The choice of the magnetic field configuration influences strongly the general design of the detector. The requirement for a good momentum resolution, without stringent requirements of muon chamber resolution and alignment, and keeping a compact spectrometer, leads naturally to the choice of a high magnetic field. We have considered both toroidal and solenoidal fields. A solenoidal field has been preferred because: due to the small size of the beam, the vertex is known in the transverse plane with high precision, and the strong bending in the transverse plane facilitates the task of triggers based on pointing to the vertex, the momentum measurement in a solenoid starts at small radius, while for a toroid it starts after the absorber; hence the overall size of a solenoid is smaller than that of a toroid, a wide experience exists in the construction of large solenoidal magnets for particle physics experiments like Aleph, Delphi and H1. We have chosen a long solenoid (L =13 m) with a free inner radius of 2.95 m and a uniform magnetic field of 4 T. The favorable aspect ratio of the solenoid allows efficient muon detection and measurement up to a rapidity of 2.4 making forward toroids unnecessary. The muon spectrometer therefore consists of a single magnet. The inner coil radius is large enough to accommodate the inner tracker and the calorimeters. The magnetic flux is returned via a 1.8 m thick saturated iron yoke instrumented with muon chambers. The overall dimensions of the detector are: a length of 22 m, a diameter of 14 m and a total weight of 12000 tons.

9.1 The iron yoke

The iron yoke is used to return the magnetic flux, act as absorber and house and support the four muon stations, both in the barrel and the end-cap regions. The system is designed as a 12 sided structure, a compromise between the size of the muon chambers and the angular acceptance for muons. The barrel is subdivided longitudinally into five 2.6 m long iron barrel rings. The central iron barrel ring in ( Fig. 9.1), which is the only part fixed around the interaction point, is used to support the superconducting coil.

The central section of the outer vacuum tank is attached to the inner part of the central barrel ring and the coil hangs symmetrically. The other four iron barrel rings can slide on rails along the beam direction to allow insertion and maintenance of the muon stations. Structural welding has been deliberately avoided and instead prestressed tiebars are used. This allows that each barrel section can be assembled in the horizontal position at the manufacturer to check the geometry, before being sent to CERN for final assembly in the vertical position. Each endcap consists of three independent iron rings which can be separated to provide access to the forward muon stations. The first endcap disk must support a 12000 ton magnetic attraction and for this reason it must be built from 600 mm thick iron blocks to be assembled on the CERN site by electroslag welding.

9.2 The superconducting coil

The diameter of the solenoid has been chosen as the maximum compatible with road transportation to CERN, i.e. 6.8 m for the outside diameter of the winding oo allow its construction at an outside manufacturer. The coil will be assembled on the central iron barrel ring as a complete sub-unit in the surface hall, where it will be tested at the full design current with the iron yoke closed. Then the coil, with the central iron barrel section will be lowered as a single 2400 ton unit into the experimental hall. Techniques developed for the construction of large solenoids such as Aleph, Delphi and H1 have been introduced in the design of the CMS solenoid. The main features which have led to the high quality and reliability of these large magnets are the use of an high purity aluminum stabilized conductor and indirect thermosiphon cooling, together with an inner layer winding mode and full epoxy impregnation. However, the large increase in parameters such as magnetic field, Ampere-turns, forces and stored energy density necessitates some changes from the previous designs, a four-layer winding, (instead of one or two), and a larger and more complex conductor designed for an higher rated current and to provide the mechanical reinforcement needed to resist the magnetic forces. The issue of stability has been an important guide in the design of the new conductor and of the winding structure. Elaborate mechanical analysis of the induced stress distribution among the components has been carried out in order to minimize shear stress concentrations which may lead to potential disturbances. Coil Parameters The Fig. 9.2 shows the cross-section of the end of the coil inside the cryostat.

The superconducting coil consists of four identical sections, and Table 9.1 summarizes the overall coil parameters.

9.3 Cost estimate

The total cost of the magnet system is 117 MCHF, out of which 73.5 MCHF is foreseen for the superconducting solenoid and 43.5 MCHF for the magnet yoke. The participation of Russian side is possible in the design, manufacturing and assembly of the magnet yoke and vacuum vessel of the superconducting solenoid.

Preliminary contacts were established with factories "Izhora" and "Atommash". The factory "Atommash" has been retained for having all necessary equipment and qualified staff to manufacture both magnet yoke and vacuum vessel at a price substantially lower, than any western firms. Cost estimate of a possible order to "Atommash" in western value is given in Table 9.2.

First contacts between CERN experts, and Atommash engineers, together with ITEP representatives, have led the conclusion that material worth 19.2 MCHF (as itemized in Table 9.2) could be manufactured at "Atommash", and transported to Rotterdam, for approximately 11 MCHF.