Performance of the SLD central drift chamber

Nuclear Instruments and Methods in Physics Research A 367 (1995) 111-114

ELSEWER

Performance of the SLD central drift chamber* M.J. Ferob, M.D. Hildreth”, A. Honma”, T.R. Junk”, T.W. Markiewiczc, R. Massetti”, H. Masuda”, B. MoursC‘*,H.A. Neal”, C.Y. Prescott’, L.S. Rochesterc”, R. Shypitd, A. Sugiyamacv3, T. Takahashic’4, T. Usher’, J. Venutif, DC. Williamsb’5, C.C. Young” “INFN, Sezione di Perugia and Universitd di Peru&a, I-061cx) Perugia, Italy bMassachusetts Institute of Technology, Cambridge, MA 02139, USA ‘Stanford Linear Accelerator Center, Stanford, CA 94309, USA dVniversity of British Columbia, Vancouver, BC, V6T IZI, Canada ‘University of Victoria, Victoria, BC, VgW 3P6, Canada ‘Vanderbilt Universiv, Nashville, TN 37235, USA

ABetract The central drift chamber (CDC) of the Stanford Linear Collider Large Detector (SLD) has been recording data since 1992. The low mass of the chamber and the use of a gas characterized by both a low drift velocity and low diffusion constant help to minimize multiple scattering and drift-distance measurement errors. We describe some of the calibrations and corrections applied to the data, and report on the resolutions achieved thus far. We measure an intrinsic drift resolution of 55-I 10 pm in the region of uniform field. Analysis of the full drift-pulse waveform allows for efficient double-hit resolution of about 1 mm. Transverse momentum resolution is given by the formula (4(P,)MY = 0.00502 + (0.010/p,)2, where p, is in GeVlc. Used in conjunction with the SLD vertex detector, the CDC permits measurements of impact parameters of high-momentum tracks to 10 km in the r-4 plane and 36 pm in the r-z plme.

1. Introduetioll The central drift chamber of SLD tracks charged particks over 80% of 4~r, and operates in a uniform solenoidal magnetic field of 0.6T. It has been taking data at the Stamford Linear Collider (SLC) since 1992, and has reamkd over 16OooO 2’ decays in that period. Analysis of the accumulated data sample has allowed us to study the performance of the CDC, and to perform the calibrations necessary to begin to realize the potential of the chamber. After briefly describing the relevant parameters of the

* Work supportedby Departmentof Energy contractDE-AC0376SFw515. ’ Corresponding author. E-mail [email protected] *Resent adders: WP, F-74941 Annecy-le-Vieux, Fraoce. ‘Resent addteas: Nagoya University, Nagoya 464, Japan. ’ Resent dress: Hiroshima University, Higashi-Hiroshima 724, Japan. ’ Present address: University of California at Sama Cruz, Santa CIUZ,CA 95064, USA.

CDC, we will highlight some aspects of the chamber performance.

2. Chamber description The CDC is a cylindrical annuh~s with a length of 2 m, an inner radius of 20 cm and an outer radius of 1 m. It was constructed by stringing wires through a pre-assembled rigid low-mass shell consisting of dished aluminum end plates, 5 mm thick, and inner and outer cylinders made of a laminate of aluminum sheet and Hexcel fiberboard. The chamber contains 80 layers of sense wires arranged in ten superlayers of eight wires each. As part of the low-mass design of the chamber, field-shaping wires are made of 15Oqn gold-coated ahnninum. Six superlayers have a 4l-mrad stereo angle with respect to the beam axis. Each superlayer is made of independent cells roughly 6cm wide by 5 cm high. The sense w& and the field wires separating adjacent cells are aligned on radii of the chamber. Fig. 1 shows a detail of one such cell. The wires are read out at both ends, which provides charge division information useful for pattern recognition. Hits are. ex-

0168-9002/95/$09.50 0 1995 Elsevier Science B.Y. AH rights reserved SSDI 0168-9002(95)00732-6

III. DRIFT CHAMBERS

M.J. Fero et al. I. Nucl. lnstr, and Meth. in Phys. Res. A 367 (1995) 111-114

i12

540 p

0.010 -

1994data

:

520 0.008 500 EE

x 480 460 0.002 -

T I I 0.005

0

I

I

I

, I 0.010

Calculated Correction to vd -40

-20

0

20

x

(mm)

40

60

Fig. 2. Measured correction to u, vs the correction calculated

from the gas parameters.

Fig. 1. Wire layout of a single cell of the CDC.

tracted from the raw wavefotrns using programmable processors, and multiple hits from a single wire are recorded.

3. Gas CO, was chosen as the primary component of the gas mixture because of its low drift velocity and low diffusion constant. Isobutane was added as a quencher at a level low enough to keep the mix nonflammable and to thus rninimize safety problems. Argon was added to increase the gain to the desired level; water was added after tests [l] indicated its presence could ameliorate the effects of wire aging. The final gas mix is 75% CO,, 21% argon, 4% isobutane, and 0.2% water, and has a drift velocity, u,, of 7.9 pm/ns in the region of uniform drift field. One disadvantage of our gas mix is that the drift velocity depends significantly on the gas density and composition, and’ on the electric field. The CDC is maintained at a constant temperature and voltage, but no attempt is made to stabilize it against fluctuations in atmospheric pressure.

this relationship by minimizing the fit residuals as a function of drift distance. The electron drift velocity, u,, varies inversely with the pressure; temperature and water vapor concentration are regulated and monitored. Reconstructed tracks are used to measure changes in u, by allowing the drift velocity to be one of the variables in the track fit. The value of u, is determined for each four-hour run by averaging the values obtained for individual tracks. The overall change in the drift velocity can be as much as 2%. The measured fractional drift velocity correction can be fit to a linear function of the pressure and the residual variation of temperature and water vapor concentration. Changes in water vapor concentration can be significant: a change of 1OOClppmresults in a 1.4% change in u,. Fig. 2 shows the measured correction plotted against that calculated from the fitted function. The width of the residual distribution between the measured and calculated correction is less than O.l%, which corresponds to an average drift-distance uncertainty of 15 km.

5. CDC Performance 4. Drift velocity calibration

5. I. Drift distance resolution

To maximize tracking efficiency, we use hits from all areas of the cell, including those regions near the sense and field-shaping wires where the drift velocity is high and changing rapidly. An electrostatic model of the cell, including the effects of adjacent superlayers, and a first estimate of tie’ relationship of drift velocity to electric field, are used to calculate the time-to-distance relationship for each layer. The data are then used to iteratively correct

The intrinsic drift distance resolution of the chamber (i.e. local to a given drift cell) is determined primarily by the diffusion coefficients of the gas, the ratio of sampling speed to electron drift velocity in the gas, the drift cell design, electronic noise, and the algorithm used to extract the time information from the raw waveforms. Since hits on adjacent wires in a cell are in nearly identical environments, many of the uncertainties which affect the measure-

113

M.J. Fero et al. I Nucl. Instr. and Meth. in Phys. Res. A 367 (1995) 111-114

ment of the absolute position of the hits cancel out if differences of track-fit residuals of adjacent hits are taken. The local resolution derived from Gaussian fits to the distributions of these differences of residuals is plotted as a function of drift distance in Fig. 3. Since only hits on tracks are used in this calculation, and since the fitter can drop hits in the tails of the residual distributions, the results shown here are for the cores of the distributions, comprising about 95% of the hits. In the region of linear field the resolution follows the curve expected from diffusion (varying as & and 68 pm at 1 cm). The mean resolution in this region, for bits lying on near-radial tracks, is 82 Fm. The resolution is degraded at the extremes of the drift region, due to the higher u, and nonuniform drift field. The global hit residual resolution, and even more importantly, the. resolution on the measured track parameters, depend on a knowledge of various systematic effects, such as: the time-to-distance relationship (its variation layer-to-layer as well as with electric field), wire positions, time offsets, and gravitational sag. The alignment of the chamber wires is performed using measured tracks. Selected tracks from hadronic decays of the Z” provide information on the deviation of the positions and orientations of individual cells from their nominal values, which leads to a set of a cell-to-cell alignment constants. After this local alignment of the cells is done, we use muon tracks from the decay Z” + p,+t~- @i-muon event) to determine the overall rotation of superlayers, by exploiting the fact that the two tracks in such an event should be back-to-back and each have the same momentum as the beam. Fig. 3 also shows the global resolution after alignment and corrections. The mean global track resolution in the region of linear drift, for hits lying on near-radial tracks, is msasuted to be 92 km. The difference between the global and local resolution reflects an uncertainty of 30 to 40 pm in alignment and other calibrations, assuming that these effects add in quadrature. 5.2. Resolving of close hits

I

I

I

I

I

I

L

300 _

J

1994 Data l

_

3

I

* +

Global Resolution

+ Local Resolution -

68 .qK)

@

.g 200 ** 5 P $

.

lloo;y;

0

I

I

0

t

I

I

-1

IO 20 DriftDistance (mm)

30

Fig. 3. Global and local drift distance resolution.

represents the effect of measurement error, and the second, that of multiple scattering. In the actual chamber, numerous effects, such as hit inefficiencies, misalignments, and tails on drift distance resolutions, conspire to degrade the resolution obtained for real tracks. The resolution at high momentum can be measured using the mono-energetic tracks in di-muon events. At low momentum, there is no such simple calibration event. However, cosmic rays passing through the CDC may be used for this purpose, by considering the upper and lower halves of each track as two separate tracks; comparing the two measured momenta yields an estimate of the resolution. Using these techniques, we measure the momentum resolution function for the CJJC to be (a(p,)/p:)* = 0.0050* + (0.010/p,)*.

*O”l

The ability to resolve two close hits depends on details of the waveform analysis as well as the intrinsic resolution of the chamber electronics. Our algorithm assigns a second hit when it senses a sufficiently large deviation on the falling edge of the pulse. Fig. 4 illustrates our two-hit resolution. An efficiency of 50% is achieved by a separation of 1 mm, rising to 100% (horizontal line) at about 1.5 mm. 5.3. Momentum resolution A model calculation for an ideal chamber with the same geometry and material as ours yields a momentum xesolution which can be parametrized as (u(p,)/p:)* = 0.fW312 + (0.0086/p,)*, where p, is the track momentum perpendicular to the. beam axis in GeV/c. The first term

0

’

1

’

’ 2

’

’ 3

’ 4

Hit Separation (mm)

Fig. 4. The number of second hits on a wire vs the separation between the first and second hits.

III. DRIFT CHAMBERS

M.J. Fero et al. I Nucl. I&r.

114

and Meth. in Phys. Res. A 367 (1995) 111-114

6. Overall tracking system performance The performance of the SLD tracking system is perhaps best characterized by the extent to which it can be used to tag decays of Z’s into heavy quarks through identification of secondary vertices. In addition to the CDC, this system includes the SLD Vertex Detector [2] (VXD). To exploit the vertexing capability of this system, tracks must be efficiently found in the CDC and linked to pixel clusters in the VXD, then extrapolated to the SLC interaction point. The VXD consists of four cylindrical layers of partially overlapped charge-coupled devices (CCDs) with 22 pm X 22 pm pixels, which typically provide two hits at radii of 3 and 4cm. The beam spot at the SLC is small [(a;, uY,crz)= (2.4,0.8,750) pm] and its position is tracked to within 7 pm. The two-prong miss distance is the distance of closest approach between the extrapolations of the two tracks in di-muon events; dividing this number by & gives a measurement of the resolution in the track position at the interaction point. Fits to CDC data alone yield u,+ = 155 pm and o;, = 1.9 mm for the resolutions. Adding the linked VXD hits to the track dramatically improves these resolutions, yielding u,+ = 10 pm and or,, = 36 pm. Without the CDC, the relatively small separation of the VXD layers would limit the resolution to 48 pm in both views, even if it were somehow possible to actually do tracking in the two layers of the VXD. The impact parameter of a track is its distance of closest

approach to the interaction point, signed by the hemisphere of the jet to which it belongs. Its quality is determined by how well one can measure the position of the interaction point and how well one can extrapolate the tracks. Fig. 5 shows the resolution in this quantity as well as the excellent agreement between measured data and the MC simulation, achieved without recourse to ad hoc smearing of the track parameters. This allows for the extraction of physics results with minimal systematic errors from this source.

7. Conclusions The central drift chamber of the SLD detector has proved to be robust and reliable, and is now performing at a level approaching its design specifications. At the present stage of data-taking, the errors in our physics results are still typically dominated by statistics, but as the statistical power of our measurements increase, we will need to improve the our understanding of the tracking efficiency and momentum resolution. The larger data sample will help us to do this. In addition, the installation, in late 1995, of an improved VXD, with three layers and increased angular coverage, will augment the capabilities of the tracking system. It should also allow us to use the CDC more effectively, and help us to improve our understanding of its performance, by providing independent tracking.

Acknowledgements The authors gratefully acknowledge M. Breidenbach for his conceptual contributions and support. Special thanks to R. Boyce as chief engineer of the project, to J. McDonald, D. Peterson, and A. Johnson for their technical help in the construction of the chamber, to G. Haller, D. Freytag, M. Freytag, and L. Paffratb for the design and commissioning of the chamber electronics. The data taken with this device would not have been possible without the inspired effort of the SLD online and offline software groups.

Y c

g

0

$ g g

0.1

0.01

4 0.001

References IEEE Trans. Nucl. Sci. NS-36 (February 1989) 595. [2] C.J. Darned et al., Proc. 26th Int. Conf. on High Energy [l] J.P Venuti and G.B. Chadwick,

impact Parameter (cm) Fig. 5. The impact parameter for data and Monte Carlo.

distribution

of tracks in Z” decays

Physics,

Dallas, TX, USA, August

1993, p. 1862.