear SPDC, and cross-beam spatial localization for Ar-II or Ar-III transitions in a helicon plasma, with the goal of enabling high-bandwidth measurements of localized plasma turbulence or impurity density [2409.08391].

ETPA also serves as a benchmark for nearby quantum-light spectroscopies. A recent microscopic comparison between entangled-photon stimulated Raman scattering and nonlinear absorption in polyatomic molecules concluded that the ESRS spectral-line intensity can be of the same order of magnitude as the one for ETPA, with reported peak-intensity ratios around \(5.60\) and \(6.10\) in pyrene-like conditions. In that treatment, vibrational coherence plays an important role for enhancing ESRS against the ETPA intensity, while ETPA remains the reference low-flux nonlinear absorption process [2601.13646].

Taken together, these developments suggest a shift in emphasis. Early ETPA work often treated linear-in-flux behavior as the central signature. More recent work focuses on state-sensitive witnesses: JSI asymmetry, HOM visibility, N00N-state interferometry, transverse-correlation structuring, eigenstate-selective excitation, and measurement-threshold analysis. A plausible implication is that the long-term significance of ETPA may lie not only in reduced-flux nonlinear spectroscopy, but also in its role as a testbed for how entangled light reshapes selection rules, pathway interference, and observability in complex matter systems.

Source: https://www.emergentmind.com/topics/entangled-two-photon-absorption-etpa