Zooming in on leptonic W decays

In the Standard Model of particle physics, the three charged lepton flavours couple to the electroweak gauge bosons W and Z with the same strength – an idea known as lepton flavour universality (LFU). This implies that differences in the rates of processes involving W or Z bosons together with electrons, muons and tau leptons should arise only from differences in the leptons’ masses. Experimental results agree with LFU at the 0.1–0.2% level in the decays of tau leptons, kaons and pions, but hints of deviations have been seen in B-meson decays, for example in the combination of measurements of B → D^(*)τν and B → D^(*)μν decays at the BaBar, Belle and LHCb experiments.

The W and Z bosons are so heavy that the probabilities for them to decay to electrons, muons and tau leptons are expected to be equal to very high precision, if LFU holds. This implies that the ratios of these probabilities such as R(μ/e), which compares W → μν and W → eν, and R(τ/μ), which compares W → τν and W → μν, should be unity. Experiments at the LEP electron–positron collider measured a surprisingly large value of R(τ/μ) = 1.070 ± 0.026, but a more precise measurement from the ATLAS collaboration at the LHC found R(τ/μ) = 0.992 ± 0.013, in agreement with LFU. This measurement made use of the large sample of top-quark pair events produced at ATLAS during Run 2 of the LHC from 2015 to 2018. These top-quark events can be cleanly selected, with each event containing two W bosons and two b-quarks produced from the decays of the top quarks.

In a new measurement, ATLAS has turned its attention to the comparison of W decays to muons and electrons, via the ratio R(μ/e). The collaboration again used top-quark pair events as a clean and copious source of W bosons. Counting the number of events with one electron from W → eν, one muon from W → μν, and one or two b-tagged jets, provides the cleanest way to measure the rate of top-quark pair production. But this rate can also be measured from the number of top-quark pair events with two electrons or two muons. If R(μ/e) = 1 and W → eν and W → μν decays occur with equal probability, the rates of such ee and μμ events should be the same, after correcting for detector efficiencies. Any difference would suggest a violation of LFU.

Some measurement uncertainties have similar effects on the ee and μμ final states, so they largely cancel in the ratio R(μ/e). However, electrons and muons behave in very different ways in the ATLAS detector, giving different detection efficiencies with differing and uncorrelated uncertainties that do not cancel in the ratio. To reduce the sensitivity of the measured R(μ/e) to these effects, the double ratio R(μ/e)/√(R(μμ/ee) was measured first, where R(μμ/ee) corresponds to the comparison of Z → μμ and Z → ee decay probabilities, determined from the same dataset. The final R(μ/e) was then obtained by making use of the very precise measurement of R(μμ/ee) from the LEP experiments and the SLD experiment at SLAC, which has an uncertainty of only 0.0028. This latter ratio acts as a calibration of the relative detection efficiencies of electrons and muons in ATLAS, reducing the associated uncertainties in R(μ/e).

The final result from this new ATLAS analysis is R(μ/e) = 0.9995 ± 0.0045, perfectly compatible with unity. The measurement is compared to previous results from LHC and LEP experiments (see figure 1). Thanks to the large data sample and careful control of all systematic uncertainties, it improves on the uncertainty of 0.006 from all previous measurements combined. At least in W decays, LFU survives intact.