The CMS detector magnet

CMS (Compact Muon Solenoid) is a general-purpose detector designed to run in mid-2005 at the highest luminosity at the LHC at CERN. Its distinctive features include a 6 m free bore diameter, 12.5 m long, 4 T superconducting solenoid enclosed inside a 10,000 tonne return yoke. The magnet will be assembled and tested on the surface by the end of 2003 before being transferred by heavy lifting means to a 90 m deep underground experimental area. The design and construction of the magnet is a 'common project' of the CMS Collaboration. It is organized by a CERN based group with strong technical and contractual participation by CEA Saclay, ETH Zurich, Fermilab Batavia IL, INFN Geneva, ITEP Moscow, University of Wisconsin and CERN. The return yoke, 21 m long and 14 m in diameter, is equivalent to 1.5 m of saturated iron interleaved with four muon stations. The yoke and the vacuum tank are being manufactured. The indirectly-cooled, pure-aluminium-stabilized coil is made up from five modules internally wound with four layers of a 20 kA mechanically reinforced conductor. The contracts for the conductor and the outer cryogenics have just been awarded, and the remaining coil parts, including winding, are being tendered worldwide in industry. The project is described, with emphasis on the present status.

The single most iinportant aspect of the overall clctector dcsigu is the configuration and parameters of the magnetic field for the mensuremmt of rniinn " m e n t a . 'l'hc requirement for a good momentum resolution, without making stringent dcmarids on the spatial resolution and the alignment of muon chambers, lcads nsturally t o the choice of n. high solenoidal magnetic field.
A long mpcrconducting solcnoid (L = 12.5 m) hns been chosen with a free inner dinmctcr of G m and B uniform magnetic field of 4 T. Thr: muon spectrometer then consists of a singlc magnct allowing for a simpler architectmc for the detector. The inner coiI radius is large enough to accommodate the inner tracker nncl thc full calorimetry.
The magnetic flux is returned via a 1.6 m thick saturated iron yoke instrumented with four slations of muon chambers, and the yoke is thick enough t n nllnrv safe idantification mid to enable a powerfill triggar on muons. The CMS experiment is built and funded by an internntional collaboration of High Energy Physics insti1;utes from thirty onc countricst: and by CERN. The experiment will he installed on the interaction point I5 of LBC ah a depth of DO m below ground.
The design field of the solenoid is 4 T . Thc opcrating field rcqucstcd by the CMS collaboration is between 3.5 to 4 T, 3 T being t.he minimum field requirecl for doing good physics. The design tias thus been carried out; to maximize the chance of reaching 4 Tesla. This explains why, except for mechanical safety margins which arc clearly conventional, other 'state of the art' safety margins like strain, stabiliby, IC margin, etc., are not. necessarily met at 4 Tesla. In fact, apart for thr! amouiit of supcrconductor, a 3,s T 'standard thin' solenciid would bc practically identical.
Thr! CMS magnet i s first nssemblcd and tested in a surfme hall (Pig. 'I) then lowered in the underground area (Fig. 9). by hcavy lifting mans, This allows to doconple the work on the magnet assembly and test from the construction of the underground area.
The CMS magnet consists of two main parts: the yoke and the coil, n pcrspectivc view of the opcn magnet can bo sccn in Fig. 2. Thc rcturn yokc is a 12-sided structure divided in thrcc main componcnts: tlic barrel yoke and the two endcap yalws. Its main parameters are given in T&lc 1.  The coil is an indirectly coolecl, aluminium shbilizerl, four layer superconducting solenoid. Its main parameters arc given in Table 11. Extensive 3D magnctic rnodels hnvc been dmdoped to optimize thr! field map and oGtairl tfhe magnitrtde of tha magnetic forccs both for the yokc slid the coil 131; thesr! results have becn used to dimonsion thc coil 171.

A . The D a m l Yoke
The barrel yoke was designed at CERN. 11; is split into fivc barrel rings, having each a mass of 1200 tonncs, which can mow iri the axial direction to give access to thc bnr-re1 mumi stations. For the barral rings, t,hc only Inad is gravity, magnetic field introducing only axial forces nn the rings. 'I'he central barrel ring supports the vtmum tank housing the coil. The vwuum tank, made of stainless stcc1 plates GO and 30 m m thick, will kw welded n r n d Ihe coil.
A large contract has been plnccd by El'H Ziirich for the construction of the bnrrd yoke and the v m " i lank with DWE (Deggcndorfer Werft und Elsenbail, Ucagendarf, Germany), This contract is financed by A CMS consortium (CMS 'common fund', Cyprus, Germany, R.ussia, USA and Switzerland). 'l'he 450 mrn thick heavy plakes, necessary for the compoutid 630 min t,liick outer hycrrs, have been subcontracted to Mora Zavod (St, Pctereburg, Russia). The siipport fect for the outor barrel rings are being manufactuwd, as an in-kind Pakistani cootribution, by SES (Islamabad, Pakistan).

Pig. 2. Parapectivc viow UT
A complete trial assembly is carried out atr DWE, using a Ferricwheel arrangement (SCC Fig. 3), and the first ring has just been nssemblecl bcfore shipment to CERN. Later, the final assembly of the b a r d yoke will lw pcriormecl at. the CERN site by i~ consortium DWE/FCI, starting in

H . 2% Endcap Yoke
Each endcap yoke, dcsigiicd nt tlic UIiiversily of Wisconsin, is biiilt from t h c o independent disks which c m be movcd 011 carts (LIS seen on Fig. 2.), and sepnratcri to provide ~C C C S S t o thc! fClrWaFd muon stations and inner subclctcctors. Duct to thc axial magnetic field tlir: two iriricr disks must withstand an attraction force nf about 85 WIN and rasjst tho largo bonding moment inducod. Therefore these disks arc GOO mm thick where# the outer disk is only 250 mm thick. Thc rlisits i~r c l)uilt, from sectors connected together by Inrgc holts (Suporbolt, Cnrncgie, PA, USA) taking

A . Dcsign of the Coil
Thc CMS coil design is Ixmetl, as Tor a numbcr of cxisting large dctcctor superconducting solenoids, on tlic or)thdpy stabilization conccpt, bccmsc it, is not subjeck to substantial external sources of rl isturha~iccs.
Important. informaticin ca.n be gained from the previous designs a.nd in particular the A LIWH solcnoid has been used in many ways as n referencr? mnrlcl for the dcsign of the CMS coil [4]. In particulnr thc thcrmosiphon cooling rriodo has bocn rctaind, Howcver, the CMS coil cannot he simply extrapolated from ALEPH; bccaiist: of thc very large increase in magnetic field (from 1.5 t.o 4 T) and tlic requirement of 1imit;ed r d i n l t.liickness, t.hc gtrsiri at 4 'I' T h e main chaiiges introduccd for thc CMS coil dcsign arc: a four-layer winding instead of a mono-layer one to provide the iiccded amperc-turns, n. construction in fivr: mudnius to ;tllow hmsporhtion, n self supporting winding mechanical structure haserl OH a rnechanically reinforced conductor wound insidr! n thin mandrel to limit shcar strcssw in thc insi.ilation in spite of the large strnin.

U . Y 5 C Coonductor
The design of a seIf supporting structure obtained by modmnically rttinffircing the conducturt makes this component, more complex than ot?her aluminium stahilizcd coritluctors prcviously iic;cd for thin solenoicls. The! dimensions arid tlic component propcirtions are rletermincd by the general coil design according to t h ! mcchnnicai strengh, quench protection and stabilily requirements.
It has been decided to reinforce khe pure aluminium conductor by welding two aluminium ~I l o y sections. Elcctrori Bcarri wclding procwscs hikvc been d~veIopcd at ETH Zurich and CERB Main Wnrltshop during the last years.  Thr! d(!sign and procurcmont of the conctuctor is ;t subcollaboration between CERN/CMS, Fermilab and ETH Ziirich. An egrccrrient has bccn signed to this effcct. with CERN ncting for CMS. Scvcrd lmge prociircments have been identified arid have bccn organized hy institutes as bllows: Sc strmds: Fwnild~, contrtlct, GO 76 to Outokumpu (Pori, Finland) arid 40 % to IGC (Waterhry, CT, container and the cryogenic lincs. T h colcl box and LHe container will be installed ncar thc magnct whereas the cornprcssors and pressure vessels will Iw installed at the siirfnrx level. 'l'he completr! systom will be run temporarily on thr! surface for refrigerator commissioning and coil tests, tis shown in Fig. 7. Thc contract for the outer cryogcnicx has been recently nwmlcd by CERN to Air Liquid0 The bi-polar power supply is located alongside the refrigerntor cold box in Lhe servicc cavern. It will deliver a " . " ' Or 2o kA 'It a rnnximum alloiving n cIiarging time of 4 11. Thew are two modw for slow rlischa.rging the coil ciirrcnt: in normal opcration dis-cIiarge will 'be pcrforrnecl using thri power supply, or the currcnt can bo cluinped into tbc rcsistor bank set at; its lowst resistance valuc! of 2 mR. In CMT! of ornergency, n fwt discharge in n 30 rriSt resistor bank c i~n be used; thc Lime constant oE Lhe curr-cnt dccay is, in this case, 460 s.
As thc trmperature of flit cold mass inny ronch GO IC after iL fast discharg(+, then needing 3 days for rc-cooling, B grcat effort will br! m& to makc s i i w that fast discharges RYE only triggered by thc most important nlarms.
IV. T l l R EXPEHIMEXT.'I\I. k . E A are repolterl in pll and jI3], l;ig, sllow,q a.

voltagr! Of 22
It has been chostm to msemble and tost h e niagnst, in a Iargc surface hall (23.5 m high that will bo reduced lator to 16 in) bcforc lowering it into thc iinderground ex~~erirricntal cavern .Wintcd a(; a depth of 00 m. rylie insertjoll of tho coil jnsidr! the vacuum tnnk is performed in the surface hall. It rcquircs to bring, and mnintain, tho 220-tonne coil horiaontally in a cmtilevererl position. Thc rotation is dorir! using the same Ferris-whd INFN, a"mguniant used to nssr!mbl~ the lmrrel rings shown in Fig. 3. ' J l e n , 'he central barrel ring. already equipped with tho oiiter shell of tlic vacuum tank will slide on licrtvy duty air pads over the coil which will then hc suspended to thc outer shell iisirig titanium tie-bars. The same proccdure will hr! used to slide thc ccritr:tl-bxrreI-ring/coil assembly ovcr the inner shcll of the vaciiuin tnrik.
As &own in Fig. 9, be:.ivy lifting means will have to be used iis the heaviest part will weigh 2000 tonnes. The coil winding procurement is a sub-collaboration be-INpN, An ngrecrncnt l;* this (,,ffcct has tween CERN been nil international tcnrlcr answers are currtmtly under cxaminabion.

D. Thc Ancillaries
'The 1.5 kW extcmal cryogenics sub-system, which i s dcsigiicd by the LHC Cryngcnic group, at C E m , cornprises thc compressors, tlic cold box, the vcsscls containing 200 Inn of prcssurized helium gas? thc 6000 1 LBc bch7v'cc~II CERN (nct,ing for CMS) heen issuecl I,y INBN, The choice of using o. krgc surface hall rather than the 'l.'he experimental cavern is scpsrnted from the service underground area, allows to construct ant1 test the magnet cavern by a 7 m-thids stiiclding wall. in pmalld with the civil engineering. In fact it allows to The magnet will be lowercd in 11 lnrge pieces; (tveigliing start working on the magnet msembly alrcady in July 2000 froin 500 to 2000 tonnes) into the iiridergrorlrld cxprirncriwhile LEP is still in operation. This solution also reduces ' tal c;Lvmn. The coil will be connected t o thc ancillaries to the minimum tlir! size of tho underground cavern. situtitcd in thc: scrvice cavern by bus bars and transfer lines, AR see11 on Fig. 10. A contract for the outer cryogenics has just been awarded by C E W t o Air-Liyiiidc The 20.4 m dianietcr shnk, giving ~C C C S S to the cxperimental cavern, will be separated from thc surface hall by a 2000-tonnc mobilc radiation Rhielding plug which will also be used as support structure for the transfer of t.he magnot to thc ctxpcrimental cavern (see also Fig. 9). . .

Pig.
10. Perspective view of tho CMS magnet inside tlie experimental c;~vcrn; tho bus bar and lransfer line conncclionr to tho power supply aitd cold box in the service cavern cross tho 7 m thick shielding wnll.

A. Participating h s t i f u t e s
main parts and systems n the yoke, consisting of the barrel, the v~wuiim tank and the two enrlcnps, I tlie coil, consisting of the general engineering, the superconductor, nnd thc coil winding, 1 thc ancillarics, consisting of the cxtcrnsl cryogenics, t h e power systems and circuit, and the control system.
This striicture is rcflected in the organization of the CMS magnet project. The management and general coordin;ition is done by a CERN based group. Work is carrird out in the institutes of trlic? CMS Mag'ar:t, Collaboration, and in pnrticular: CERN/CMS is in charge of the bnrrel yoke, The CMS rnag'nct projoct can be grouped into three endcnp yokc in r:oIl;iboration with CERN/CMS, powcr supply, Lrexkcrs, outer cryogenics, process control etc.. , I

L1. Uvgnnbation oj Proczirements
Major procurements are orE;mized, whcn possible, following this distribution of tasks by asking pwticipting institutes to place direct contrtwts in industry. In fact, procurements for the magnet; project, which is a 'Common Project' of the CMS Collnboration, can be done in 3 ways !