Spontaneous parametric down-conversion

Christophe Couteau1

1
University of Technology of Troyes
Light, Nanotechnologies & Nanomaterials (L2n)
ICD CNRS, 12 rue Marie Curie, 10000 Troyes, France.

Spontaneous Parametric Down-Conversion (SPDC), also known as parametric fluorescence, parametric

noise, parametric scattering and all various combinations of the abbreviation SPDC, is a non-linear
optical process where a photon spontaneously splits into two other photons of lower energies. One
would think that this article is about particle physics and yet it is not, as this process can occur fairly
easily on a day to day basis in an optics laboratory. Nowadays, SPDC is at the heart of many quantum
optics experiments for applications in quantum cryptography, quantum simulation, quantum
metrology but also for testing fundamentals laws of physics in quantum mechanics. In this article, we
will focus on the physics of this process and highlight few important properties of SPDC. There will be
two parts: a first theoretical one showing the particular quantum nature of SPDC and the second part,
more experimental and in particular focusing on applications of parametric down-conversion. This is
clearly a non-exhaustive article about parametric down-conversion as there is a tremendous literature
on the subject, but it gives the necessary first elements needed for a novice student or researcher to
work on SPDC sources of light.

1
Even though SPDC has been predicted in the 1960s [1] and first demonstrated experimentally by
Burham and Weinberg (from NASA!) in 1970 [2], this process was really used for practical purposes
only in the late 80s and early 90s. Since then, this process is at the heart of many quantum optics
experiments for applications in quantum cryptography [3], quantum computing [4], quantum
metrology [5] but also simply for testing fundamentals laws of physics in quantum mechanics [6]. This
review article is clearly not exhaustive as a massive literature exists on the subject and cannot be
summarized in 20 pages only. Rather, this article intends to provide the necessary basic formalism and
knowledge there is to know if one wants to work with SPDC and what it is good for. We will not go on
to explain how to build your own SPDC source as plenty of literature already exists on this matter [7,8].
In that context, we will first develop some basic non-linear optics derived from Maxwell equations with
a focus on the exact reverse process of SPDC, namely second-harmonic generation (SHG). Then we will
go on to introduce some quantum optics in a nutshell before going into the theoretical derivation of
SPDC. In the second part, we will develop what SPDC is used for with the notion of correlated photons
[9], indistinguishable photons [10], entangled photons [11] but also how it can be used to engineer a
single photon source [12]. To finish, we will give some examples of how parametric fluorescence was
used for the demonstration of quantum teleportation [13], quantum cryptography [14], CNOT gates
[15], quantum candela [16] or for quantum lithography [17]. We will finish by giving some perspectives
of engineering new sources of SDPC in the hope of increasing the efficiency of this physical effect.

I- Theoretical derivation of SPDC: classical case

Without going into the details of non-linear optics, let us refresh our mind with some concepts [18].
The idea is that when an electromagnetic (em) wave interacts with a medium full of charges, there is
a dipolar-type of interaction between the dipoles (charges) in the medium and the incoming em field
at the frequency ω. The classical harmonic oscillator model holds which is the model of an electron
attached to an atom-core behaving like a spring and thus like an harmonic oscillator interacting with
the incoming wave. We then obtain the standard linear complex refractive index of the medium, which
would depend on ω. But if one starts to increase significantly the energy of the incoming field then this
harmonic oscillator will start to be anharmonic i.e. not respond linearly with the incident
electromagnetic wave. One can simulate this behaviour by the following equation of motion for an
electron in the medium under such em wave by:

𝑑2 𝑥 𝑑𝑥 𝑒
+ 2𝛾 + 𝜔02 𝑥 + 𝑎𝑥 2 = 𝐹 = − 𝐸(𝑡) (1)
𝑑𝑡 2 𝑑𝑡 𝑚

where the first term on the left hand side is the motion acceleration contribution, the second term is
a damping term that tends to slow down oscillations mainly due to non-radiative contribution and the
last two terms on the left side of the equality cause the restoring force of the electrons towards its
initial position. The anharmonicity is contained in the last term (a.x2) which we can neglect in linear
optics. If ignored, then one can see that the dependence of the motion equation is linear with the
position x. This is no longer true at high em power, then the electromagnetic force given by the right-
hand side of equation (1) will induce non-linearities (we note that higher terms than a.x2 can also exist).
Now if one takes two different incoming frequencies ω1 and ω2, one can show that this equation of
motion, thanks to the last quadratic term, will lead to the generation of waves at ω1 and ω2 of course,
but also at ω1 + ω2, ω1 - ω2, 2ω1, 2ω2 and 0 (so-called optical rectification) leading to 6 frequency
components [18]. We can solve this equation by assuming that the non-linear term is a perturbation
case of the linear one in order to end up with these 6 frequencies. A particularly interesting case occurs

2
when ω1=ω2=ω as this will lead to the so-called second-harmonic generation at 2ω, so useful in every
optics laboratory. Higher terms can also exist leading to other non-linear phenomena with more
interacting waves. This classical approach is very useful to understand the physics of the interaction
but is rapidly complex when using the perturbation theory. There is another equivalent way of treating
the problem by starting with Maxwell's equations with the displacement term 𝐷 ⃗ and the magnetic
⃗ in a medium which is not the vacuum. We suppose no external charges, no currents
field strength 𝐻
and we suppose a non-magnetic medium, thus we have the standards relations 𝐷⃗ = 𝜀0 𝐸⃗ + 𝑃⃗ for the
displacement, where 𝑃⃗ is the polarisation vector of the medium, and 𝐵
⃗ = 𝜇0 𝐻
⃗ for the magnetic field.
The polarisation vector 𝑃⃗ accounts for the usual charge displacement in a medium due to an applied
external em field. By playing around with Maxwell's equations, one can show the following known
wave equation:

1 𝜕2 𝐸⃗ 1 𝜕2 𝑃⃗
⃗ 2 𝐸⃗ −
∇ = (2)
𝑐 2 𝜕𝑡 2 𝜀0 𝑐 2 𝜕𝑡 2

with the polarisation 𝑃⃗ = 𝑃⃗𝐿 + 𝑃⃗𝑁𝐿 containing both the linear (L) and the non-linear (NL) terms. SPDC
(and SHG for that matter) is a three-wave interaction only and thus we will only consider the first non-
linear term for the polarisation given by:

(2)
𝑃𝑖𝑁𝐿 = 𝜀0 ∑𝑗 ∑𝑘 𝜒𝑖𝑗𝑘 . 𝐸𝑗 𝐸𝑘 (3)

(2)
with {𝑖, 𝑗, 𝑘} = {𝑥, 𝑦, 𝑧}, meaning that the second order term 𝜒𝑖𝑗𝑘 is actually a tensor and is called the
second order nonlinear susceptibility of the medium. Figure 1-a) represents the Feynman’s diagram of
the SPDC process where the photon γ3 at energy ℏ𝜔3 splits into two twin photons (γ2 and γ1) at
energies ℏ𝜔2 and ℏ𝜔1 . SHG would be the exact time-reverse process. Even though very odd looking at
a first glance, SPDC does follow energy conservation and momentum conservation rules so that:

ℏ𝜔3 = ℏ𝜔1 + ℏ𝜔2 (4)

𝑘⃗3 = 𝑘⃗2 + 𝑘⃗1 and ∆𝑘⃗ = 𝑘⃗3 − 𝑘⃗2 − 𝑘⃗1 = ⃗0 (5)

These two equations are called the phase-matching conditions and will be discussed in details later on
as they are central for the mechanism to happen. These conditions are represented in Figure 1-b
(energy conservation) and Figure 1-c (momentum conservation). Now one can solve the wave equation
from Eq. (2) with the usual assumptions where we consider a lossless medium, collimated and
monochromatic input beams, continuous excitation, normal incidence and we will neglect the double
refraction. We will also make a less-reasonable assumption which is to be in the non-depleted regime
meaning that the incoming beam can give rise to SHG (reverse diagram than the one presented in
Figure 1-a) but without significant energy loss for the excitation beam or so (the so-called pump beam,
with 2 photons at frequency ω coming from the same pump laser in this case). This assumption can be
quite wrong as the incoming beam can be more or less completely depleted for SHG if one is careful
with phase-matching conditions.

3
Figure 1: a) Feynman’s diagram of SPDC where one photon at energy ℏ𝝎𝟑 splits into two twin photons at
energies ℏ𝝎𝟐 and ℏ𝝎𝟏 and b) representation of energy conservation of SPDC, c) is the representation of
momentum conservation of SPDC and d) is a schematic of SPDC where a UV pump laser generates red SPDC
photons coming out of the crystal in a cone shape.

For SPDC on the other hand, we are (unfortunately) completely in the regime of no-pump depletion as
parametric down-conversion is an extremely inefficient process. Under such assumptions, one can
solve the equation (2) by looking for a solution of the form:

𝐸3 (𝑧, 𝑡) = 𝐴3 𝑒 𝑖(𝑘3 𝑧−𝜔3 𝑡) + 𝑐. 𝑐. (6)

‘c.c.’ being the complex conjugate term and k3=n3.ω3/c and 𝑛32 = 𝜀 𝐿 (𝜔3 ) where L stands for the linear
part only. We thus look for a solution for the polarisation of the form:

𝑃3 (𝑧, 𝑡) = ℘3 . 𝑒 𝑖(𝑘3 𝑧−𝜔3 𝑡) + 𝑐. 𝑐. (7)

with

℘3 = 4𝜀0 𝑑𝑒𝑓𝑓 𝐸1 𝐸2 (8)

E1 and E2 being the amplitudes of the em fields 1 and 2. 𝑑𝑒𝑓𝑓 = 1/2𝜒𝑒𝑓𝑓 is a complicated but known
parameter depending on the strength of the non-linearity of the material and on geometrical factors
all of which are properties of the material (as an example, for BBO type II, we have 𝑑𝑒𝑓𝑓 =
(𝑑11 sin(3𝜙) + 𝑑22 cos(3𝜙))cos(2𝜃)). One can then find the following equation from equations (2-
3) and (6-8):

𝑑 2 𝐴3 𝑑𝐴3 4𝑑𝑒𝑓𝑓 𝜔3
𝑑𝑧 2
+ 2𝑖𝑘3 𝑑𝑧
=− 𝑐2
𝐴1 𝐴2 𝑒 𝑖(𝑘1 +𝑘2 −𝑘3 )𝑧 (9)

4
This last equation can be further reduced by assuming the slow varying amplitude approximation and
thus neglect the first term compared to the second term (also a standard approximation in non-linear
optics) on the left-hand side of this equation. It is then straightforward to find the solution of such
equation under the non-depleted pump approximation by integration between 0 and L (length of the
crystal) and considering that 𝐼𝑖 = 2𝑛𝑖 𝜀0 𝑐|𝐴𝑖 |2 is the optical intensity of beam 𝑖:
2
8𝑑𝑒𝑓𝑓 𝜔32 𝐼1 𝐼2 2 sin ∆𝑘.𝐿/2 2
𝐼3 = 𝑛1 𝑛2 𝑛3 𝜀0 𝑐 3
𝐿 ( ∆𝑘.𝐿/2 ) = 𝐼3𝑚𝑎𝑥 . 𝑠𝑖𝑛𝑐 2 (∆𝑘. 𝐿/2) (10)

with the so-called phase-matching parameter ∆𝑘 = 𝑘1 + 𝑘2 −𝑘3 (we use scalar notations for easier
use). One can see from equation (10) that ideally, one would want to have ∆𝑘~0 in order to have a
quadratic increase (first order Taylor expansion of the sinc function) of the SHG intensity as a function
of the length of the crystal. Otherwise, the sinc function ensures a periodic energy exchange between
the em pump field (the incoming beam) and the created waves. This phase-matching condition is
crucial for an efficient non-linear effect and accounts for the wave vector conservation of the process.
In order to get the phase-matching condition right, one needs to have:
𝑛3 𝜔3 𝑛1 𝜔1 𝑛2 𝜔2
𝑘1 + 𝑘2 = 𝑘3 = = + (11)
𝑐 𝑐 𝑐

which is an impossible condition to be satisfied with most materials as 𝑛1 (𝜔1 ) < 𝑛2 (𝜔2 ) < 𝑛3 (𝜔3 )
for 𝜔1 < 𝜔2 < 𝜔3 , i.e. for ‘normal’ dispersion. This is where birefringent crystals are used as they
possess two (uniaxial) or three (biaxial) different refractive indices along different symmetry axes
which can satisfy the phase-matching condition. In the particular case of second-harmonic generation,
it is fairly easy to have 𝑛𝑒 (𝜔1 ) = 𝑛𝑜 (𝜔3 = 2𝜔1 ) with o and e the so-called ordinary and extraordinary
axes for uniaxial non-linear crystals.

SPDC is the reverse process of sum-frequency generation (SFG which is non-degenerate SPDC, i.e.
different wavelengths for the two output photons) or SHG (degenerate SPDC, i.e. same wavelengths
for the two output photons) which would correspond to the process called parametric amplification in
classical non-linear optics. This process occurs when a strong pump beam at 𝜔3 is combined with a
weak signal beam at 𝜔1 to give rise to an amplified signal at 𝜔1 but also to a second signal called the
idler beam at 𝜔2 . Using the same type of equation than the one derived in (9), we have this time:
𝑑𝐴1 𝜔 (2)
𝑑𝑧
= 𝑖 𝑛 1𝑐 𝜒𝑒𝑓𝑓 𝐴3 𝐴∗2 𝑒 𝑖Δ𝑘.𝑧 (12)
1

𝑑𝐴2 𝜔2 (2)
=𝑖 𝜒 𝐴 𝐴∗ 𝑒 𝑖Δ𝑘.𝑧 (13)
𝑑𝑧 𝑛2 𝑐 𝑒𝑓𝑓 3 1

These equations leading to the so-called Manley-Rowe relations can be solved supposing that Δ𝑘 = 0
and lead to:

𝜔 𝑛 𝐴
𝐴2 (𝑧) = 𝑖 √𝜔2 𝑛1 |𝐴3 | 𝐴1∗ (0) sinh(𝛼𝑧) (14)
1 2 3

𝐴1 (𝑧) = 𝐴1 (0) cosh(𝛼𝑧) (15)

(2)
𝜒𝑒𝑓𝑓 |𝐴3 | 𝜔2 𝜔1
𝛼= 𝑐
√𝑛 (16)
2 𝑛1

5
From these solutions of the parametric amplification process, we can clearly see that if there is no
initial incident power at frequency 𝜔1 , i.e. if 𝐴1 (0) = 0, then the signal and the idler beams would
not exist and SPDC would not be possible. In other words, classical non-linear optics “does not allow”
spontaneous parametric down-conversion to exist. Clearly, we see the limitations of the classical
treatment of non-linear optics in order to derive SPDC. For that, a quantum description of the
phenomena is needed [19,20].

II- Theoretical derivation of SPDC: quantum case

In order to do so, let us recall some basic quantum optics properties. Quantum optics is the realm of
electromagnetism and Maxwell's equations when one deals with single photons. Indeed, a process
such as SPDC tells you that spontaneously, a photon pair can suddenly 'appear' somewhere in the
universe ex-nihilo. Once again, energy and momentum conservation hold but nevertheless, a photon
in a particular optical mode can spontaneously be created. In order to account for such phenomena
(and many others in quantum optics), one needs to quantise the electromagnetic field. This article is
certainly not about that and one can find excellent textbooks on the matter [21]. We will just give a
quick overview of what is required to know. The general principle is to start with Maxwell's equations
and use a spatial Fourier expansion of the em field in order to solve these equations. It is also wiser to
work with the vector potential 𝐴 rather than directly with the electric field 𝐸⃗ . If one works in the so-
called Coulomb gauge where ⃗∇. 𝐴 = 0 (which tells you that photons are transverse!) then one can end
up with an equation that looks very much like the one of an harmonic oscillator with the derivation of
conjugate canonical variables just like in the case of a quantised harmonic oscillator. It is then
straightforward to apply the rules of a quantum harmonic oscillator but this time for the vector
potential of the electromagnetic field. By doing so, one can show the following relation:

ℇ ⃗ ⃗
𝐴(𝑟) = ∑𝑙 𝜖𝑙 𝜔𝑙 (𝑒 𝑖𝑘𝑙 .𝑟 𝑎̂𝑙 + 𝑒 −𝑖𝑘𝑙 .𝑟 𝑎̂𝑙† ) = 𝐴(+) (𝑟) + 𝐴(−) (𝑟) (17)
𝑙

with the sum over the 𝑙 optical modes accessible in this universe (quite a few then) and the two
quantum operators, 𝑎̂ being the photon annihilation operator (accounting for the disappearance of a
photon) and 𝑎̂† the photon creation operator (accounting for the appearance of a photon). By the
same token, 𝐴 is also an operator. 𝜖𝑙 is the vector direction of the em field associated to these photons
and ℇ𝑙 is the amplitude of the associated classical field with energy ℏ𝜔𝑙 equals to that of a single
photon in the 𝑙 mode. 𝑎̂ and 𝑎̂† are the equivalent creation and annihilation operators of a quantum
of energy for an harmonic oscillator, allowing to go up and down in the energy ladder of the oscillator,
which in the case of quantum optics, is called a photon. We have then the following relations:

𝑎̂𝑙† |𝑣𝑎𝑐⟩ = |1𝑙 ⟩ (18)

𝑎̂𝑙 |1𝑙 ⟩ = |𝑣𝑎𝑐⟩ (19)

where a single photon in mode 𝑙 is created from the vacuum state |𝑣𝑎𝑐⟩ and a single photon in mode
𝑙 “disappears” for instance due to absorption from a single dipole. One can show that the quantised
free radiation Hamiltonian would look like:

𝐻 ̂𝑙 + 1) = ∑𝑙 ℏ𝜔𝑙 (𝑎̂𝑙† 𝑎̂𝑙 + 1)

̂𝑅 = ∑𝑙 ℏ𝜔𝑙 (𝑁 (20)
2 2

6
Where 𝑁 ̂𝑙 is the photon number operator giving rise to the number 𝑁𝑙 of photons in the
electromagnetic mode 𝑙. Within this context, quantum optics allows and explains perfectly the
possibility of spontaneously creating a photon of a certain energy in a given optical mode. Thus a
photon of high energy can split into two (or more but less likely) photons of lower energies that were
not previously around when the interaction with the non-linear medium takes place. In order to
describe the phenomena, we can infer an effective interaction Hamiltonian for SPDC which would look
like:

̂𝑆𝑃𝐷𝐶 = 𝑖ℏ𝜅 (𝑎̂1 𝑎̂2 𝑎̂3† 𝑒 𝑖∆𝑘⃗.𝑟−𝑖∆𝜔.𝑡 + 𝑎̂1† 𝑎̂2† 𝑎̂3 𝑒 −𝑖∆𝑘⃗.𝑟+𝑖∆𝜔.𝑡 )
𝐻

⃗ ⃗
= 𝑖ℏ𝜅 (𝑎̂𝑖 𝑎̂𝑠 𝑎̂𝑝† 𝑒 𝑖∆𝑘.𝑟−𝑖∆𝜔.𝑡 + 𝑎̂𝑖† 𝑎̂𝑠† 𝑎̂𝑝 𝑒 −𝑖∆𝑘.𝑟+𝑖∆𝜔.𝑡 ) (21)

with the first term of the Hamiltonian accounting for SFG or SHG (if 𝜔1 = 𝜔2) and the creation of a
SHG photon at 𝜔3 = 2𝜔2 with ∆𝜔 = 𝜔3 − 𝜔1 − 𝜔2 . The second term of the Hamiltonian accounts for
SPDC where one photon (the pump beam) 'disappears' to create one photon at 𝜔1 and one photon at
𝜔2 in the non-degenerate case where the two frequencies are different. We can clearly see in (21) that
SPDC and SHG do compete and do exist at the same time. In most of what follows in this article, we
will focus on degenerate SPDC with 𝜔1 = 𝜔2 thus 'giving birth' to twin photons in this case. Note that
from now on, we will use the indices {𝑠, 𝑖, 𝑝} rather than {1,2,3} when talking about the three-wave
mixing phenomenon SPDC. We have 𝑝 for the incoming pump photon at high energy and the twin
photons 𝑖 for idler and 𝑠 for signal (note that these names are given more for historical reasons). We
must stress that the constant 𝜅 is given by:

2 𝑑𝑒𝑓𝑓 𝜔𝑝 𝜔𝑠 𝜔𝑖
𝜅=3 𝜀0 𝑉
√ 2𝜀 𝑉 (22)
0

with the usual meanings for 𝑑𝑒𝑓𝑓 , V and the 3 frequencies. Now in a more concrete way we can
estimate what happens to an incoming pump beam with 𝑁𝑝 number of pump photons impinging on a
non-linear crystal by looking at the effect of the SPDC Hamiltonian on the incoming state |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩
describing 𝑁𝑝 pump photons and 0 photon in the signal mode and 0 photon in the idler mode. We use
the Schrödinger equation in time and look for the state:
1 𝑡 ′ )𝑑𝑡 ′
̂
|𝜓(𝑡)⟩ = 𝑒 𝑖ℏ ∫0 𝐻𝑆𝑃𝐷𝐶 (𝑡 |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩ (23)

𝜅 is rather small so it means that SPDC is very inefficient and that most of the pump beam is
unperturbed by the SPDC process. We can thus make a Taylor expansion of the exponential
Hamiltonian up to the first order and find that:
1 𝑡
|𝜓(𝑡)⟩ ≈ 𝐶0 |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩ + 𝐶1 ∫ 𝐻̂ (𝑡 ′ )𝑑𝑡 ′ |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩ + ⋯ (24)
𝑖ℏ 0 𝑆𝑃𝐷𝐶

where 𝐶𝑖 are the coefficients of the Taylor expansion present for the normalization of the wave
function. In the case of 1st order, we only have 𝐶0 and 𝐶1 ≈ 1 − 𝐶0 again for the sake of working with
normalised states. As we are looking for perfect phase-matching, including in energy too, then we have
∆𝜔 ≈ 0 thus the integral term in (24) is a Dirac function and we can apply the operators 𝑎̂𝑖† , 𝑎̂𝑠† , 𝑎̂𝑝 on
the incoming state |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩ and find:

7
 |𝜓(𝑡)⟩ = 𝐶0 |0𝑠 , 0𝑖 , 𝑁𝑝 ⟩ + 𝜅𝐶1 𝑒 −𝑖∆𝑘⃗.𝑟 |1𝑠 , 1𝑖 , 𝑁𝑝 − 1⟩ (25)

In order to bring out a true parallel with the SHG case (from Equation (10)), we must recall that the
measured em intensity 𝐼𝑆𝑃𝐷𝐶 will be proportional to the square of the wave function from (25). Thus
the phase matching condition in wave-vector does appear (like in SPDC). Unlike SHG, SPDC is a linear
function with the pump power which is not trivial to see at first when looking at (25). For that, we need
to go back to (21) and assume that the pump power is so strong that it can be considered like a classical
em field 𝑎̂𝑝 ≈ 𝐸𝑝 and as a result, 𝐼𝑆𝑃𝐷𝐶 will be proportional to 𝐼𝑝 only, unlike the SHG phenomena from
(10). For SPDC, one pump photon only splits into two photons but only one photon in each mode so
the dependence is clearly linear even though it is a non-linear process. We also recall that 𝐶0 ≫ 𝐶1 .
This is actually one of the major issues of SPDC, it is the fact that SPDC is a very inefficient process
which hinders many applications in quantum information and quantum cryptography for instance. We
also assumed that we only have the first 2 terms in the Taylor expansion of (24) but if 𝑎̂𝑝 ≈ 𝐸𝑝 is strong
enough (sometimes necessary if one wants to get enough SPDC effect) then higher order terms appear.
This expansion is thus a Poisson distribution that characterizes SPDC and thus not an on-demand
supply of photon pairs but rather a probabilistic one. From (25), we see that for a certain probability
to have one pair (proportional to 𝜅 2 ) we have a non-zero chance to have two pairs (with probability to
be 𝜅 4 /4) when one add the second order term etc… In other words, most of the incoming state (the
pump beam) is left untouched and only a small portion (10-5 to 10-12) goes into producing a twin pair
of down-converted photons. We also see that the double-pair generation increases if the pump power
is increased too much and that is not something one would necessarily want for most experiments
using single twin photon pairs. Finally, we must stress that in formula (25), we ‘increase’ the probability
of being in the |1𝑠 , 1𝑖 , 𝑁𝑝 − 1⟩ by performing a so-called coincidence measurement between the idler
and the signal photon in a quantum optics experiment. Practically, we look at a double-detection event
on two detectors (one pointing towards the idler photons and the other one towards the signal
photons) in a very short time-scale of 10 ns or so. One can show that accidental events in such a short
time window are very low and thus a coincidence event corresponds to the detection of a pair of SPDC
photons. This is the practical way of getting rid of the vacuum state, or rather the initial state
|0𝑠 , 0𝑖 , 𝑁𝑝 ⟩, where we collapse the wave function unto the one-and-more-pair-coincidence state, thus
getting rid of the contribution from this dominant non-creation of pairs state. Mandel et al. also
demonstrated that the number of events for parametric down-conversion has to the same as for the
pump photons. In other words, one pump photon might split into two SPDC photons but the number
of pair-event created has to be at the most the number of incoming photons, assuming every photon
is converted and no higher order occur. It is obvious experimentally but not trivial. Finally, unlike the
reverse case of SHG which requires a non-zero ‘seed’ signal 𝐴1 (0) in order for any SPDC photons to
exist, thanks to the properties of the creation operators, we can start from zero photons in the idler
and signal states and create them using the Hamiltonian in (21). To wrap up on the theory of SPDC, we
can claim that the fact that only quantum formalism can explain the existence of SPDC photons, is
actually one of the many peculiar properties of parametric fluorescence and that what makes it unique
for applications in quantum optics as we will learn later on.

III- Practical case: BBO crystal for SPDC

Practically in the laboratory there are many aspects that should be taken care of in order to generate
'efficient' SPDC in an optical set-up. First of all, the phase-matching condition Δ𝑘 = 0 has to be fulfilled
as much as possible. As it was mentioned earlier it is quite unlikely to find a natural material which

8
fulfill the condition 𝑛𝑖 (𝜔𝑖 ) = 𝑛𝑠 (𝜔𝑠 ) < 𝑛𝑝 (𝜔𝑝 ) with 𝜔𝑖 = 𝜔𝑠 < 𝜔𝑝 except when using birefringent
crystals which may possess two (uniaxial crystals) or three (biaxial crystals) different indices for a given
wavelength. Another important point is that we want non-collinear phase-matching leading to
different wave vectors directions for the three waves involved thus more easily selected thereafter
(see schematic Fig. 1-d). This makes the theory more difficult but ensures at the end the spatial
separation of the twin photons. Let us take the most common non-linear crystal used for SPDC up to
now: barium borate 𝛽 − 𝐵𝑎(𝐵𝑂2 )2 or BBO. It is a negative (meaning that the direction having a low
refractive index is the fast axis. At right angles to it is the slow axis, with a high index of refraction,
𝑛𝑒 < 𝑛𝑜 ) uniaxial crystal (meaning it has an optical axis OA) with two different indices 𝑛𝑒 and 𝑛𝑜 for
extraordinary and ordinary indices. BBO is transparent from 190 nm to 3300 nm and with indices given
by the Sellmeier equations which are, for example, for type I down-conversion (𝑒 ↦ 𝑜 + 𝑜):
0.0184
𝑛𝑜 = 2.27405 + 𝜆2 −0.0179 − 0.0155𝜆2 (33)

0.0128
𝑛𝑒 = 2.3730 + 𝜆2 −0.0156 − 0.0044𝜆2 (34)

with 𝜆 given in 𝜇𝑚. We recall the relation of the extraordinary index of refraction with the angle 𝜃
from the optical axis is given by the ellipsoid equation for indices:

𝑛2 𝑛2
𝑛2 (𝜃) = 𝑛2 cos2 𝜃+𝑛
𝑒 𝑜
2 sin2 𝜃 (35)
𝑒 𝑜

So typically, when the pump beam enters the crystal with a certain angle 𝜃 with the optical axis OA,
the extraordinary index will be different for different angles, as opposed to the ordinary index. For
BBO, phase-matching conditions occur with an extraordinary beam only and one can have two types
of phase-matching:

𝑒 ⟼𝑜+𝑜 type I

𝑒 ⟼𝑒+𝑜 type II

𝑒 ⟼𝑜+𝑒 type II

Figure 1-d represents a type II configuration. Figure 2 represents the two curves for the Sellmeier
equations of BBO with 𝑛(𝜃) for 𝜃 = 90° (in black) and 𝑛𝑜 (in red) as a function of the wavelength 𝜆.
We will take the simple case where we have collinear phase matching and with a pump at 400 nm and
hope to get a pair of 800 nm photons. The pink curve is 𝑛(𝜃 = 31.7°) where we can see that for a
pump beam at 400 nm (very common in the laboratory) we can have phase matching at 800 nm (the
two grey circles) where we fulfill a condition where 𝑛𝑒 (𝜔𝑠 ) = 𝑛𝑜 (2𝜔𝑝 ). This is the phase matching
condition for this particular case at these particular wavelengths for type I and in a collinear
configuration. This simple case is for illustration but the most general case cannot be evaluated without
the help of numerics. The idea is to solve equation (5) in a vector form |∆𝑘⃗ = 𝑘⃗𝑝 − 𝑘⃗𝑠 − 𝑘⃗𝑖 | = 0. In
order to do so, one needs to work in spherical coordinates and determine 9 parameters to fulfill the
phase matching conditions given by the equations (4) and (5): 3 parameters for the wavelength 𝑖, 𝑠, 𝑝,
3 angles 𝜃 for each wave and 3 angles 𝜑 for each wave. Clearly one can set 𝜆𝑝 , 𝜆𝑠 , 𝜆𝑖 for the 3 waves if
we want to work at certain wavelengths and in the degenerate case. We can also decide to set 𝜃𝑠 for
instance (and not be in the collinear case such as in Fig. 1-d) and we can show that for a uniaxial crystal,

9
Figure 2: Graphs representing the Sellmeier equations of BBO with ne(θ) for θ=90° (in black) and no (in red) as
a function of the wavelength λ. The pink curve is n e(θ=31.7° ). We have phase-matching between 400 nm and
800 nm (the two grey circles) when we fulfill a condition where ne (ωs )=no (2ωp). This is the phase matching
for SHG condition for this particular case at these particular wavelengths for a type I non-linear BBO crystal
and in a collinear configuration.

Figure 3: Graph representing the two outcoming cones of SPDC for a type II BBO crystal. This is an experimental
transverse cut of SPDC at 810 nm behind a BBO crystal. The two circles represent the two extraordinary and
ordinary beams. In this particular case, we can observe two intersection points (where polarization
entanglement occurs). We can also observe that one circle is slightly bigger than the other one due to crystal
tilting and meaning that the two wavelengths for signal and idler are slightly different.

10
there is no dependence with 𝜑𝑝 and that there is the relation 𝜑𝑠 = 𝜋 + 𝜑𝑖 . We are finally left with 3
parameters (𝜃𝑝 , 𝜃𝑖 , 𝜑𝑠 ) to determine using equations (4) and (5). For more information in the
calculation part, see the article from NIST by Boeuf et al. from 2000 [22]. The following website
maintained by NIST: spdcalc.org allows you to enter your own parameters and gives you all sorts of
information such as the phase-matching curves or the joint spectral distribution. In particular, one
derives from these calculations that for a type II SPDC in BBO with a pump laser at 405 nm, giving rise
to 810 nm down-converted photons, we have an angle of about 3° between the idler and the signal
waves coming out of the crystal in a cone shape (see Figure 1-d for a schematic). Figure 3 presents an
experimental result of the double cone emission of SPDC in type II BBO. In this case, we have two
different circles as opposed to the type I case. Figure 3 represents the notorious picture of the
parametric down-converted cones coming out of the crystal with photons with orthogonal
polarisations for the two different cones (extraordinary e/V cone and ordinary o/H photons). This is an
experimental transverse cut of SPDC at 810 nm behind a BBO crystal done by scanning across a 2D
plane parallel and away from the nonlinear crystal using an optical fibre linked to a single photon
detector. We will see later on that it can lead to polarisation entangled pairs of photons. For that, one
will need to take two things into account that are inherent to SPDC: the temporal walk-off and the
spatial walk-off. For the temporal walk-off, also known as the longitudinal walk-off, it simply shows
that as the crystal has two indices along which the 'e' and the 'o' photons follow, they will travel
differently (different group velocities) and be temporally different when coming out of the crystal (the
length of the crystal is typically on the order of 1 mm). One photon will come out of the crystal before
the other one. This temporal difference is given by:

1 1
Δ𝑡 = |𝐿𝑐 (𝑛 − 𝑛 )| (36)
𝑒 𝑜

The spatial walk-off, or transverse walk-off is due to the fact again that the material is birefringent and
as such the energy propagation or the Poynting vector is not necessarily in the same direction as the
wavevector of the down-converted photons. This is the case for the ordinary wave o but not for the
extraordinary wave e and we thus end up with an angle 𝜌 between the two waves at the output of the
crystal given by:
1 𝑑𝑛𝑒
𝜌 = −𝑛 (37)
𝑒 𝑑𝜃

Once again, this will be very important for the production of entangled photons as one must not have
any kind of information (temporal or otherwise) in order to produce pure entanglement. There are
easy ways to avoid these ‘labelling’ by using compensating crystals (see [8] for instance). One more
thing we should emphasize is the fact that for spontaneous parametric down-conversion, the
conservation of momentum has an uncertainty to it (equation (5)) and we have rather:

𝑘⃗3 = 𝑘⃗2 + 𝑘⃗1 + Δ𝑘⃗ (38)

and the equation still holds. That is the reason why, these two cones actually form two rings when one
makes a transversal view-cut along the propagation direction of the SPDC photons (see Fig. 3). One
way to overcome this situation is to select spectrally with narrow-band filters (from 1 to 10 nm) the
down-converted photon pairs, on top of a spatial (with angles) and temporal (with coincidence
detections) selection [11,23].

11
 IV- SPDC used for applications
Single-photon source
The first striking effect with SPDC is the fact that we end up with twin photons created at the same
time and which are indistinguishable. One foresees that this could be useful for some low-signal level
communications and was predicted by Mandel and Rarity (c. 1984) as a joint measurement can ensure
their 'twin-ness'. Derive from that, the first application was the localisation of a one-photon state using
SPDC by [24]. This led to heralded single photon sources using SPDC providing strong photon
antibunching which is again a clear quantum property of light. This is good for quantum information
processing (QIP) although more for proofs of principle. Real single photon sources such as quantum
dots, NV centers, atoms, etc... are still preferable. Figure 4-a is an example of an heralded single photon
source where the authors used this time collinear source of orthogonal twin photons that were
separated in polarisation [12]. One photon was detected (idler photon in Fig 4-a) and served to herald
the arrival of the signal photon on the other side which was then detected using a standard Hanbury-
Brown and Twiss (HBT) interferometer known in quantum optics for the determination of single
photons. Fig 4-b presents the photon antibunching characteristics of this system as a function of the
time coincidence window for the detection. We can show that the quality of this autocorrelation g(2)
degrades as the intensity of the pump increases and that is due to the fact that the emission of
spontaneous parametric down-conversion follows a Poisson statistics as discussed previously with
equation (24). Thus SPDC has a limited use as a source of single photons as multiple pair creation is
always present.

Source of indistinguishable photons

In 1987, the group of Leonard Mandel realised a two-photon interferometry and observed this time a

Figure 4: a) Example of an heralded single photon source using SPDC from a periodically poled KTP crystal
(PPKTP). The crystal produces a pair of collinear photons with orthogonal polarisations which are first
separated by a polarizing beamsplitter (PBS), so that the idler photon serves as an heralding event for the
(𝟐)
signal photon sent onto a non-polarising beamsplitter (BS) for a measurement of the 𝒈𝒄 function (so-called
(𝟐) (𝟐)
HBT). b) 𝒈𝒄 (𝝉) function for 3 different coincidence windows showing 𝒈𝒄 (𝝉) ≈ 0.07 for the lowest
coincidence window [12].

12
Figure 5: a) Experimental set-up of the first two indistinguishable photon experiment pioneered by Hong, Ou
and Mandel [10]. This was done by doing SPDC in a KDP in a non-collinear phase-matching condition. The BS
is moved to change the delay between the arrival times of the photons onto the BS. b) Coincidence
measurements of the photon pairs detected by D1 and D2 as a function of the path delay at the BS with a clear
dip observed.

photon bunching effect. This experiment consisted in creating two twin photons by SPDC and then
made them interfere on a beamsplitter (cf Figure 5). Quantum-mechanically, one can show that the
two outcomes where the photons come out in different ports interfere destructively and remain only
the cases where the photons come out together from the BS, they bunch. This shows bosonic
properties for photons due to the fact that the quantum operators of annihilation and creation of
photons are used (oppositely, fermions like electrons would go away from each other). The photons
are said indistinguishable as they must be from the same optical mode in order for the interferences
to occur and this is called the Hong-Ou-Mandel effect [10].

Entangled photons
Following these experiments, researchers used parametric down-converted photons for creating
entangled photons in polarization for violating Bell’s inequalities. Figure 6 presents the original article
where they managed, using a type-II BBO crystal, to create two intersections of the output cones by
tilting their BBO crystal (see also Figure 3). They then selected the intersection of the two parts of the
cones of the twin photons with orthogonal polarisations. As with the right conditions it is impossible
to tell which photon is which anymore, you cannot predict the result of the measured polarisation of
the photon at one intersection. What you do know though is the fact that parallel measurements will
give you perfect anti-correlation in polarisation. Figure 1-d is a schematic of the two output cones from
idler and signal and Figure 3 provides experimental data where we have a transverse cut of the two
twin beams produced. At the intersection, one ends up with the following entangled state:
1
|𝜓⟩ = (|𝐻𝑖 ⟩|𝑉𝑠 ⟩ + 𝑒 𝑖𝜙 |𝑉𝑖 ⟩|𝐻𝑠 ⟩) (39)
√2

where H is for horizontal polarization and V is for vertical polarisation, s and i stand for signal and idler
like before and φ is a phase that can be modified. Figure 6-a represents a schematic of the experiment
carried on by Kwiat et al. [11], while Fig 6-b shows a diagram of the geometry of the process. Figure 6-

13
c demonstrates clearly that there are strong correlations in polarization between the two entangled
photons. This entangled state and this technique was used to violate Bell’s inequalities and brought up
a lot of discussions regarding the so-called Einstein-Podolsky-Rosen (EPR) paradox.

Quantum teleportation
One of the most famous use of entanglement is certainly when it was applied to demonstrate quantum
teleportation. The idea of teleportation is that two protagonists (usually called Alice and Bob) share a
common pair of EPR particles (in this case photons) that they will use in order to teleport a quantum
state (which does not have to be known) from Alice to Bob. Figure 7-a shows the general principle of
creating a pair of entangled/EPR photons (2 and 3) distributed between Alice and Bob and Alice uses
her photon 2 to make a so-called Bell state measurement (BSM) with the photon to be teleported
(photon 1) [13]. Using a classical channel of information, Alice tells Bob what BSM-type of
measurement she made (but not the result) so that Bob can act accordingly on his photon n°3 and
recover the quantum state of the teleported photon 1. Figure 7-b represents the experimental set-up
developed in 1997 at the University of Innsbruck. A first pair of entangled photon is created by a pump
beam where photon 3 will be going to Bob and photon 2 will be going to Alice. Alice then makes a BSM
on a beamsplitter with photon 1 to be teleported which is in fact coming from the creation of another
pair of EPR photons by sending back the pump beam into the crystal. This time one photon of the pair
is photon 1 and the other one will be used to herald the arrival of photon 1 for Alice to make the BSM.
We note that in order to synchronise the whole process, the authors use a pulse excitation with a well-
known arrival time for each time (within the pump pulse) and photon 4 is used to herald the arrival of
photon 1 to be teleported. Finally, the equivalent classical communication between Alice and Bob is
done by the fact that they have to communicate together to see when they made a coincidence
measurement. Using a similar set-up, entanglement swapping was also realized [25] and now people
try to store single photons and entangled photons into quantum memories from specific materials
[26].

Figure 6: a) Set-up of the first experiment of two entangled photon experiment done by Kwiat et al. where a
BBO crystal is pumped at 351 nm for SPDC in a non-collinear critical phase-matching condition. b) Schematic
of the famous two SPDC cones coming off the crystal with orthogonal polarisations and spatial selection to
select only the photons at the intersection. c) Visibility measurements leading to a Bell parameter S > 2.5 and
proving that the photons are indeed entangled [11].

14
Figure 7: a) Principle of the teleportation experiment where we send an initial state 1 and Alice proceed in
doing a Bell-state measurement (BSM) with one entangled photon from the EPR source (2). This initial state is
the one to be teleported. The other entangled photon of the EPR pair (3) is sent to Bob who receives a classical
information from Alice onto what measurement he should do in order to recover Alice’s initial state, thus
effectively teleporting the initial state 1 to state 3 from Alice to Bob. b) Schematic of the experiment where a
strong pulse of pump photons first creates an EPR entangled pair (2 and 3) to share between Alice and Bob.
This pump pulse of light is then sent back again into the non-linear crystal to create another pair of SPDC
photon (1 and 4) where one photon of the pair will be the initial state to teleport (path 1) and the other one
will be used as a trigger to create the heralded photon source (4) [13].

Quantum cryptography
This teleportation experiment and some others at the time certainly sparked a lot of interest in the
broad field of quantum information. In particular, it was shown earlier on that using entanglement and
the no-cloning theorem in quantum mechanics, one could make a very secure protocol based on

15
quantum cryptography (at least on paper). The basic idea proposed is that the polarisation state of a
photon (say horizontal or vertical) can be thought of as a quantum bit of information, or a qubit. Thus
a pair of EPR photons can represent a two-qubit source. Figure 8-a and –b presents two experiments
realising quantum cryptography using entangled photons. In Figure 8-a, the authors use polarisation
entangled photons [14] but in Figure 8-b, the group from Nicolas Gisin in Geneva used the so-called
time-bin entanglement based on James-Franson’s idea from 1991 [27]. In this case, the idea is simply
to create twin photons (say A and B) and to send each of them in a Mach-Zehnder interferometer with
a long (L) and a short path (S). Then if one gets a coincidence measurement, assuming that beforehand

Figure 8: a) Schematic of the quantum cryptography protocol using polarisation entangled photons using the
SPDC method of Kwiat et al. (Figure 6). Each photon from a pair is sent to Alice and Bob where they verify that
the photons are indeed entangled in order to use them as an encryption key. H polarization can be the “+1”
bit of information and V polarization can be the “-1” bit of information [14]. b) Quantum cryptography was
also realised in telecommunication fibres where polarization is not preserved. For that, Tittel et al. used the
so-called time-energy entanglement or time-bin entanglement to generate entangled photons [28].

16
we use a pulse pump beam itself in a superposition of L and S paths, there is no way to tell whether
the photons took both the short paths or the long paths [28]. All we know is that they have both taken
the same path in order to get a coincidence measurement. This very powerful method of entangling
photons allowed the Geneva group (and others thereafter) to use non-degenerate SPDC to get one
photon at the pair at, say 900 nm in the silicon detection window, and a photon at 1550 nm [29]. This
wavelength is not chosen randomly as it corresponds to the wavelength used in the world
communication network thus bringing quantum cryptography into the scene of a mature
application.Of course, science is not that easy and other problems have been encountered like
inefficient detectors, problem with maintaining stable interferometers etc… The carrier of information
has been tested on optical tables but also in fibers (visible or telecom) and in free-space. Now there
are projects to point to satellites to do satellite communication quantum cryptography [30].

Quantum computation
Following these remarkable advances in quantum communication, it did not take long before SPDC
was used for running experiments on quantum computation. This is still a very much on-going task
where groups around the world demonstrated basic algorithms, error-correction codes and more using
entangled photons from SPDC. SPDC can only generate two quantum bits (qubits) at the time (at least
in polarisation) and people have been working actively in order to get 2, 3 up to 10 pairs of entangled

Figure 9: Number of polarisation entangled photons in time using SPDC. The black squares are experimental
results with the reference and the red line is linear fit. At this linear rate, we will reach 20 entangled photons
in 2037.

17
Figure 10: a) Schematic of the CNOT gate realized in [15]. b) Polarisation-encoded photonic qubits can
be converted into spatially encoded qubits, suitable for the gate shown in a) and c) schematic of the
experimental CNOT gate. Pairs of energy degenerate photons are incident from the left of the
diagram. They then follow the gate procedure in the middle before being detected on the right-hand
side [15].

photons in a given experiment [11,31,32,33,34,35,36]. Figure 9 plots the number of polarisation

entangled photons created by SPDC as a function of time. At this linear rate, we will reach 20 entangled
photons in 2037. Unfortunately, the trend does not follow Moore’s law and it seems pretty clear it
may even reach a plateau at some point using this technique. Nevertheless, it is still primordial to show
basic quantum information protocols using SPDC and entangled photons. Also shows the need for new
sources of entangled photons.The first real practical implementation of an optical gate was realized in
2003 with entangled photons [15]. This was the first experimental all-optical quantum controlled-NOT
gate (CNOT gate). This experiment and many others followed the so-called KLM proposal (published
by E. Knill, R. Laflamme and G. Milburn in 2001 and it was the first article published in Nature of the
twenty first century [37]. The KLM proposal showed that using non-deterministic gates, single photons,
efficient and discriminating single-photon detectors, it was possible to perform an all-optical computer
using only linear optics such as beamsplitters, wave-plates etc… Figure 10-a presents the basic principle
of this CNOT gate with a control qubit and a target qubit. Each qubit can have a 0 or a 1 state and the
CNOT gate tells you if the control qubit is in the 0 state, then the state of the target state is unchanged
whereas if the control qubit is in the 1 state, then the target state will have its state flip from 0 to 1 or
from 1 to 0. This scheme used by O’Brien et al. was a coincidence basis gate which is non-deterministic
[15]. Fig 10-b represents the so-called dual-rail encoding where spatial superposition with a
beamsplitter is realised in order to implement this scheme. This, combined with entangled photons in
polarization and conditional detections showed that the gate was realized 1/9th of the time. Since
then, this scheme was improved. Using SPDC, it has been used to implement basic concepts in
quantum processing through the proof of CNOT gates [15], Shor's algorithm [38], one-way quantum

18
computation [39] or boson sampling type of gates [40]. There is a wide literature on the subject and it
is still very much on-going [4].

Quantum metrology and more

SPDC allows also other types of less-obvious applications. One of them is the use of entanglement in
order to perform super-resolution phase measurements [17]. The basic idea is to create a so-called
NOON state written as |𝑁, 0⟩ +|0, 𝑁⟩ where N is the number of photons in a given mode. In this case,
using interferometry, one can show a phase resolution given by ∆𝜑 = 1/√𝑁. The concept of ‘quantum
lithography’ was thus born although there is still much more improvement to be done before this
concept can actually be put in practice. One example was to use entangled photons for measuring
protein concentration [41].

Another idea that came out fairly earlier on when spontaneous parametric down-conversion photons
were characterised, is the idea that one could use twin photons for metrology. The concept of quantum
metrology uses light for setting the standards that define units of measurement (the candela) for light.
The quantum candela project aims at developing standards for photon metrology from the signal level
of existing radiometric standards and reach the smallest grains of light which are single photons.
Through photon correlation measurements, it was shown that one could calibrate very accurately
photons detectors, again a very important application for radiometric metrology [16,5,42].

Finally, we must mention that originally, parametric fluorescence and entangled photons generated
with them where used for testing the very foundations of quantum mechanics such as the wave-
particle duality, non-locality via the violation of Bell’s inequalities where in 2015, 3 experiments finally
closed all the loop-holes associated to these tests and put an end to hidden variables introduced by
John Bell in the 60s. Two of them used SPDC for that purpose [43,44]. Born’s rule of quantum
mechanics was also tested with an heralded single-photon source using SPDC going through a three
slit interferometer [6]. This was done in order to show that quantum interference is a two-amplitude
effect and that no three-amplitude term exists in quantum mechanics.

V- Future for SPDC

The future for SPDC is still very bright and nowadays, SPDC is used as a tool more than anything, for
various applications and we can classify the applications into three main categories: a-foundations of
physics and in particular of quantum mechanics, b-quantum information processing and c-quantum
metrology via the quantum candela project and quantum sensors. Within QIP, we can split into 3 sub-
categories: quantum cryptography, quantum computing and quantum communications. There are
other secondary potential applications for optical lithography but the odds that any application will
come out of this are pretty low, quantum imaging, also very narrow field and quantum sensing which
has a higher potential.
Now the challenge of tomorrow is to actually engineer new technologies to create efficient SPDC and
not just using the natural crystal birefringence. Several groups are attempting to do so by several
means, either by using confined light into optical fibres [45], in SOI waveguides [46] but also by careful
engineering of semiconductor Bragg-type of structure in order to obtain parametric down-conversion
directly on an optical chip [47]. By the same token, using two four-wave mixing processes, Silverstone
et al. managed to create SPDC photons directly on a chip made of silicon-on-insulator [46]. A direct 3-
photon has also been shown using cascaded SPDC [48]. Recently, an electrically injected photon-pair
at room temperature was even demonstrated with is the first evidence of SPDC obtained by electrical

19
pumping rather than optical pumping [49]. More recently, we should mention the very interesting
work on enhancing SPDC effect using non-linearities in metamaterials [50]. As such, it is safe to say
that research in SPDC has thus a bright future ahead.

ACKNOWLEDGEMENTS
The author would like to point out that most of this work was done when the author was at the
University of Oxford from 2000 to 2002 and as such, would like to thank, Dik Bouwmeester, John Rarity,
Christoph Simon, Antia Lamas-Linares, Gabriel Durkin and John Howell but also Gregor Weihs, Chris
Erven, Rolf Horn and Kevin Resch for fruitful and sometimes very entangled discussions. The author
thanks L. Le Cunff for Figure 1-d and C. Altuzarra and S. Vezzoli for working on the data of Figure 3.

REFERENCES:
[1] Quantum Fluctuations and Noise in Parametric Processes I. W. H. Louisell, A. Yariv, & A. E.
Siegman, Phys. Rev. 124, 1646 (1961).
[2] Observation of simultaneity in parametric production of optical photons pairs. D. C. Burnham, &
D. L. Weinberg, Phys. Rev. Lett. 25, 84 (1970).
[3] Quantum cryptography. N. Gisin, G. Ribordy, W. Tittel, & H. Zbinden, Rev. Mod. Phys. 74, 145
(2002).
[4] Optical quantum computing. J. L. O’Brien, Science 318, 1567 (2007).
[5] The quantum candela: a re-definition of the standard units for optical radiation. J. Y. Cheung, C. J.
Chunnilall, E. R. Woolliams, N. P. Fox, J. R. Mountford, J. Wang, & P. J. Thomas, J. Mod. Opt. 54, 373
(2007).
[6] Ruling out multi-order interference in quantum mechanics. U. Sinha, C. Couteau, T. Jennewein, R.
Laflamme, & G. Weihs, Science 329, 418 (2010).
[7] Entangled photon apparatus for the undergraduate laboratory. D. Dehlinger, & M. W. Mitchell,
Am. J. Phys. 70, 898 (2002).
[8] Entangled photons, nonlocality, and Bell inequalities in the undergraduate laboratory. D.
Dehlinger, & M. W. Mitchell, Am. J. Phys. 70, 903 (2002).
[9] Measurements of time delays in the parametric production of photon pairs. S. Friberg, C. K. Hong,
& L. Mandel, Phys. Rev. Lett. 54, 2011 (1985).
[10] Measurements of subpicosecond time intervals between two photons by interference. C. K.
Hong, Z. Y. Ou, & L. Mandel, Phys. Rev. Lett. 59, 2044 (1987).
[11] New high-intensity source of polarization-entangled photons pairs. P. G. Kwiat, K. Mattle, H.
Weinfurter, A. Zeilinger, A. V. Sergienko, & Y. Shih, Phys. Rev. Lett. 75, 4337 (1995).
[12] Coherence measures for heralded single-photon sources. E. Bocquillon, C. Couteau, M. Razavi, R.
Laflamme, & G. Weihs, Phys. Rev. A 79, 035801 (2009).
[13] Experimental quantum teleportation. D. Bouwmeester, J.-W. Pan, K. Mattle, M. Eibl, H.
Weinfurter, & A. Zeilinger, Nature 390, 575 (1997).
[14] Quantum cryptography with entangled photons. T. Jennewein, C. Simon, G. Weihs, H.
Weinfurter, & A. Zeilinger, Phys. Rev. Lett. 84, 4729 (2000).
[15] Demonstration of an all-optical quantum controlled-NOT gate. J. L. O’Brien, G. J. Pryde, A. G.
White, T. C. Ralph, & D. Branning, Nature 426, 264 (2003).
[16] Absolute self-calibration of the quantum efficiency of single-photon detectors. X.-H. Chen, Y.-H.
Zhai, D. Zhang, & L.-A. Wu, Opt. Lett. 31, 2441 (2006).

20
[17] Super-resolving phase measurements with a multiphoton entangled state. M. W. Mitchell, J. S.
Lundeen, & A. M. Steinberg, Nature 429, 161 (2004).
[18] Nonlinear optics and quantum electronics. M. Schubert, & B. Wilhelmi, John Wiley and Son
(1986).
[19] Theory of parametric frequency down conversion of light. C. K. Hong, & L. Mandel, Phys. Rev. A
31, 2409 (1985).
[20] Theory of two photon entanglement in type-II optical parametric down-conversion. M.H. Rubin,
D.N. Klyshko, Y.H. Shih, & A.V. Sergienko, Phys. Rev. A 50, 5122-5133 (1994).
[21] The quantum theory of light. R. Loudon, Oxford University Press 3rd Edition (2000).
[22] Calculating characteristics of noncollinear phase matching in uniaxial and biaxial crystals. N.
Boeuf, D. Branning, I. Chaperot, E. Dauler, S. Guérin, G. Jaeger, A. Muller, & A. Migdall, Opt. Eng. 39,
1016 (2000).
[23] High efficiency entangled photon pair collection in type II parametric fluorescence. C. Kurtsiefer,
M. Oberparleiter, & H. Weinfurter, Phys. Rev. A. 64, 023802 (2001).
[24] Experimental realization of a localized one-photon state. C. K. Hong & L. Mandel, Phys. Rev. Lett.
56, 58 (1986).
[25] Experimental Entanglement Swapping: Entangling Photons That Never Interacted. J.-W. Pan, D.
Bouwmeester, H. Weinfurter, and A. Zeilinger, Phys. Rev. Lett. 80, 3891 (1998).
[26] Preparation and storage of frequency-uncorrelated entangled photons from cavity-enhanced
spontaneous parametric downconversion. H. Zhang, X.-M. Jin, J. Yang, H.-N. Dai, S.-J. Yang, T.-M.
Zhao, J. Rui, Y. He, X. Jiang, F. Yang, G.-S. Pan, Z.-S. Yuan, Y. Deng, Z.-B. Chen, X.-H. Bao, S. Chen, B.
Zhao, & J.-W. Pan, Nature Photonics 5, 628 (2011).
[27] Bell inequality for position and time. J. D. Franson, Phys. Rev. Lett. 62, 2205 (1989).
[28] Quantum Cryptography Using Entangled Photons in Energy-Time Bell States. W. Tittel, J. Brendel,
H. Zbinden, & N. Gisin, Phys. Rev. Lett. 84, 8737 (2000).
[29] High-quality asynchronous heralded single-photon source at telecom wavelength. S. Fasel, O.
Alibart, S. Tanzilli, P. Baldi, A. Beveratos, N. Gisin, H. Zbinden, New J. of Phys. 6, 163 (2004).
[30] Experimental quantum key distribution with simulated ground-to-satellite photon losses and
processing limitations. J.-P. Bourgoin, N. Gigov, B. L. Higgins, Z. Yan, E. Meyer-Scott, A. K. Khandani,
N. Lütkenhaus, and T. Jennewein, Phys. Rev. A 92, 052339 (2015).
[31] Observation of three-photon Greenberger–Horne–Zeilinger entanglement. D. Bouwmeester, J.-
W. Pan, M. Daniell, H. Weinfurter, A. & Zeilinger, Phys. Rev. Lett. 82, 1345 (1999).
[32] Experimental demonstration of four-photon entanglement and high-fidelity teleportation. J.-W.
Pan, M. Daniell, S. Gasparoni, G. Weihs, & A. Zeilinger, Phys. Rev. Lett. 86, 4435 (2001).
[33] Experimental demonstration of five-photon entanglement and open-destination teleportation.
Z. Zhao, Y.-A. Chen, A.-N. Zhang, T. Yang, H. J. Briegel & J.-W. Pan, Nature 430, 54 (2004).
[34] Experimental entanglement of six photons in graph states. C.-Y. Lu, X.-Q. Zhou. O. Guhne, W.-B.
Gao, J. Zhang, Z.-S. Yuan, A. Goebel, T. Yang, J.-W. Pan, Nature Phys. 3, 91 (2007).
[35] Observation of eight-photon entanglement. X.-C. Yao, T.-X. Wang, P. Xu, He Lu, G.-S. Pan, X.-H.
Bao, C.-Z. Peng, C.-Y. Lu, Y.-A. Chen, & J.-W. Pan, Nature Photonics 6, 225 (2012).
[36] Experimental ten-photon entanglement. X.-L. Wang, L.-K. Chen, W. Li, H.-L. Huang, C. Liu, C.
Chen, Y.-H. Luo, Z.-E. Su, D. Wu, Z.-D. Li, H. Lu, Y. Hu, X. Jiang, C.-Z. Peng, L. Li, N.-L. Liu, Y.-A. Chen, C.-
Y. Lu, J.-W. Pan, Phys. Rev. Lett. 117, 210502 (2016).
[37] A scheme for efficient quantum computation with linear optics. E. Knill, R. Laflamme, & G. J.
Milburn, G. J., Nature 409, 46 (2001).

21
[38] Experimental realization of Shor's quantum factoring algorithm using qubit recycling. E. Martín-
López, A. Laing, T. Lawson, R. Alvarez, X.-Q. Zhou, & J. L. O'Brien, Nature Photonics 6, 773 (2012).
[39] Experimental one-way quantum computing. P. Walther, K. J. Resch, T. Rudolph, E. Schenck, H.
Weinfurter, V. Vedral, M. Aspelmeyer, & A. Zeilinger, Nature 434, 169 (2005).
[40] Experimental boson sampling. M. Tillmann, B. Dakić, R. Heilmann, S. Nolte, A. Szameit, & P.
Walther, Nature Photonics 7, 540 (2013).
[41] Measuring protein concentration with entangled photons. A. Crespi, M. Lobino, J. C. F.
Matthews, A. Politi, C. R. Neal, R. Ramponi, R. Osellame, & J. L. O’Brien, Appl. Phys. Lett. 100, 233704
(2012).
[42] Advances in quantum metrology. V. Giovannetti, S. Lloyd, & L. Maccone, Nature Photonics 5, 222
(2011).
[43] Significant-Loophole-Free Test of Bell’s Theorem with Entangled Photons. M. Giustina, M. A. M.
Versteegh, S. Wengerowsky, J. Handsteiner, A. Hochrainer, K. Phelan, F. Steinlechner, J. Kofler, J.-Å.
Larsson, C. Abellán, W. Amaya, V. Pruneri, M. W. Mitchell, J. Beyer, T. Gerrits, A. E. Lita, L. K. Shalm, S.
W. Nam, T. Scheidl, R. Ursin, B. Wittmann, & A. Zeilinger, Phys. Rev. Lett. 115, 250401 (2015).
[44] Strong Loophole-Free Test of Local Realism. L. K. Shalm, E. Meyer-Scott, B. G. Christensen, P.
Bierhorst, M. A. Wayne, M. J. Stevens, T. Gerrits, S. Glancy, D. R. Hamel, M. S. Allman, K. J. Coakley, S.
D. Dyer, C. Hodge, A. E. Lita, V. B. Verma, C. Lambrocco, E. Tortorici, A. L. Migdall, Y. Zhang, D. R.
Kumor, W. H. Farr, F. Marsili, M. D. Shaw, J. A. Stern, C. Abellán, W. Amaya, V. Pruneri, T. Jennewein,
M. W. Mitchell, P. G. Kwiat, J. C. Bienfang, R. P. Mirin, E. Knill, & S. W. Nam, Phys. Rev. Lett. 115,
250402 (2015).
[45] Optical-Fiber Source of Polarization-Entangled Photons in the 1550 nm Telecom Band. X. Li, P. L.
Voss, J. E. Sharping, & P. Kumar, Phys. Rev. Lett. 94, 053601 (2005).
[46] On-chip quantum interference between silicon photon-pair sources. J. W. Silverstone, D.
Bonneau, K. Ohira, N. Suzuki, H. Yoshida, N. Iizuka, M. Ezaki, C. M. Natarajan, M. G. Tanner, R. H.
Hadfield, V. Zwiller, G. D. Marshall, J. G. Rarity, J. L. O'Brien & M. G. Thompson, Nature Photonics 8,
104 (2014).
[47] Monolithic Source of Photon Pairs. R. T. Horn, P. Abolghasem, B. J. Bijlani, D. Kang, A. S. Helmy,
& G. Weihs, Phys. Rev. Lett. 108, 153605 (2012), Phys. Rev. Lett. 108, 153605 (2012).
[48] Direct generation of three-photon polarization entanglement. D R. Hamel, L. K. Shalm, H. Hübel,
A. J. Miller, F. Marsili, V. B. Verma, R. P. Mirin, S. W. Nam, K. J. Resch, & T. Jennewein, Nature
Photonics 8, 801 (2014).
[49] Electrically Injected Photon-Pair Source at Room Temperature. F. Boitier, A. Orieux, C. Autebert,
A. Lemaître, E. Galopin, C. Manquest, C. Sirtori, I. Favero, G. Leo, & S. Ducci, Phys. Rev. Lett. 112,
183901 (2014).
[50] Quantum nonlinear light emission in metamaterials: broadband Purcell enhancement of
parametric downconversion. A. Davoyan & H. Atwater, Optica 5, 608 (2018).

22