m (→Gallery) |
m (→Antibunching) |
||
Line 7: | Line 7: | ||
for all $\tau$ in a neigborhood of~$\tau=0$. Note that this allows $g^{(2)}(0)$ to be larger than unity and thus to have a super-Poissonian antibunched source, just as one can have sub-Poissonian bunched ones. The confusion (reconciliation?) with $g^{(2)}$ for typical cases is that they could be monotonic (or monotonic enough) to take the case $\tau\to\infty$ where $g^{(2)}(\infty)=1$. We discuss a bit the definition of antibunching for instance in Section~II of Ref. {{onlinecite|lopezcarreno22a}} but one could make a more thorough analysis, extending that of Zou and Mandel.{{cite|zou90a}} | for all $\tau$ in a neigborhood of~$\tau=0$. Note that this allows $g^{(2)}(0)$ to be larger than unity and thus to have a super-Poissonian antibunched source, just as one can have sub-Poissonian bunched ones. The confusion (reconciliation?) with $g^{(2)}$ for typical cases is that they could be monotonic (or monotonic enough) to take the case $\tau\to\infty$ where $g^{(2)}(\infty)=1$. We discuss a bit the definition of antibunching for instance in Section~II of Ref. {{onlinecite|lopezcarreno22a}} but one could make a more thorough analysis, extending that of Zou and Mandel.{{cite|zou90a}} | ||
− | The first antibunching is from Kimble{{etal}}{{cite|kimble77a}} | + | The first antibunching is from Kimble{{etal}},{{cite|kimble77a}} although according to [[Walls]],{{cite|walls79a}} this was also simultaneously reported by [[Leuchs]]{{etal}} who did not publish their findings, but that Walls provides in his review (along with, interestingly, much better results than the ones published by Kimble{{etal}} in their text): |
+ | |||
+ | <center><wz tip="Antibunching by Leuchs et al, unpublished! but reported by Walls.">[[File:Screenshot_20240218_173113.png|250px]]</wz><wz tip="Antibunching by Dagenais and Mandel (without Kimble) and also as reported by Walls, much better than the one they published (with Kimble).">[[File:Screenshot_20240218_173157.png|250px]]</wz></center> | ||
== Gallery == | == Gallery == |
Contents |
Antibunching is the tendency of photons to repulse each other in time, i.e., to be less likely to be found together than later apart in time. At such, this is a particular case of photon statistics, as measured by the so-called $g^{(2)}$. But this is a case of great importance as it signals, or is usually linked, to the quantum rgime. Unfortunately, there have been several differing definitions, based around the idea that antibunching describes single-photons, or suppression of two-photon (or multiphoton) coincidence. In particular, various authors (ourselves included, when convenient) understand antibunching as the condition $g^{(2)}(0)\approx 0$ (or even equal to zero, or less than unity). There is no good term that I know for the latter condition (some speak of "purity", which I also don't like very much; a better term would be sub-Poissonian but that's awkward). A more accurate, widespread, agreed-upon definition of antibunching itself is:
$$g^{(2)}(0)< g^{(2)}(\tau)$$
for all $\tau$ in a neigborhood of~$\tau=0$. Note that this allows $g^{(2)}(0)$ to be larger than unity and thus to have a super-Poissonian antibunched source, just as one can have sub-Poissonian bunched ones. The confusion (reconciliation?) with $g^{(2)}$ for typical cases is that they could be monotonic (or monotonic enough) to take the case $\tau\to\infty$ where $g^{(2)}(\infty)=1$. We discuss a bit the definition of antibunching for instance in Section~II of Ref. [1] but one could make a more thorough analysis, extending that of Zou and Mandel.[2]
The first antibunching is from Kimble et al.,[3] although according to Walls,[4] this was also simultaneously reported by Leuchs et al. who did not publish their findings, but that Walls provides in his review (along with, interestingly, much better results than the ones published by Kimble et al. in their text):
Here follows a collection of antibunching traces $g^{(2)}(\tau)$. It is of course impossible to be comprehensive, but hopefully this will grow to be representative enough of everything and everybody:
From Kimble et al.[3]
From Diedrich et al.[5]
From Basché et al.[6]
From Ambrose et al.[7]
They consider histograms of time-difference so with no normalization of their signal back to unity at long time delays. The identity of this histogram method with a complete~$g^{(2)}$ is valid only over times much smaller than the mean time between detections.[8]
From Lounis et al.[9]
From Fleury et al.[10]
From Kurtsiefer et al.[11]
The bunching elbows are accounted here for the first time with a rate-equation model (yielding a bi-exponential curve).
From Michler et al.[12]
From Zwiller et al.[13]
From Messin et al.[14]
From Hübner et al.[15]
From Ampem-Lassen et al.[16]
From Neu et al.[17]
From Nothaft et al.[18]
Top: raw-data; Bottom: with correction. But their optical pumping provides antibunching similar to the "corrected" electrically-pumped one. This is a log-linear plot, the red line is actually a single exponential.
From Davanço et al.[19]
From Berthel et al.[20]
From Koperski et al.[21]
From Wang et al.[22]
From Boll et al.[23]
From Nahra et al.[24]
From Fiedler et al.[25]
The 'photon bunching at non-zero correlation times' is still attributed to a metastable (shelving) state, like in the original paper.[11]
Something which frequency-filtering does very neatly.
|