Validation of Garfield with test-beam data of ATLAS MDT chambers in magnetic fields

The precision tracking in the ATLAS muon spectrometer is performed by chambers consisting of 6 or 8 layers of pressurized drift tubes. The drift tubes are filled with an Ar:CO2(93:7) mixture at an absolute pressure of 3 bars. The muon chambers are run in a toroidal magnetic field of 0.4 T created by eight air core coils. The magnetic field B influences the electron drift inside the tubes: the maximum drift time, typically about 700 ns, changes by about 70 ns T-2. B varies by up to 0.4 T along the tubes of the chambers mounted near the magnet coils which translates into a variation of the maximum drift time of up to 40 ns. In order to obtain the r(t) relationships of these chambers with the required accuracy better than 20 μm, the dependence of r(t) on B must be taken into account. Test-beam measurements show that, on the level of 20 μm, the influence of B on r(t) relationship significantly depends on the Ar-CO2 mixing ratio and the gas temperature and pressure. The dependence of r(t) on B will have to be provided by a detailed drift simulation for the given operating conditions. The comparison of predictions of the two most recent versions of the simulation Garfield with test-beam data indicate that only the latest version is capable of predicting the dependence of r(t) on B with the required accuracy. (orig.)