Contents 1 Introduction 2 2 Overview of High-Luminosity Background Extrapolations Methodology of Background Characterization

Transcription

1 BaBar Note October 2000 Report of the High-Luminosity Background Task Force C. Hast, W. Kozanecki (chair), A. Kulikov, T.I. Meyer, S. Petrak, S. Robertson, T. Schietinger, M. Sullivan, and J. Va'vra

2 Contents 1 Introduction 2 2 Overview of High-Luminosity Background Extrapolations Methodology of Background Characterization Extrapolation to High Luminosity (assuming no improvements) Expected Impact of Planned Accelerator Improvements Analysis of Potential Long-term Improvements BaBar subdetector-specic issues Silicon Vertex Tracker (SVT) Drift Chamber (DCH) Detector of Internally Reected Cherenkov Light (DIRC) Contributions to the DIRC Background Rate Potential Improvements Electromagnetic Calorimeter (EMC) Current Status Extrapolation to High Luminosity Conclusions Trigger Contributions to the L1 Background Rate Trigger Upgrades Accelerator-specic issues Systematic Uncertainties aecting Background Predictions Short-Term Scrubbing Long-term Vacuum Improvements in the PEP-II Rings Improvements in Injection Eciency Conclusions 28 1

3 1 Introduction The High-Luminosity Background Task Force (HLBTF) was set up in June 2000, in response to a request by the PEP-II management to estimate the BaBar background levels that could be expected as the B-factory luminosity increases by about an order of magnitude over the next few years. The mandate, as spelled out by J. Seeman, calls for the following: starting from today's measured levels, and assuming no improvements, estimate BaBar backgrounds for luminosities of and of ; then estimate the levels including expected improvements: new IR2 collimators, DIRC shielding, HER low pressure chamber, beam scrubbing in the IR, LER vacuum lifetime improvement; identify additional improvements, if any, which could help future backgrounds. Serving on the Task Force were the following BaBar and PEP-II physicists: T.I. Meyer (SVT and BaBar radiation protection), T. Schietinger (DCH), C. Hast and J. Va'vra (DIRC), S. Robertson (EMC; detector background simulations), S. Petrak (Trigger), A. Kulikov (LER vacuum), M. Sullivan (synchrotron radiation), and W. Kozanecki (accelerator experiments; beam-line simulations). In order to identify and characterize the various components of the background, a number of machine physics experiments were carried out in June and July These consisted primarily in measuring the dependence on beam currents and on luminosity, of background levels in each BaBar subdetector. In addition, highly valuable data, particularly on scrubbing rates, were collected parasitically during BaBar physics running, or extracted from the BaBar and PEP-II history buers. These experimental studies were complemented by existing or new simulations of beam-gas scattering around the PEP-II ring (at the Turtle level), and of the propagation of beam-gas induced showers in the BaBar detector (at the Geant level, in the BBSIM framework). Finally, a small number of collimation and vacuum bump experiments were also carried out, but these remain to be analyzed in detail, and provide only qualitative input to the present document. This report, which was prepared collectively by the Task Force members, is organized as follows. Sec. 2 provides an overall summary of the ndings. For the avid reader, a more detailed discussion of some of the subdetector-specic issues appears in Secs. 3 and in the bibliography; accelerator-related issues are addressed more extensively in Sec. 4. Sec. 5 spells out the conclusions. 2

4 2 Overview of High-Luminosity Background Extrapolations 2.1 Methodology of Background Characterization Based on the rst year of running experience, the response of the BaBar detector to machineinduced backgrounds can be quantied by the following measures: the instantaneous dose rate in the radiation protection diodes BW:MID and FE:MID, which are representative, within about a factor of two, of the harshest radiation levels hitting SVT modules in the horizontal plane. The radiation hardness of SVT detector modules is currently estimated at 2 MRad; the total current drawn by the drift chamber, limited to about 1000 A by the existing HV power supplies; the counting rate in DIRC phototubes, which, with the present electronics, starts inducing signicant dead time when it exceeds 200{300 khz; the fractional EMC crystal occupancy above a 1 MeV threshold, and the number of crystals above 10 MeV, which characterize, respectively, the potential degradation of the calorimeter energy resolution, and the number of fake neutral clusters produced by machine backgrounds; the level-1 (L1) trigger rate, currently limited to about 2{2.5 khz by DAQ bandwidth considerations. Each of these observables potentially receives contributions from synchrotron radiation, beam-gas-induced lost-particle backgrounds, and two-beam backgrounds. The rst two are single-beam eects; the latter occur only with beams in collision 1. Reliably extrapolating these distinct contributions - which aect in a dierent way the various subdetectors - therefore requires measuring separately, as a function of beam currents and of luminosity, the response of individual subdetectors to: electrons (HER) only: backgrounds from lost particles (beam-gas bremstrahlung and Coulomb scattering), and from synchrotron radiation (if any); positrons (LER) only: backgrounds from lost particles (beam-gas); both beams, out-of-collision: beam-gas cross-term (if any), plus the above two; both beams, in collision: backgrounds from luminosity and/or beam-beam tails, plus the above three. Several such measurements were carried out, under tightly controlled conditions, during background-dedicated machine developement (MD) sessions; as such they represent the lowest backgrounds ever achieved in BaBar. In some cases MD data were complemented by parasitic measurements carried out while BaBar was recording physics data, thereby providing valuable consistency checks. 1 Although clearly present in early running, no beam-gas LER-HER cross-term was detectable after a few weeks of scrubbing. 3

5 2.2 Extrapolation to High Luminosity (assuming no improvements) As discussed in more detail in Sec. 3, the resulting data sets were parametrized in terms of polynomials of beam currents and luminosity. Good ts could typically be achieved using a linear dependence only, with an occasional small additional quadratic dependence on single-beam current(s); the dierences between quadratic and purely linear extrapolations are typically around 20 to 50 %, but can reach a factor of two. Background formulas The corresponding expressions are listed below; they form the basis of all the high-luminosity extrapolations presented in this report (beam currents are in A, and the luminosity is in units of cm,2 s,1 throughout). SVT dose rate (as measured by pin diodes in the horizontal plane, including an estimate of injection dose integrated over one running year): FE : MID [krad=year] = 128 I LER +16I 2 LER BW : MID [krad=year] = 246 I HER +9:1I 2 HER DCH current (for a HV setting of 1900 V; at 1960 V the currents would be 65 % higher): I DCH [A] =,10+35:3I LER +23:5I 2 LER +77:2I HER +46:3I 2 HER +41:9L DIRC (data from a well-shielded sector; optimistic in that it reects a well-tuned machine): BG [khz] = 8:5 I HER +35I LER +25 L EMC (the best-t scenario, given here, diers by only 30 % from the (optimistic) linear extrapolation. Note, though, that this formula describes only a median behavior. In the hottest areas, backgrounds are up to a factor of 10 higher): Occ E>1MeV [%] = 9:8+2:2I HER +2:2I LER +1:4L N E>10MeV = 4:7 I HER +0:23 I 2 HER +2:4I LER +0:33 I 2 LER +0:6L Trigger (optimistic in that it reects a well-tuned machine): L1 [Hz] = 130 (cosmics) I LER I HER +70L High-Luminosity Model Predictions The background extrapolations above can then be combined with assumptions about the time-evolution of beam currents and luminosity, to yield a projected time-line for background levels. Such a model [1], expressed in terms of year-averaged beam currents and luminosity, is presented in the table below, together with the corresponding background projections. A graphical representation of the same results is presented in Sec. 3, one subdetector at a time, in Figs. 1, 2, 5, 8 and 10. These show that while single-beam backgrounds will typically be LER-dominated (because of the higher currents), the dominant background source is in most 4

6 cases the luminosity itself (the SVT being a notable exception). The present interpretation is that energy-degraded electrons or positrons abundantly produced by radiative-bhabha scattering, hit aperture limitations within a few meters of the interaction point (IP), thereby spraying BaBar with electromagnetic shower debris 2. Year HER LER Lumi SVT H SVT L DCH DIRC EMC 1 EMC 10 Trig [A] [A] [10 33 ] [krad/y] [krad/y] [A] [khz] ocp [%] # clstr [Hz] ? Systematic Errors Even though the quality of the input data is exceptionally good for this type of measurement (10{20 %), these predictions are aected by several systematic eects (Sec. 4.1). Overall, a global factor of about 2 is the best estimate of the uncertainty aecting the extrapolated average background levels, that is available at present. It should be stressed that the conclusions derived from the table above, and discussed in the remainder of this document, are, in some cases, sensitive to the time-line assumed. Apart from the uncertainties associated with the background prediction itself, the actual evolution of beam currents and luminosity may be quite dierent from the model used here, especially for luminosities in excess of ; the same holds for any ndings of the task force, that have schedule or funding-prole implications. Background-Tolerance Projections Taking the above predictions at face value, neglecting systematic uncertainties, and assuming no improvements, long-term background-tolerance issues can be summarized as follows. SVT: if the silicon detectors can be shown to survive a dose of 2 MRad (which at present remains an assumption), it may be possible to postpone the currently planned replacement of horizontal SVT modules until 2003 or If radiation levels after 2004 remain at, or exceed, the 400{900 krad per year level (including injection losses), and the maximum acceptable dose remains at 2 Mrad, then a module replacement every other year could become necessary. Drift chamber: if the DCH HV setting can be lowered back to 1900 V for future datataking, as envisaged at the time of this writing, then power-supply upgrades are not strictly necessary until 2004 if the machine remains well-behaved. Wire aging from accumulated charge becomes a concern (at the same voltage) on a comparable, or slightly longer, time scale. It should be pointed out, however, that already at present the headroom for buering background spikes occasionnally gets uncomfortably small; with increasing steady-state DCH current, operational ineciencies associated with 2 The case of the L1-trigger rate is slightly dierent: its two-beam contribution is consistent with that of large-angle Bhabha events. 5

7 frequent DCH trips may become an issue. An early HV upgrade, such as the one planned for the upcoming shutdown, is therefore highly prudent. DIRC: the lead-shield upgrade foreseen for the Winter 2000 shutdown focuses on improving access to critical detector and machine components, but may not signicantly improve the DIRC background averaged over all sectors, when compared to the measurements reported here. If this is borne out by 2001 running experience, the DIRC will run into dead-time problems at the latest in Once the currently planned electronics upgrade is completed, the DIRC should remain background-tolerant until at least EMC: typical occupancies above the 1 MeV threshold will reach 30 % in 2002, indicating a potentially large degradation of the energy resolution, and possibly causing timing problems with the readout as well as diculties with the clustering algorithm. While the latter two can be easily handled by raising the EMC "digi" energy threshold, the real issue here is degraded performance. In addition, the number of cluster candidates will exceed 20 around 2001, which could induce timing problems in the level-3 (L3) trigger. Whether such background levels aect the physics capabilities of BaBar depends to a large extent on the physics channel considered; as for radiation damage to the crystals, it most likely will not be of signicant concern in the foreseeable future. L1 Trigger: the steady-state rate has to remain 50 % below the DAQ rate limit, in order to maintain a dead-time-free readout even during background spikes. This appears achievable, even at high currents and luminosity, by a combination of already-planned tracking-trigger improvements, level-3 and DAQ upgrades, and background reductions anticipated from PEP-II vacuum and collimation improvements. Overall therefore, the main operational concerns are the need to replace some of the SVT modules more than once during the lifetime of the experiment; and the diculty in simultaneously maximizing the DCH tracking eciency and its long-term background tolerance (trip rate, wire aging). The main physics impact of future backgrounds is likely to be in the area of low-branching ratio, pion- and photon-rich physics channels. The impact of higher backgrounds on the data volume and its relation to L3-trigger and background- lter rejection rates, fall outside the mandate and the competence of this Task Force. A luminosity as high as cm,2 s,1 would clearly stress or often exceed the capabilities of most BaBar systems as they are expected to exist about ve years from now. 2.3 Expected Impact of Planned Accelerator Improvements HER S2A Vacuum Upgrade During the upcoming Fall 2000 shutdown, the section of incoming HER beam pipe between -21 and -12 m will be replaced, and outtted with considerably increased pumping capacity. This will hopefully result (after scrubbing) in reducing the average pressure in that section by a factor of four to ve. Detailed pressure-prole and background-improvement calculations are still in progress; very rough, preliminary estimates suggest this could reduce the HERinduced SVT backgrounds by about 30 %, with smaller gains for the DCH and the EMC. 6

8 S2 Collimation Upgrade in HER and LER The above-mentioned HER vacuum upgrade is expected to reduce a dominant (resp. signicant) background source for the SVT (resp. the DCH). If this is successful, the existing adjustable collimator which is already in place in the HER, 12 m upstream of the IP, and is intended to reduce bremsstrahlung background generated in the next HER section (upstream of the upgraded vacuum pipe), may start becoming eective. An additional adjustable horizontal collimator will be installed this winter in the HER, 30 m upstream of the IP. This is designed to stop distant-bremsstrahlung electrons: for a 10 setting, Turtle simulations predict it will reduce the bremsstrahlung contribution from the HER arcs by a factor of 9, and at 7 the contribution from the rst 20 m of S2A (upstream of B3) is expected to drop by a factor of 2 [3]. This will mostly impact EMC backgrounds, and may also reduce somewhat the L1 beam-wall trigger rate. Finally - but this remains to be veried - this collimator might help further reduce the dynamic pressure in the upgraded S2A pipe by intercepting some of the synchrotron-radiation ux from upstream bends. In the LER, two adjustable horizontal collimators will be installed this winter at -25 and -12 m from the IP. The former complements the collimation of large-amplitude, distant Coulomb tails currently performed in IR4. The latter should deplete the opposite-phase Coulomb tails, as well as some of the bremsstrahlung positrons produced beyond -20 m. Together these two new collimators are expected to reduce the distant tails by about a factor of two compared to present levels; the reduction in the S2B bremsstrahlung contribution is comparable [3]. The expected projected gain from improved Coulomb collimation is largest in the DIRC, where it amounts to 10{ 20 % of the total background at high current and high luminosity; the DCH is also expected to benet, although somewhat less. Quantitative estimates of the impact of these collimators on the EMC are not yet available. 2.4 Analysis of Potential Long-term Improvements Scrubbing Scrubbing measurements have been performed by most subdetectors during the 2000 run. The cleanest signature is that of the DCH current, described in Sec During the rst month after the winter shutdown, backgrounds were observed to decrease by factors between 2 and 8, depending on the subdetector considered. After two months of running, no signicant changes in background level could be attributed to scrubbing (except for the level-1 trigger). Most of the quadratic dependence of the backgrounds on beam current died away during that time. For the case of the SVT and the L1 trigger at least, short-term scrubbing rates are known to be dominated by the last few ten meters in the HER and/or LER straights. Although long-term gains from further scrubbing in the detector straight (S2) cannot be excluded, it appears unlikely that these would signicantly aect the overall background picture. The DCH current and DIRC PM rate receive signicant contributions from distant Coulomb scatters in the LER. Based on an extrapolation of ring vacuum measured by ion pumps in the LER arcs as a function of integrated current (Sec. 4.3), dynamic-pressure reductions of more than a factor of two are not expected - if they occur at all. This would correspond to a decrease of about 15 % in DIRC backgrounds at the highest currents and luminosities. 7

9 Further Vacuum Upgrades At high current and luminosity, the dominant SVT background, and from a fth to a third of the backgrounds in the other subdetectors, are attributed to beam-gas scattering in the LER. Very preliminary Monte Carlo studies suggest that with the denite exception of the DIRC (and maybe that of the DCH), most of this radiation is generated in the incoming LER straight (S2B), possibly in the last 10 to 20 m. Such backgrounds cannot be collimated; but reducing the pressure in that section by a factor of two to four (if technically feasible) could provide much-needed headroom. This might, for instance, postpone the need for SVT module replacement by a year, reduce the DCH sensitivity to radiation bursts, or lower the impact of soft background photons on precious low-branching ratio physics channels. Further IP Shielding Upgrades The DIRC PM rate, the DCH current, and - to a lesser extent - the low-threshold EMC occupancy are dominated by luminosity-induced backgrounds. The sources of these showers are roughly known, at least under the assumption that the problem is caused by radiative- Bhabha products. More detailed measurements 3 and simulations of how the shower debris propagate, should be carried out in order to design enhanced IP shielding. While it may not be possible to achieve signicant background reductions for all three subdetectors, one might gain some valuable headroom for at least one of them. 3 using, for instance, the recently developped CsI sensors. 8

10 3 BaBar subdetector-specic issues 3.1 Silicon Vertex Tracker (SVT) The most signicant concern for the SVT with respect to machine backgrounds is integrated radiation dose. It is expected that an integrated dose of 1-2 MRad can dramatically aect the performance of some of the silicon modules through the process of type-inversion. The consequences of this process are not yet fully understood. Additionally, the front-end electronics will suer a modest increase in noise and a reduction in gain at 2 MRad. The SVT has installed a system of 12 PIN photodiodes mounted around the beam-pipe in an eort to more accurately monitor the delivery of radiation dose to the SVT bulk. The diode dose rates and dose integrals are the main tools by which the SVT gauges its radiation exposure. The standard gure of merit is therefore the total accumulated dose as measured by continuously integrating the instantaneous radiation dose rates in the SVT diodes. Another useful gure is the ratio of cumulative radiation dose to integrated luminosity, which measures how eciently the radiation budget of the SVT is being used to record data. Present running conditions are usually stable, albeit with strong departures during recovery from invasive procedures. Typically, the MID-plane diodes exhibit the highest dose rates due to coincidence with the bend-plane of the machine, averaging mr/s during stable beams (1.2 A LER x 0.7 A HER). The TOP and BTM diodes typically measure 2 mr/s during the same conditions (the East TOP and BTM diodes show a strong sensitivity to LER tunes, however, and can be as high as 7 mr/s). Because of their location, the most heavily exposed diodes are FE:MID and BW:MID. These are considered worst-case scenarios and used in the extrapolations below. The MID-plane diodes are also found to exhibit purely single-beam dependence, at the levels of measurement precision, with the East diodes being LER-sensitive, and the West diodes HER-sensitive. The frequency and quality of injection dramatically impacts the diode dose rates. Typically, the TOP and BTM diodes exhibit dose rates in excess of 30 mr/s during dual beam, 15 Hz injection each. The MID plane diodes are less sensitive (in part because of the much larger stored beam contribution), and report dose rates typically below 40 mr/s. Using data from dedicated background studies during June and July 2000, limited predictions are possible given a model of beam-current evolution. In separate dedicated experiments, no two-beam contribution to SVT diode backgrounds was measured. Although slightly surprising, this conclusion is built into the extrapolations. Quadratic dependence of diode dose rate on beam current has been assumed, although in almost all cases the quadratic predictions dier from pure linear extrapolation by less than a factor of 2. Dose rates have been converted to integrated dose by examining 1999 and 2000 performance and nding the equivalent Snowmass year for the SVT diodes. From the proposed current-evolution model, it is apparent that the FE:MID diode will be in grave danger of exceeding its lifetime radiation budget in 2004 (see Fig. 1 top left). The BW:MID diode will reach 1.5 MRad by 2004 (Fig. 1 top right), nearly compromising the radiation dose budget at the level of extrapolation error. The BE:TOP diode represents the worst of all of the other diodes, and is predicted to achieve an integrated dose of only 250 krad through 2004 (Fig. 1 bottom). The SVT is therefore most concerned about the exposure to the inner two layers of horizontal-plane silicon modules (a total of eight). The SVT diodes have been shown elsewhere to be dominated by local pressure and background sources. Distant collimation is expected to have a negligible eect on dose rates, 9

11 but local vacuum improvements that successfully combat the pressure within a few ten meters of the IP, could oer a direct and signicant improvement. The gain could, optimistically, scale directly with the pressure improvements. The SVT has scheduled a replacement of the horizontal plan silicon modules in the inner two layers in This would reset the dose integrals for those modules. Assuming that PEP-II delivers 640 krad per year after the replacement (an average of the 2004 and 2003 projected dose integrals), the new horizontal plane modules would last until 2006 when they once again exceed 2 MRad. Additional material and analyses can be found at the SVT background analysis Web site [4]. FE:MID Int. Dose / krad Y1999 (a) Y2000 Composition of SVT Diode Background LER HER Y2001 Y2002 Y2003 Y2004 BW:MID Int. Dose / krad Y1999 (b) Y2000 LER HER Y2001 Y2002 Y2003 Y2004 BE:TOP Int. Dose / krad Y1999 (c) Y2000 LER HER Y2001 Y2002 Y2003 Y2004 Figure 1: Top left: Integrated dose predictions using the beam-current evolution model described in the text for the most sensitive LER diode, FE:MID. Note that the accumulated dose reaches the 2 MRad lifetime limit in Top right: Integrated dose predictions for the most sensitive HER diode, BW:MID. Note that the accumulated dose reaches 1.5 MRad in Bottom: Integrated dose predictions for the most sensitive TOP/BTM diode, BE:TOP. Note that the LER contribution dominates by 2003, although the total dose integrated by 2004 is only 250 krad. 10

12 3.2 Drift Chamber (DCH) Machine backgrounds can aect the operation and usefulness of the Drift Chamber in three ways: The current drawn by the wires is dominated by the charge released by beam-related showers (in fact, since the chamber integrates these charges over a large central volume of the BaBar detector, the total DCH current is by far our most sensitive background monitor, an indispensable tool for many background studies). The high-voltage system can supply a limited amount of current. Above that limit, the Drift Chamber, and thus BaBar, becomes non-operational. The wire currents also contribute to the aging of the chamber. A wire is expected to deteriorate once the integrated charge collected on a wire exceeds approximately 0.1 C per cm of wire. The occupancy in the chamber due to machine backgrounds (single hits and tracks), can hamper the reconstruction of physics events. Ionizing radiation can permanently damage the read-out and digitizing electronics mounted on the rear end-plate of the chamber. From experience during the rst two years of running we know that only the rst item is, and will continue to be, a major concern. The rest of this section will therefore focus on the current drawn by the chamber. The Drift Chamber consists of 10 radial, so-called super-layers, each of which contains 4 layers of cells. Each super-layer is powered by four high-voltage supplies according to its segmentation into quadrants (top, bottom, inside and outside), except for the innermost super-layer, which is powered by eight supplies. Earlier studies have shown that the current per high-voltage supply shows a fairly at distribution as a function of both radius (layers) and azimuth (quadrants). The quadrant on the inside of the ring receives a large contribution from the HER whereas the LER produces currents predominantly in the outside, but also the top and bottom quadrants. The almost constant distribution in radius is understood as being due to a fortunate compensation of the falling current density and of the increasing number of wires per layer. The at distribution of currents means that the total current is an excellent gure of merit for backgrounds in BaBar (it is not dominated by one sector of the detector). It also means that the current limit for the entire chamber is not very far below the sum of the limits of the individual high-voltage supplies (less than a factor of two). The hardware current limit of a high-voltage supply is 40 A, currently we run at a trip threshold of 30 A. Thus a conservative estimate of the upper edge of the operational range for the total Drift Chamber current is about 1000 A. Over most of the 2000 run, the Drift Chamber has been operated at a voltage of 1900 V, slightly lower than design. It was recently (July 2000) switched back to the design voltage (1960 V) in order to improve tracking eciency. The increase in gain theoretically leads to currents 75% higher than at 1900 V; in practice, an increase of around 65% was observed. It should be emphasized that the measurements and extrapolations presented here are based on the lower voltage of 1900 V. At the moment it is not clear whether we will be able to go back to the lower voltage. Initial results from an improved amplier chip (stealth chip) do not show the expected increase in signal-to-noise ratio. It is however possible that advances 11

13 in track-nding software may allow us to return to the lower voltage without compromising track-nding eciency. The measurements performed during the summer of 2000 show an almost linear dependence of the Drift Chamber current on each of the two beam currents and on the luminosity. For the extrapolations we take into account small quadratic terms for the single-beam components. In the present conguration, the HER, the LER, and the luminosity contribute in about equal parts to the total Drift Chamber current. For a beam-current evolution according to the \High-Luminosity Model", extrapolations show that backgrounds from the LER and from the luminosity itself are going to dominate the Drift Chamber current (Fig. 2). When operated at the lower voltage, the Drift Chamber current will reach its pain threshold in 2004, at design voltage about two years earlier. The total integrated charge will reach around 0.05 C per centimeter wire in 2004 (1900 V). The Drift Chamber receives background contributions from both distant and local sources. Recent LER collimator tests have shown an overall background reduction of 30{40% due to collimation at IR-4. This reduction is included in the background extrapolations. HER backgrounds are dominated by the residual pressure in the incoming HER straight. Composition of Drift Chamber Background total DCH current [µa] Seeman model (V1 = 1900 V) Luminosity LER HER A.D. Figure 2: Total Drift Chamber current prediction using the "High-Luminosity Model", assuming a chamber voltage of 1900 V and no improvement. The operational limit of the chamber lies around 1000 A. Additional material can be found at the DCH background analysis Web site [5]. 12

14 3.3 Detector of Internally Reected Cherenkov Light (DIRC) The present background rates in the DIRC are signicantly higher than predicted in the TDR. They aect the DIRC DAQ in such a way that the readout eciency is rapidly degrading when counting rates exceed certain thresholds. With the present Front-End- Electronics TDC chips, the dead time reaches several percent as soon as DIRC scaler-rates exceed 300 khz; at 500 khz it climbs to 20 % and rises even faster above 700 khz Contributions to the DIRC Background Rate From the measurements performed in the summer of 2000, it became clear that the dominant source of background comes from a term proportional to luminosity. Fig. 3 shows a series of scans during which beams are moved vertically across each other: the correlation between the time-dependence of the luminosity signal and that of the DIRC Scaler-2 rate is apparent. Figure 3: Time-history of the luminosity signal (in green, bottom) and of a DIRC background scaler (in magenta, third from the bottom) during a series of beam-beam scans Plotting the measured counting rate vs. the luminosity (Fig. 4), reveals a linear correlation between the two, whose slope is independent of the beam currents and of the bunch pattern (the dierent vertical osets reect the single-beam contributions to the background). One therefore concludes that part of the DIRC backgrounds are luminosityinduced; counting-rate measurements by CsI counters placed around the beam pipes between quadrupoles QF2 and QD4, conrm this interpretation [6]. Given the high rate, the most likely candidate is radiative Bhabha scattering: energy-degraded positrons over-bent by the beam-separation magnets B1 and Q1, hit an aperture restriction somewhere between Q1 (end of the support tube) and Q5, leading to direct hits of electromagnetic shower debris in the water of the Stand-O Box (SOB). 13

15 Figure 4: DIRC background rate vs. luminosity during beam-beam scans performed with dierent beam-current combinations (labelled in the gure as L[ER]/H[ER]). The blue dots (L/H=200/100) represent data taken on July 17th, in a by-5 bunch pattern; the other data sets correspond to a by-4 pattern Fitting to an extensive set of single- and colliding-beam data yields the parametrization presented in Sec. 2.2: DIRC BG [khz] = 8:5 I HER +35I LER +25L Higher levels have often been observed at the top of lls, or when LER tunes drift away from their optimum; the expression above therefore reects running experience only under well-tuned conditions. Combining the above equation with the "High-Luminosity Model" time line, results in the background projection displayed in Fig. 5. The dominance of luminosity-induced backgrounds is apparent Potential Improvements The DIRC background studies are based on the analysis of the behavior of Scaler 2, which approximately corresponds to the horizontal inner plane of the detector. This area is believed to be the best-shielded one at this time. The new lead shield, which will be installed in the winter shutdown, will most likely not improve the protection of this particular area. However, as this new shielding will cover the full azimuth (in contrast to the temporary one currently in place), it should considerably reduce the background in other sectors, which at present record substantially higher rates than Scaler 2. It is dicult to quantitatively predict the improvement factor, as the shielding design was based not on a detailed Monte Carlo simulation, but rather on a succession of empirical improvements. 14

16 DRC Scaler 2 in khz Composition of DIRC Background Seeman model Luminosity LER HER Figure 5: DIRC background extrapolation, assuming no improvements A.D. A reduction of up to a factor of two in LER single-beam backgrounds, is expected from the installation of the new LER collimators in S2B. Given, however, that luminosity-induced backgrounds already dominate, the only avenues left are to increase the rate capability of the readout, and/or to improve the shielding of the SOB even further. To make the DIRC DAQ more robust, the group has started an upgrade program of the TDCs. The goal is to reduce the dead time to a few percent for rates around 1 MHz. Completion is scheduled for As predicted steady-state backgrounds reach the 300 khz level somewhat sooner, diculties will arise if large background uctuations - due for instance to beam-beam tails or tune-related instabilities in the LEB - start aecting DIRC performance before the electronics upgrade is completed. This underscores the importance of a continued measurement and simulation program to improve IP shielding near the SOB. Eventually, the electronics upgrade, together with additional local shielding between Q1 and Q2 and between Q2 and Q4, should allow satisfactory performance at even the highest luminosities. A detailed account of DIRC background studies can be found in Refs. [6] and [7]. 15

17 3.4 Electromagnetic Calorimeter (EMC) The general background issues facing the EMC are the integrated radiation dose, fake-neutral cluster occupancy and the eective loss of energy resolution due to the inclusion of energy from photons produced by machine backgrounds into EMC clusters produced by physics events. Integrated radiation dose contributes to light loss and non-uniformity in the calorimeter CsI crystals. Contributions to the radiation dose come from both machine backgrounds during injection and stable beams running, as well as from luminosity-related eects. The cumulative dose is monitored using an array of RadFET's distributed approximately uniformly over the front face of the EMC. The highest dose is observed in the horizontal plane in the innermost ring of the calorimeter endcap. This maximum dose is used as a gure-of-merit for radiation damage in the crystals. The calorimeter currently imposes a hardware threshold of 1 MeV on crystal energies, below which the DAQ system treats the energy as zero. Electronics noise in the crystal readout chain typically results in about 10% of EMC crystals being above this threshold in the absence of any energy deposition. Energy deposited in EMC crystals by a ux of lowenergy photons from machine backgrounds degrades the eective energy resolution of the calorimeter, since this energy may be incorrectly associated with an EMC cluster produced by a physics event. The average single-crystal occupancy above the 1 MeV threshold in background-sensitive regions of the EMC, can therefore be used as a gure of merit which characterizes this eect. If the energy spectrum of the background photons is known, the eect on the energy resolution can in principle be estimated. Calorimeter clusters are reconstructed following an algorithm which searches for crystals containing greater than 10 MeV of energy, and then computing a weighted sum of the energies of adjacent crystals which are above the 1 MeV threshold. This energy sum is required to be above 20 MeV for the cluster to be recorded. The expected number of calorimeter neutral clusters due to machine backgrounds is therefore dependent on the number of crystals containing more than 10 MeV of energy Current Status The HER and LER produce distinct patterns of energy deposition in the EMC with an average energy on the order of 1 MeV and a tail extending above 100 MeV. The HER produces a harder energy spectrum than the LER, and the two beams produce signicantly dierent patterns of energy deposition in the EMC. The HER produces the highest crystal occupancies in the EMC endcap in the horizontal plane towards inside of the PEP ring. Much lower occupancy is observed in the direction of the backward barrel. The LER produces high background rates in the backward barrel, peaking in the horizontal plane of the detector with a preference for the outside of the ring. The LER also produces occupancy in the innermost rings of the EMC endcap due to scattering o of the upstream Q2 septum chamber. The left and right plots in Fig. 6 show, respectively, the average single crystal occupancy above the 1 MeV threshold, and the number of crystals above 10 MeV as a function of the beam currents in the HER and LER. In both cases the background dependence on beam current is very nearly linear; however a quadratic dependence is assumed in order to give a more conservative extrapolation to higher currents. It should be noted that beam scrubbing resulted in a reduction in the EMC background rates between February and June 2000, consistent with that observed in other sub-detectors. 16

18 E > 10 MeV Single Crystal Occupancy above threshold (%) HER single-beam data LER single-beam data Number of crystals above 10 MeV HER single-beam data LER single-beam data Beam Current (ma) Beam Current (ma) Figure 6: Average occupancy of EMC crystals above the 1 MeV readout threshold (left), and number of crystals containing more than 10 MeV (right), as a function of HER and LER currents Data collected with both beams circulating simultaneously indicate that the calorimeter does not see a HER-LER cross term other than that attributable to luminosity. Fig. 7 shows the measured calorimeter background rates as a function of HER current, for a xed LER current of 1:1 A.Data taken with the beam separated at the IP produce backgrounds consistent with those produced in the single-beam experiments, while data collected with the beams in collision at the IP produced a higher rate. The high rate of increase of the number of crystals above the 1 MeV threshold compared to the number of crystals above 10 MeV suggests that the bulk of the luminosity contribution is due to very low angle radiative Bhabha events which shower into the EMC. The background rates can be parameterized as a function of the beam currents (A) and luminosity (10 33 )asfollows: Occ E>1MeV [%] = 9:8+2:2I HER +2:2I LER +1:4L N E>10MeV = 4:7 I HER +0:23 I 2 HER +2:4I LER +0:33 I 2 LER +0:6L Note that non-uniform distribution of backgrounds in the EMC produce occupancies in the hottest regions which are typically an order of magnitude higher than the mean Extrapolation to High Luminosity In the rst six months of 2000 the hottest regions of the EMC endcap received an integrated dose of approximately 100 Rad, while the barrel typically integrated 40 Rad in the same period. It is not known what fraction of this dose is attributable to injection and tuning and how much is from stable-beam running; however the endcap dose rate so far has scaled approximately linearly with HER current. Naive scaling of the observed dose rates to machine congurations for the \High-Luminosity Model" predict dose rates of less than 500 Rad/year anywhere in the calorimeter, comfortably within the TDR dose budget. Fig. 8 shows the expected contributions of the HER, LER and the luminosity to the EMC background rates extrapolated to future running conditions according to the \High- Luminosity Model" and assuming no other improvements in the machine performance. Note 17

19 Collisions data (I_LER = 1100 ma) Collisions data (I_LER = 1100 ma) Single Crystal Occupancy above threshold (%) Beams in collision Beams separated HER single beam data Number of crystals above 10 MeV Beams in collision Beams separated HER single beam data HER Current (ma) HER Current (ma) Figure 7: Average occupancy above the 1 MeV readout threshold (left), and number of crystals containing more than 10 MeV (right), as a function of HER current at a xed LER current of 1100mA, with the beams in (upper points) and out (lower points) of collision. The lower curve is the t to the single-beam data of Fig. 6, while the upper curve is a t of the luminosity term to the collisions data with the single-beam terms xed that the 10% occupancy due to noise has not been suppressed in the left-hand plot. The assumed quadratic dependence of the single-beam backgrounds on the HER and LER currents contributes about 30% of the total non-luminosity component by Although the EMC backgrounds are currently dominated by the HER, it is expected that the LER and luminosity contributions will be more important in the future. In the \High-Luminosity Model" extrapolation, the luminosity component due to zero-angle radiative Bhabhas dominates the low energy background, which will ultimately limit the EMC energy resolution. The higher energy component, which contributes to spurious neutral cluster formation, has large components from both the HER and LER. TURTLE ray simulation studies indicate that the EMC is sensitive to backgrounds from both beams due to bremsstrahlung events from relatively local sources (< 60 m from the IP). Additionally, the HER produces backgrounds from distant Coulomb and bremsstrahlung events, while the LER contributes through distant Coulomb scatters in which the positron strikes in the vicinity of either the upstream or downstream Q2 magnet. Since the HER and LER backgrounds have signicant components from both local and distant scattering events, it is expected that both HER and LER backgrounds could be reduced somewhat through improvements in either the local or distant vacuum. Additional shielding in the vicinity of Q2 could possibly reduce both single beam and luminosity related backgrounds Conclusions Integrated radiation dose is not expected to be a problem, assuming that the crystal light loss with integrated dose is consistent with expectations. Backgrounds are currently HER-dominated, but expect that in the future the LER will dominate, although the more localized pattern produced by the HER may result in a larger number of spurious neutral clusters. 18

20 Composition of Calorimeter Background Composition of Calorimeter Background Mean crystal occupancy above 1 MeV (%) Seeman model Luminosity Non-collision cross term LER HER Noise Number of crystals above 10 MeV Seeman model Luminosity Non-collision cross term LER HER A.D A.D. Figure 8: EMC backgrounds extrapolation, assuming no improvements The eect of the large numberoflow energy photons on the EMC energy resolution has not been studied; however it will in general depend on the clustering algorithm, digital ltering and other details of the EMC reconstruction chain. It is dicult to assess a priori what impact future background rates will have on EMC data quality; however this study suggests that the total energy will be dominated by the contribution from luminosity rather than from single-beam backgrounds. The background rates in the EMC in the \High-Luminosity Model" extrapolation appear to be large, but most likely survivable. Improvements in the vacuum near the IP, combined with eective collimation against positrons from distant Coulomb scattering in the LER are expected to be the most eective strategy for reducing these backgrounds in the future. Additional material and analyses can be found at the EMC background analysis Web site [8]. 19

21 3.5 Trigger The main concern of trigger backgrounds is the output rate of the rst-level trigger (L1). The data acquisition system (DAQ), as well the nal-level trigger (L3) need to be capable of handling this data volume. At present, both the L3 trigger and the DAQ are able to keep up with a L1 rate of up to 2.5 khz, measured for an event size corresponding to good running conditions. But as the DAQ rate limit is very sensitive to the event size, the growth in occupancy with increasing backgrounds could lower this limit to 2.0 khz or below. Based on the background uctuations experienced during the 2000 run, the L1 steadystate rate should not exceed 50 % of the rate limit above, if one wants to achieve deadtime-free running not only during the best machine conditions, but at all times. A factor of two headroom should be sucient to cope with background bursts or other rate spikes that suddendly double the trigger rate, last a couple of minutes, and cannot always be correlated with known changes in machine conditions. Besides the L1 rate, there might be concerns about the L3 output rate written to disk and the amount of oine computing needed to analyze the data. These issues could not be considered in the present report Contributions to the L1 Background Rate Most of the charged-particle trigger background is caused by inelastic scattering of lost beam particles with the material near the interaction point. The analysis of L1 pass-through events shows that the L1 tracking-trigger background is dominated by beam-wall interactions at the two anges between the beryllium beam pipe and the B1 magnets 20 cm from the IP (Fig. 9). Typical HER background originates on the inside of the ring, whereas the LER background comes from the outside of the ring. The background is dominantly in the horizontal plane. Single Beam Background - HER Number of L1 Passthrough Tracks Longitudinal Position in Beam Direction (cm) Figure 9: z 0 distribution of L1 pass-through tracks at their point of closest approach to the beam line (HER single-beam data at 700 ma, recorded on July 6, 2000) 20

22 Fits to single- and colliding-beam data yield the following parametrization: L1 [Hz] = 130 (cosmics) I LER I HER +70L A graphical representation of the contributions to the L1 rate is shown in Fig. 10. The underlying assumption of linearity in the currents and luminosity is a good approximation for a stable and well-tuned machine with the present vacuum conditions: in early 2000, there were signicant quadratic contributions from both the HER and LER currents, which progressively faded away as scrubbing progressed. Composition of Trigger Background total L1 Rate [Hz] Seeman model Luminosity (lin) LER (lin) HER cosmics A.D. Figure 10: L1 rate extrapolation, assuming no improvements At present, the HER background is twice as high as that of the LER, and the luminosityrelated trigger rate is only 40 % of the combined HER and LER single-beam backgrounds. With the machine proceeding to higher currents, the LER single-beam and luminosity terms will become more important. By 2004, the LER and HER single-beam backgrounds are expected to contribute equally, with a luminosity-related term comparable to the sum of the rst two. At least 90 % of the luminosity-related trigger rate can be attributed to large-angle Bhabha scatters into the ducial volume of the detector (rather than to the shower debris from radiative Bhabhas hitting aperture limitations, that are responsible for the DIRC and DCH luminosity backgrounds). Such events are not of general physics interest and are only used for calibration and alignment purposes. The Bhabha rate is currently prescaled at level 3; this could in principle be done at level 1 already. It turns out, however, that the L1 discriminating power is insucient to uniquely identify Bhabha events without losing eciency in other physics channels: to avoid impacting two-prong physics, only one-prong 21