Thursday, June 10, 2021

A Stunning New Neutral D Meson Oscillation Anomaly

Shorter Summary 

For the first time ever, an experiment at the Large Hadron Collider (LHC) has seen a highly statistically significant (albeit slight) difference in mass between a particle and its antiparticle, contrary to the Standard Model. This has suddenly risen to the most statistically significant anomaly in all of high energy physics.


It isn't clear why this is the case, or whether there is any good reason to suspect underestimated systemic error (other than the fact that it contradicts the Standard Model and wasn't strongly predicted by any of the front runner beyond the Standard Model physics theories currently in circulation as viable proposals in light of other HEP experimental data).

If this results is independently replicated by another experiment (since the LHCb operates at lower energies than the ATLAS and CMS experiments, there are other colliders in the world that can do so), this will be a very big deal, probably implying "new physics" of some undetermined nature. 

But, despite the high statistical significance of the error, because the discrepancy is so small in both absolute and relative terms, it is easy to imagine that an overlooked source of systemic error in the measurement that could resolve the anomaly on highly technical grounds (although I have no specific ones in mind).

Is This A Discovery Yet?

While "5 sigma" statistical significance is the standard for making a new discovery in high energy physics, this isn't the only requirement. 

The preprint results still need to be peer reviewed, the 5 sigma observation needs to be independently replicated, and the proponents of new physics based upon the observation need to propose some theory to explain that result that is consistent with the rest of the laws of physics supported by empirical observations in contexts where anomalies aren't seen. 

This long road is a bit like the Roman Catholic church's arduous process for declaring someone to be a "saint." But this is enough to get the ball rollings towards a widely recognized beyond the Standard Model physics result that all credible high energy physicists would have to accept and reckon with somehow.

The New Results In Context

A Dº meson has two valence quarks, a charm quark and an anti-up quark, bound by gluons. The Dº mass is about 1865 MeV. Its antiparticle, anti-Dº meson also has two valence quarks, an anti-charm quark and an up quark. Since it is neutral, the electromagnetic charge of the particle and the antiparticle is the same. Both the particle and its antiparticle are pseudoscalar (i.e. spin-0 odd parity) mesons.

The particle and its antiparticle can't oscillate in "tree level" (i.e. single step Feynman diagram) processes. But, they can oscillate via a number of two step processes.

For example, the charm quark can decay to a strange quark which can decay to an up quark, while the anti-up quark can become an anti-strange quark which can become an anti-charm quark, in each case, via two rounds of simultaneous W boson transitions, one W- and one W+ each, and those four weak force bosons can be virtual ones that cancel each other out.

Because it can happen, the Standard Model predicts that it will happen, so Dº-anti-Dº meson oscillation while nice to observe to confirm that hypothesis, is no big deal.

The probability of going from a Dº to an anti-Dº meson and the probability of going from an anti-Dº meson to a Dº meson are not identical due to the CP violating phase in the CKM matrix. So, the observed oscillating Dº-anti-Dº pair, isn't exactly a 50-50 mix of each of them, although the difference between what was observed and a 50-50 mix wasn't statistically significant (as predicted in the Standard Model given the precision of the measurements done).

But, as a PDG review paper explains, in the Standard Model, the mass and decay width of the Dº meson and the mass and decay width of the anti-Dº meson should be identical.

preprint of a June 7, 2021 letter from the LHCb experiment, however, concludes that there is a small, but still 7.3 sigma mass difference (without considering look elsewhere effects) between a Dº meson and an anti-Dº meson, the first time ever that a particle and its antiparticle of any kind have been observed to have statistically significantly different masses, and a statistically significant 3.1 sigma difference in decay widths.

Normalized by the decay width, the mass difference is 0.397% of the decay width, and the decay width difference is 0.459% of the average decay width.  The average decay width is less than 2.1 MeV. So, the observed discrepancy between the masses of the Dº and anti-Dº meson is on the order of 0.008 MeV or less, and the observed discrepancy between the decay widths of the Dº and anti-Dº meson is on the order of 0.009 MeV or less. 

In both absolute terms (less than 0.2% of the electron mass), and relative to the mass of the Dº meson (on the order of two or three parts per million), these differences are very small. 

But, as noted above, the differences are reported to be statistically significant despite the fact that in the Standard Model they shouldn't occur at all. 

Look elsewhere effects probably reduce the differences in decay widths to a non-statistically significant level of 2 sigma or less because over the years probably something like a thousand or more HEP matter-antimatter mass comparisons have been done, all producing null results. But, even with look elsewhere effects, the locally 7.3 sigma mass difference should be at least 4-5 sigma, and it is also worth noting that the mean percentage mass difference in reported in this preprint does replicate the prior world average for this measurement (which previously was "marginally compatible with" no mass difference due to a much larger experimental uncertainty in previous measurements).

The Letter contains no meaningful or insightful commentary regarding what could be leading to this bombshell conclusion.


Onur Dincer said...

So, the observed discrepancy between the masses of the Dº and anti-Dº meson is on the order of 0.008 MeV or less, and the observed discrepancy between the masses of the Dº and anti-Dº meson is on the order of 0.009 MeV or less.

I assume there is a typo in this sentence.

andrew said...

Indeed. The second sentence should say "decay width" and not mass. I'll fix it.

DDeden said...

I'm stunned all right.

Tom Andersen said...

Thanks for pointing this out!

If there are 1000 other experiments that get null, does that really translate into ~5sigma? I get 1000 as 4 sigma, leaving a 3 sigma discrepancy. But its likely my math is wrong.

Also systemics seem hard, but I guess the other 1000 experiments with null results actually help the systemics enormously.

Mitchell said...

Please correct this...

Graham Dungworth said...

"this will be a very big deal, probably implying "new physics" of some undetermined nature. ".
Forget the esoteric physics and try a thermodynamics and kinetic approach.
There are two types of particle here; you deal with an up quark and a charm quark along with the "bonds that bind them. A chemist might call them different elements bonded together.
There a reaction between them that is reversible and kinetically controlled. Is the reaction favoured , that's the thermodynamic control.
With just one particle and its antiparticle one may get a 50/50 mix; a chemist would consider its chirality. For such a reaction the free energy change is zero but the reaction kinetics still occurs. In this case there is an enthalpic change given by g deltaH =T*deltaS where S is the entropy change.
Now if there are 2 species one can get 4 different products and there is a Gibbs Free energy change as well as enthalpic energies. The consequence is that one does not expect a 50/50 mix.
In practice there is a spatial reorganisation occurring. Call it a spontaneous breaking or phase change; the reordering involves say for a chemist a bond is broken and the system eqilibrates by moving from a chiral symmetry to a planar symmetry; it drops a spatial dimension and takes on mass; a negligeable amount for a chemist, a rotational or vibrational degree of freedom. The bond will reform.
So- for this 1865 MeV meson what is its half-life 10^-8 to 10^-16 sec ?
I'd guess it's a 1st order process- we need a rate constant for the kinetics. Also the Arrhenius function requires a pre-exponential factor- it's probably within an order of 10^-13 second.
The pair production temperature for these mesons at 1.865GeV is ca. 4.33*10^12Kelvin. A lot of other products could be formed at such a temperature!
The particles may be relativistic in velocity but not light speed so parity doesn't double the complexity.
The free energy difference delta E is 0.008 MeV ie. tiny but measured.

Many years ago the presence of amino acids in the Miller Urey experiment raised the question of whether the chiral centres were generated in the flask at ca. 393Kelvin or the spark discharge. So in 1973 I did this experiment and noted that the isoleucine/alloisoleucine epimers with 2 chiral centres gave different thermodynamic ration that were not quite 50/50 mixes. At low temperatures the ratios in the colder geologic environment were ca. 1.29 but in the Miller experiment the ratio was down to 1.06; a consequence of formation of chirality at ca. 2500K; only possible in the spark! That fact dumbfounded many exobiologists..

At 10^12 Kelvin the effect should be small but measurable. For the chemist the activation energy that causes the effect is no more than 2 eV, a molehill. For these mesons it's probably the Higgs energy 125Gev/2, a mountain. One day they may get the entropy of activation.

andrew said...

I didn't formally calculate what a factor of 1000 would do in terms of look elsewhere effect, but my intuition isn't far off.

I am guessing 7.3--> 4 sigma or 5 sigma with look elsewhere effect
and 3.1--> 2 sigma with look elsewhere effect.

3.1 sigma is about one part per 1800. 2 sigma is about 1 part per 20. So, you only need 100 other trials for the look elsewhere effect to reduce 3.1 locally to 2.0 globally.

Four sigma is 63000 parts per billion. Five sigma is 600 parts per billion. Six sigma is 20 parts per billion. With a look elsewhere effect factor of 1000, 63 parts per billion locally converts to 4 sigma globally, and 0.6 parts per billion locally converts to 5 sigma globally.

Couldn't find an easy conversion of 7.3 sigma to parts per billion. Seven sigma should be on the order of 0.6 to 0.2 parts per billion.

Of course, my 1000 factor for look elsewhere effect is also very rough.

"Please correct this..." Which part?

Mitchell said...

' "Please correct this..." Which part? '

Everything that says this is an anomaly, violated the standard model, etc.

Look at equation 69.3 in the PDG review that you link to. m1 and m2 are the masses. They equal M + q/p M12 and M - q/p M12 respectively, where M = M11 and M12 are elements of a mixing matrix.

The analogue of this for kaons was observed almost fifty years ago, although the mass difference is not observed as directly.

Mitchell said...

More on kaons in chapter 10 of draft book on weak force by Howard Georgi

Still hoping you will correct this post.

andrew said...

I've been a nomad the last couple of weeks due to asbestos removal, natural gas related renovations (replacing a 96 year old boiler), and electrical renovations, in my house, and I have also spent about four of those days completely out of commission with a bad summer cold (only awake about 4 hours a day for the worst two days - missing father's day festivities and my daughter's birthday party).

I'll take a look when I get a chance, and I will update if I gain an understanding contrary to what I have posted from doing so.