Construction of an Entangled Photon Source

[Marilyn67]

TL;DR Summary

An interesting montage made by an electrical and computer engineering student girl

Hello,
Happy New Year 2023 !

Interested in spontaneous parametric down-conversion (SPDC), and the possibility of creating entangled photon pairs, I studied several possible configurations with type 1 and type 2 nonlinear crystals.
When using these crystals, the dominant phenomenon is second harmonic generation (SHG, frequency doubling).

The inverse phenomenon that occurs in parallel (SPDC), the one that interests me, unfortunately has a very low return rate (less than 4 pairs per 1 million...!).

I discovered a very interesting montage made by an electrical and computer engineering student girl :



Can you give me your informed opinion on this assembly?

Is this configuration correct ?

Can it be improved or, on the contrary, simplified to lower the assembly cost ?

(at the moment I'm mainly interested in the first part of the assembly, I mean the production of entangled photon pairs at 810 nm thanks to a source of photons of 405 nm).

Cordially,
Marilyn

 

[Hyperfine]

Is this configuration correct ?

Can it be improved or, on the contrary, simplified to lower the assembly cost ?

(at the moment I'm mainly interested in the first part of the assembly, I mean the production of entangled photon pairs at 810 nm thanks to a source of photons of 405 nm).

The configuration looks fine to me--actually it is quite simple. It could probably be improved upon, but that would almost certainly mean a more complex set of apparatus and higher cost.

Is there something special about the 810 nm and 405 nm wavelengths for what you have in mind?

You might want to take a look at Observation of two-photon emission from semiconductors published in Nature Photonics (2008). The abstract to that paper claims that two-photon production is three orders of magnitude more efficient than SPDC. It is my understanding that with the semiconductor, the two emitted photons may not be of identical wavelength, but are entangled. Almost certainly there is much more literature available since that was published.

 

[Marilyn67]

Hello @Hyperfine,

Thank you very much for your answer.

Your link is very interesting but unfortunately it can't correspond to what I have in mind.

I absolutely need pairs of photons of identical wavelength.

My equipment (transparency) is adapted to 405 nm and 810 nm and these wavelengths being very close to the visible, I can "see" them thanks to CCD adjustment screens.

Regarding the price, I have the following question :

To "maximize" the number of exploitable entangled photons, two crossed crystals are used, and I don't understand why only one would not be enough, as in the first diagram, if the detection is correctly placed at the level of the first cone of light, that from the center with an angle of 3 degrees, where the wavelength of the photons is exactly twice that of the photon pumps.

Can we easily use a single type 1 crystal instead of two crossed type 1 crystals with the same maximization ?

Cordially,
Marilyn

 

[Marilyn67]

Hello,

To "maximize" the number of exploitable entangled photons, two crossed crystals are used, and I don't understand why only one would not be enough, as in the first diagram, if the detection is correctly placed at the level of the first cone of light, that from the center with an angle of 3 degrees, where the wavelength of the photons is exactly twice that of the photon pumps.

Can we easily use a single type 1 crystal instead of two crossed type 1 crystals with the same maximization ?

I found this paper with different configurations of type 1 and type 2 crystals (page 9) :

https://arxiv.org/pdf/2007.15364.pdf

I still don't understand why we must use (to maximize) 2 crossed crystals of type 1, if we correctly identify the cone of entangled photons respecting λ=2λ0 so as not to receive photons of wavelength < or > at 2λ0.

Anyone know why?
In terms of cost, it's twice as expensive...

Cordially,
Marilyn

 

[DrChinese]

I found this paper with different configurations of type 1 and type 2 crystals (page 9) :

https://arxiv.org/pdf/2007.15364.pdf

I still don't understand why we must use (to maximize) 2 crossed crystals of type 1, if we correctly identify the cone of entangled photons respecting λ=2λ0 so as not to receive photons of wavelength < or > at 2λ0.

Anyone know why?
In terms of cost, it's twice as expensive...

You must use 2 Type I crystals if you seek polarization entangled pairs. A single Type I crystal emits 2 entangled photons, but they are not polarization entangled because they always come out in an H>+H> cone. Put a second flush up against the first - rotated 90 degrees - and you get a V>+V>cone. At first glance, you wouldn't think you would get polarization entanglement from this setup. But when the cones overlap such that the "which crystal" paths are indistinguishable, you get the desired polarization entanglement. See figure 2 in reference 1 below. It shows graphically what I was trying to describe. Reference 2 includes a parts list for their setup (designed for a 405nm laser like you want).

https://arxiv.org/abs/quant-ph/0205171
1. Entangled photons, nonlocality and Bell inequalities in the undergraduate laboratory

http://people.whitman.edu/~beckmk/QM/grangier/Thorn_ajp.pdf
2. Observing the quantum behavior of light in an undergraduate laboratory

 

[Hyperfine]

Two quick comments:

In the set up pictured in the initial post, there is a mirror used to redirect the light path to the BBO crystal. I would suggest you use a prism instead of a mirror. That reduces the likelihood of burning an optic.

Scattered light is a total nuisance. As you mentioned above, there are 10⁶ pump photons per entangled pair. Those scattered pump photons will hit your detectors, so be prepared to address that problem.

 

[Marilyn67]

Hello @DrChinese,

Your answer is really very concise and your links are a gold mine of information !
I looked at the diagram you are talking about.
All this is really interesting and deserves that I read these two papers with a clear head.

The list of parts and different distributors you provided me with is invaluable !

I recognized in this list the distributor to which I have already gone (I dared not send a direct link corresponding to certain parts from this supplier because certain links are prohibited on certain forums because of advertising).

What I can say is that some parts are (fortunately) much cheaper today than in 2003 (I was scared when I saw some prices displayed).

I have to think about some things that are not yet clear for the editing I want to achieve, and your two papers will help me a lot !

I will come back to you if I have other important questions to ask you.

Thanks again for everything !

Best regards
Marilyn

 

[Marilyn67]

Hello @Hyperfine,

Thank you for your valuable advice !

Two quick comments:

In the set up pictured in the initial post, there is a mirror used to redirect the light path to the BBO crystal. I would suggest you use a prism instead of a mirror. That reduces the likelihood of burning an optic.

I asked myself the question precisely concerning the power (100 mW is a lot for a 1 mm beam but it seems to me that you need to have this power on the crystal to have a high field and produce non-linear phenomena, no ?)

At the same time, can the crystal support this power, especially if the beam is concentrated by means of a lens ?
I also have other mirrors and 50/50 flat beam splitters (not in prisms).

Scattered light is a total nuisance. As you mentioned above, there are 10⁶ pump photons per entangled pair. Those scattered pump photons will hit your detectors, so be prepared to address that problem.

I take good note of it, thank you!
I take this opportunity to ask your opinion on the weakness of the signals that I will have to visualize :

A digital camera, with its CCD screen is able to "see" the near infrared (810 Nm).
If I consider that my 100 mW is reduced by a factor of 1 million, I drop to 100 nW on the camera.

Do you think the camera can "see" this light?

Shanni Prutchi talks about single photons in the end, but can we already talk about single photons at 100 nW ?

Thank you in advance for your answers

Cordially,
Marilyn

 

[Cthugha]

You might want to take a look at Observation of two-photon emission from semiconductors published in Nature Photonics (2008). The abstract to that paper claims that two-photon production is three orders of magnitude more efficient than SPDC. It is my understanding that with the semiconductor, the two emitted photons may not be of identical wavelength, but are entangled. Almost certainly there is much more literature available since that was published.

Just for the record. That very paper is considered to be questionable by a relevant part of the scientific community working on semconductor quantum optics. No other group has ever been able to reproduce the data on bulk GaAs in the 15 years after the publication and most people assume that the efficiency of the two-photon emission claimed in the manuscript is too large by several orders of magnitude compared to what is is in reality.

 

[Hyperfine]

I asked myself the question precisely concerning the power (100 mW is a lot for a 1 mm beam but it seems to me that you need to have this power on the crystal to have a high field and produce non-linear phenomena, no ?)

At the same time, can the crystal support this power, especially if the beam is concentrated by means of a lens ?
I also have other mirrors and 50/50 flat beam splitters (not in prisms).

You will need power. I would not worry about the BBO crystal as they are often used for harmonic generation in high power, pulsed LASER systems.

A digital camera, with its CCD screen is able to "see" the near infrared (810 Nm).
If I consider that my 100 mW is reduced by a factor of 1 million, I drop to 100 nW on the camera.

Do you think the camera can "see" this light?

A CCD should have no difficulty detecting the 810 nm photons. At least in principle. What I believe you need to be concerned about is the scattered 405 nm photons overwhelming the desired signal. They will be everywhere! There will be many many more of them than the ones you want to detect. One possible approach would be to get a high-quality, narrow-band optical bandpass filter centered on 810 nm. That would be placed in front of your CCD. Yes, it will block some of the 810 nm photons, but it should help a lot in blocking the scattered 405 nm photons. I have used that approach successfully in 2-photon LIF excitation spectroscopy experiments. I know--another added expense!

Folding prisms are cheap--I reiterate my recommendation to use one instead of a mirror.

 

[Marilyn67]

Thanks @Hyperfine for your quick response !

I am reassured as to the resistance of the BBO and for the possible detection at 810 Nm at this emission rate.

I will do my best to receive only entangled photons because of this wide scattering of photons at 405nm, but also all others in the environment.

It will also be necessary to darken the assembly as much as possible and perhaps "cool" it in certain places, but I don't know at what temperature... (not too low I hope..)

Anyway, I will follow your recommendations to protect my optics (dust is also an enemy )

Have a good evening,
Marilyn

 

[Hyperfine]

Anyway, I will follow your recommendations to protect my optics (dust is also an enemy )

Do not get me started on dust!

I would be remiss were I to fail to make the following point. With the very best of intentions, I may have lead you astray as exemplified by the citation corrected by Cthugha. It is difficult for me to understand some instrumental issues when I cannot put my hands on the apparatus itself. Thus, I do not know the full context of the problems under discussion. You need to regard me as an anonymous individual whose credentials and experience are unknown to you. Thus, Caveat lector!

I do hope I have been of some modest assistance, and I most certainly do wish you the best with respect to your endeavors.

 

[Marilyn67]

All the help is precious to me, and I still have tons of questions (I have the list and it isn't finished !)

Have a good day,
Marilyn

 

[Marilyn67]

Hello everybody,
Hello @DrChinese,

I have studied your two papers with great interest.
One week has already passed !
I think they will serve as a base for my assembly but I have a very important question for you !

I thought I would find the answer myself, but I can't.

You must use 2 Type I crystals if you seek polarization entangled pairs. A single Type I crystal emits 2 entangled photons, but they are not polarization entangled because they always come out in an H>+H> cone. Put a second flush up against the first - rotated 90 degrees - and you get a V>+V>cone. At first glance, you wouldn't think you would get polarization entanglement from this setup. But when the cones overlap such that the "which crystal" paths are indistinguishable, you get the desired polarization entanglement. See figure 2 in reference 1 below. It shows graphically what I was trying to describe. Reference 2 includes a parts list for their setup (designed for a 405nm laser like you want).

https://arxiv.org/abs/quant-ph/0205171
1. Entangled photons, nonlocality and Bell inequalities in the undergraduate laboratory

http://people.whitman.edu/~beckmk/QM/grangier/Thorn_ajp.pdf
2. Observing the quantum behavior of light in an undergraduate laboratory

I well understood the need to have photons entangled in polarization in the assemblies we have spoken about.

With simple entangled photons (without worrying about polarization), this could not work because you have to be able to compare them.

In fact I would like to use my source of entangled photons in a variant delayed-choice quantum eraser experiment like this:

https://en.wikipedia.org/wiki/Delayed-choice_quantum_eraser

I noted that in this experiment, one also uses entangled photons in polarization, and I don't understand why simple entangled photons would not do the trick (here moreover, one has orthogonal polarizations with a crystal of the type 2, instead of having identical polarizations with two crossed type 1 crystals...)

Let me explain :

In this experiment, we show (when the paths are indistinguishable, by deleting D3 and D4) that we can reconstruct (a posteriori of course) the two interference signals which are indistinguishable as long as they are not correlated to the signals detectors D1 and D2, OK.

Indeed, without this correlation of measurements carried out by a classic channel between the signal photons and the idler photons, the two interference signals are mixed (phase shifted by π) and give a diffuse signal, (it isn't possible to do the two samplings ) OK.

The reason is that in D1, a path corresponds to two reflections, the other path to a single reflection and a transmission, and the same thing in the opposite direction in D2 : this leads to two interference signals phase shifted by π, OK.

Here's what I don't understand:

In each of the two samplings, we obtain an interference signal, because the idler photons interfere with themselves, OK.

My question :

Why are signal photons and idler photons polarization entangled ?
If the photons are simply entangled (without being polarization entangled), the signal photon must also behave like the idler photon, and participate in an interference signal, right ? (once the sample has been determined of course).

A few weeks ago, I made a Mach-Zehnder interferometer, and I was able to observe these interferences on two screens by placing a lens on each side of the quantum eraser.

My goal is to achieve something similar without a coincidence counter at the start (just patterns on CCD screens), something that would schematically look more like this :

https://fr.wikipedia.org/wiki/Expérience_de_la_gomme_quantique_à_choix_retardé#/media/Fichier:Experience_Sculley.png

Cordially,
Marilyn

 

[Marilyn67]

Hello,

The description of the delayed-choice quantum eraser experiment isn't very explicit on the Wikipedia.

In fact I would like to use my source of entangled photons in a variant delayed-choice quantum eraser experiment like this:

https://en.wikipedia.org/wiki/Delayed-choice_quantum_eraser

I noted that in this experiment, one also uses entangled photons in polarization, and I don't understand why simple entangled photons would not do the trick (here moreover, one has orthogonal polarizations with a crystal of the type 2, instead of having identical polarizations with two crossed type 1 crystals...)

There are more details here :

https://arxiv.org/pdf/quant-ph/9903047.pdf

It is written (page 2, left, 2nd paragraph) :

Argon ion pump laser beam is divided by a double-slit and incident onto a type-II phase matching nonlinear optical crystal BBO (β −BaB2O4) at two regions A and B. A pair of 702.2nm orthogonally polarized signal-idler photon is generated either from A or B region.

Cordially,
Marilyn

 

[DrChinese]

Why are signal photons and idler photons polarization entangled ?
If the photons are simply entangled (without being polarization entangled), the signal photon must also behave like the idler photon, and participate in an interference signal, right ? (once the sample has been determined of course).

I am following your efforts with interest!

I don't want to comment on the DCQE itself, as I find it almost impossible to get the discussion on a single line of inquiry. It's complicated!!

As to your question above: I think you know the answer already. The entangled photons are produced by the BBo crystal, which is specifically cut to enable collection of polarization entangled pairs. To achieve polarization entanglement, the H and V exit cones must overlap so that there is indistinguishability. They are also entangled on other bases (wavelength, momentum, etc.) which is relevant for the DCQE.

Type II:
https://arxiv.org/abs/quant-ph/0101074

 

[Marilyn67]

Hello,

Thanks @DrChinese for quick and detailed reply !

I am following your efforts with interest!

You do me a lot of honor

I don't want to comment on the DCQE itself, as I find it almost impossible to get the discussion on a single line of inquiry. It's complicated!!

I understand.
It's vast, it's still the subject of much debate and the exact description of what I want to do is also complicated to explain on a forum, especially since all the pieces of my puzzle are not yet in place, but it's starting to get clearer...

As to your question above: I think you know the answer already

In this case, I will try to formulate it, hoping not to be mistaken !
I wrote (post #14) :

I well understood the need to have photons entangled in polarization in the assemblies we have spoken about.

With simple entangled photons (without worrying about polarization), this could not work because you have to be able to compare them.

It's the same problem with the DCQE presented on arXiv :

When coincidence counters and therefore SPADs are used, a "usable base" is needed to "sort out" (in this case the polarization).
The "management" of this sorting is complicated, and your new paper dealing with entangled pairs in a type 2 crystal confirms this. (Thanks for this paper !).

First, as I said :

My goal is to achieve something similar without a coincidence counter at the start (just patterns on CCD screens), something that would schematically look more like this :

https://fr.wikipedia.org/wiki/Expérience_de_la_gomme_quantique_à_choix_retardé#/media/Fichier:Experience_Sculley.png

I will not use SPADS to count and compare idler and signal photons :

I want to stay in pure optics (with the eyes on camera) and witness what happens on a CCD screen when I try to make the signal photons coming from the two different paths interfere, whatever the behavior of the idler photons.

As you know, I shouldn't see signal photon interference when I have idler photon interference.

Obviously, I don't want this to be due to bad editing, and I want the editing to be correct...!!!
(I intend to present it, for educational purposes among other things).

Since the photons are supposed to interfere with themselves in each sample, (regardless of their polarization), I'm going to settle for a single type 1 crystal for the SPDC.

In reality, I would need one on each path (the complete source from which I extracted the drawing is in French, sorry) :

https://fr.wikipedia.org/wiki/Expérience_de_la_gomme_quantique_à_choix_retardé

I think I can save a crystal by changing the path of the two beams after the first BS to focus them at two different points on a single crystal (a few mirrors cost less than a BBO crystal ).

The crystal I want to use is this :

https://www.edmundoptics.fr/p/6-x-6-x-05mm-800nm-shg-type-i-bbo-nonlinear-crystal/40609/

(It is originally intended for the SHG but I use it for the SPDC)

I hope that my answer is worthy of your interest in my efforts !

If I make a mistake, please tell me !

(I wonder if the presence of the polarizer on the path of the violet laser diode is useful and if its absence would not allow the gain to be increased...).

I will keep you informed of my progress, in particular on how I will do the two samplings later to observe the two interferences in phase opposition thanks to a classic channel .

Cordially,
Marilyn

 

[Marilyn67]

The crystal I want to use is this :

https://www.edmundoptics.fr/p/6-x-6-x-05mm-800nm-shg-type-i-bbo-nonlinear-crystal/40609/

(It is originally intended for the SHG but I use it for the SPDC)

I have three crystal-related documents that might work for me :

https://www.edmundoptics.fr/document/download/462520
https://www.edmundoptics.fr/document/download/462115
https://cdn.livechat-files.com/api/file/lc/att/5237381/0a377b95cc8819f7a16034e1aaa5355f/087_BB.pdf

The diagram (section A-A) looks weird to me (?)

 

[Marilyn67]

Hello,

In order not to be off-topic, I opened a new thread dedicated to a question that bothers me and that hinders my future studies concerning the behavior of entangled photons in the Mach-Zehnder interferometer that I built there a few weeks ago :

https://www.physicsforums.com/threa...ferometer-but-with-an-entangled-beam.1049772/

Cordially,
Marilyn