Detector simulations for EIC at BNL Alexander Kiselev

Detector simulations for EIC at BNL Alexander Kiselev POETIC Workshop Sep, 2 -5 2013

Contents § Dedicated e. RHIC detector concept § Eic. Root framework § § General idea Tracking studies Calorimeter R&D Tools for detector development § EIC detector solenoid field modeling § Update on EIC smearing generator § Summary Sep, 3 2013 A. Kiselev

e. RHIC detector concept Sep, 3 2013 A. Kiselev

Interaction Region design y E A. Kiselev 1 4 m Sep, 3 2013 m x ions 678 ns 1 0 m bea 733 P 0 m Ion 0. 1 el 0 m 4. 5 ole Dip rad s 0 mtron q=1 neu 1 2 rad 2 m ro ect m 2 8 3 3 m 0. 8 1 m 0. 8 6. Q= 2. 5 6 4 3 m m. 512 1 m m. 9 c 2. 5) / (p o E 4. 5 93 0. 9 rad m 10 2 1. 5 5 m q= I Qu 1. 9 5. 4 OU ad d ine mb ion o C nct fu gnet ma m 75 ID S T ad Qu 0. 1 ed n lig m a oid bea n n le So ith io w z Z D C 0. 3 2 m ID INS rad 3 m 0. 0 q=1 1 6 Electron quadrupoles

Interaction Region design Q 4 D 5 m q=4 3 m q=3. 6745 mrad 10 m 75 m 5. 4 4. 5 10 15. 1 m d q=9. a r m 20 30 40 50 60. 0559 m 90. 08703 m Sep, 3 2013 A. Kiselev 0. 44843 m d ra m 2 8 0. 316 m 8 8. 1 Q 5

e. RHIC detector layout (Ee ~ 20 Ge. V) Sep, 3 2013 A. Kiselev

e. RHIC detector layout (Ee < 10 Ge. V) Sep, 3 2013 A. Kiselev

e. RHIC detector view (August’ 2013) Forward EMC RICH Central EMC SOLENOID Backward EMC n Sep, 3 2013 EMC and tracking detectors A. Kiselev ~implemented so far RICH

Eic. Root framework Sep, 3 2013 A. Kiselev

EIC detector in Fair. Root framework n q q q Fair. Root is officially maintained by GSI O(10) active experiments O(100) users n n n Fair. Base n Interface to GEANT Interface to ROOT event display Parameter database MC stack handling … Panda. Root n n n Interface to Fair. Base classes Ideal track finder Interface to Gen. Fit … Generator output file import Fast smearing codes Sep, 3 2013 Fopi. Root Eic. Root n n eic-smear A. Kiselev solenoid modeling TPC digitization Real life track finder(s) Another Gen. Fit interface Interface to RAVE

End user view n No executable (steering through ROOT macro scripts) simulation digitization reconstruction -> MC points -> Hits -> “Short” tracks -> Clusters n n “PID” Pass -> “Combined” tracks -> Vertices @ IP ROOT files for analysis available after each step C++ class structure is well defined at each I/O stage Sep, 3 2013 A. Kiselev

Tracking studies Sep, 3 2013 A. Kiselev

Requirements for EIC main tracker n n n As low material budget as possible Complete geometric coverage in the h range [-3. . 3] Momentum resolution not worse than ~3%, even at large rapidities Sep, 3 2013 A. Kiselev

Tracking code implementation n n Magnetic field interface exists Detector geometry is described in 0 -th approximation: n n n n Silicon vertex tracker Silicon forward/backward tracker TPC GEM forward/backward tracker MRS-B 1 solenoid design per default Digitization exists (simplistic, except for TPC where Fopi. Root codes allow to perform a complete digi chain) “Ideal” track reconstruction inherited from Panda. Root codes Real-life track finding&fitting as well as vertex finding&fitting inherited from Fopi. Root codes (work in progress) Sep, 3 2013 A. Kiselev

Vertex silicon tracker n n MAPS technology; ~20 x 20 mm 2 chips, ~20 mm 2 D pixels STAR detector upgrade “building blocks” (cable assemblies) Sep, 3 2013 A. Kiselev

Vertex silicon tracker n n n 6 layers at [30. . 160] mm radius 0. 37% X 0 in acceptance per layer simulated precisely digitization: single discrete pixels, one-to-one from MC points Sep, 3 2013 A. Kiselev

Other tracking elements forward/backward silicon trackers: n n n 2 x 7 disks with up to 180 mm radius N sectors per disk; 200 mm silicon-equivalent thickness digitization: discrete ~20 x 20 mm 2 pixels TPC: n n n ~2 m long; gas volume radius [200. . 800] mm 1. 2% X 0 IFC, 4. 0% X 0 OFC; 15. 0% X 0 aluminum end-caps digitization: 1) idealized, assume 1 x 5 mm GEM pads; 2) complete (Fopi. Root source codes adapted, GEM pad shape tuning in progress) GEM trackers: n n n 3 disks behind the TPC end-caps STAR FGT design digitization: 100 mm resolution in X&Y; gaussian smearing Sep, 3 2013 A. Kiselev

EIC tracker view (August’ 2013) Backward Silicon Tracker Forward Silicon Tracker Vertex Silicon Tracker TPC Backward GEM Tracker Sep, 3 2013 A. Kiselev Forward GEM Tracker

Example plots from tracking code 1 Ge. V/c p+ tracks at h=0. 5: <ndf> = 206 32 Ge. V/c p+ tracks at h=3. 0: <ndf> = 9 Sep, 3 2013 -> look very reasonable from statistical point of view A. Kiselev

Momentum resolution plot #1 p+ track momentum resolution vs. pseudo-rapidity -> expect 2 -3% or better momentum resolution in the whole kinematic range Sep, 3 2013 A. Kiselev

Momentum resolution plot #2 p+ track momentum resolution at h = 3. 0 vs. Silicon thickness -> ~flat over inspected momentum range because of very small Si pixel size Sep, 3 2013 A. Kiselev

Momentum resolution plot #3 p+ track momentum resolution at h = 3. 0 vs. Silicon pixel size -> 20 micron pixel size is essential to maintain good momentum resolution Sep, 3 2013 A. Kiselev

EIC solenoid modeling Sep, 3 2013 A. Kiselev

EIC solenoid modeling main requirements: n n n Yield large enough bending for charged tracks at large h Keep field inside TPC volume as homogeneous as possible Keep magnetic field inside RICH volume(s) small -> use OPERA-3 D/2 D software Presently used design: MRS-B 1 Sep, 3 2013 A. Kiselev

EIC solenoid modeling Other options investigated, like 4 -th concept solenoid design -> obviously helps to cancel “tails” Sep, 3 2013 A. Kiselev

EIC smearing generator Sep, 3 2013 A. Kiselev

General architecture MC generator output Django h PYTHIA PEPSI Rapga p Sep, 3 2013 eic-smear MC tree code: Smearer: Builds ROOT Performs fast tree detector containing smearing events DPMjet Milou • gmc_tran C++ code running in ROOT s • Builds with configure/Make • Single libeicsmear. so to load in ROOT A. Kiselev

Functionality built in n n Easily configurable acceptance definitions Kinematic variable smearing declarations (single) quantity, X, to smear: E, p, θ, φ + Function defining σ(X) = f([E, p, θ, φ]) + Acceptance for X in E, p, θ, φ, p. T, p. Z either a priori knowledge of detector resolutions is needed or parameterization based on a full GEANT simulation -> try out resolutions provided by Eic. Root fits … Sep, 3 2013 A. Kiselev

Hadron identification with RICH consider hadrons in pseudo-rapidity range ~[1. 0. . 3. 0] -> pion/kaon/proton separation must be possible up to momenta ~40 Ge. V/c Sep, 3 2013 A. Kiselev

Migration in (x, Q 2) bins 10 Ge. V x 100 Ge. V beams -> “survival probability” is above ~80% in the region where tracking has superior resolution compared to calorimetry Sep, 3 2013 A. Kiselev

Eic. Root Interface to eic-smear q Eic. Root input n n q directly uses eic-smear library calls to import ASCII event files after MC generators … … as well as “unified” ROOT format event files Eic. Root output n n is available in eic-smear format with charged particle momentum variables “smeared” by Kalman Filter fit after track reconstruction … … while other variables modified by smearing generator according to its recipes Sep, 3 2013 A. Kiselev

Calorimeter R&D Sep, 3 2013 A. Kiselev

EM calorimeter requirements n n n Complete geometric coverage in the h range [-4. . 4] Very good (~1 -2 %/√E) backward end-cap (electron going direction) resolution Moderate (~10 -12 %/√E) resolution forward and central parts High granularity (angular resolution better than 1 o) Compactness of barrel calorimeter Sep, 3 2013 A. Kiselev

Calorimeter code implementation n Written from scratch Unified interface (geometry definition, digitization, clustering) for all EIC calorimeter types n Rather detailed digitization: n n n configurable light yield exponential decay time; light collection in a time window attenuation length; possible light reflection on one “cell” end Si. PM dark counting rate; APD gain, ENF, ENC configurable thresholds Sep, 3 2013 A. Kiselev

Backward EM Calorimeter (BEMC) n n 10 Ge. V/c electron hitting one of the four BEMC quadrants Sep, 3 2013 PWO-II, layout a la CMS & PANDA (but no cooling) -2500 mm from the IP both projective and non-projective geometry implemented digitization based on parameters taken from PANDA R&D Same event (details of shower development) A. Kiselev

BEMC energy resolution plot #1 electrons at h = 2. 0 -> projective geometry may lag behind in terms of resolution? Sep, 3 2013 A. Kiselev

BEMC energy resolution plot #2 non-projective geometry; h = 2. 0 n “Realistic” digitization stands for: light yield 17 pe/Me. V; APD gain 50, ENF 2. 0, ENC 4. 2 k; 10 Me. V single cell threshold; Sep, 3 2013 A. Kiselev

Forward EM Calorimeter (FEMC) n n n STAR e/m calorimeter upgrade building blocks (tungsten powder scintillating fiber sampling technology) 1 mm fibers; sampling fraction for e/m showers ~2. 6% +2500 mm from the IP; non-projective geometry “medium speed” simulation (up to energy deposit in fibers) reasonably detailed digitization; “ideal” clustering code Sep, 3 2013 A. Kiselev

FEMC energy resolution study 3 degree track-to-tower-axis incident angle n “Realistic” digitization stands for: 40 MHz Si. PM noise in 50 ns gate; 4 m attenuation length; 5 pixel single tower threshold; 70% light reflection on upstream fiber end; -> good agreement with original MC studies and measured data Sep, 3 2013 A. Kiselev

FEMC tower “optimization” original mesh -> optimized mesh design can probably decrease “constant term” in energy resolution Sep, 3 2013 A. Kiselev optimized mesh

Central EM Calorimeter (CEMC) -> barrel (central) calorimeter collects less light, but response (at a fixed 3 o angle) is perfectly linear n n n same tungsten powder + fibers technology as FEMC, … … but towers are tapered non-projective geometry; radial distance from beam line [815. . 980]mm Sep, 3 2013 A. Kiselev

CEMC energy resolution plot #1 3 degree track-to-tower-axis incident angle -> simulation does not show any noticeable difference in energy resolution between straight and tapered tower calorimeters Sep, 3 2013 A. Kiselev

CEMC energy resolution plot #2 8 Ge. V/c electrons -> energy response goes down with polar angle because of effectively decreasing sampling fraction; quite reasonable Sep, 3 2013 A. Kiselev

Eic. Root as a tool for detector development Sep, 3 2013 A. Kiselev

Eic. Root availability n n SVN -> http: //svn. racf. bnl. gov/svn/eicroot eic 000* cluster -> /eic/data/Fair. Root n n Sep, 3 2013 README & installation hints Few basic usage examples A. Kiselev

Tracker “designer” tools § Allow, among other things § to add “simple” tracking detector templates to the “official” geometry § to try out various magnetic field maps § Require next to zero coding effort Which momentum resolution for 10 Ge. V/c pions will I get with 10 MAPS layers at h=3? -> see tutorials/designer/tracking directory for details Sep, 3 2013 A. Kiselev

Tracker “designer” tools -> workflow sequence: n Create geometry file (few dozens of lines ROOT C script) Include few lines in “standard” sim/digi/reco scripts: n Analyze output ROOT file n Sep, 3 2013 A. Kiselev

Calorimeter “designer” tools § Allow to easily add “simple” calorimeter detector templates to the “official” geometry § Require next to zero coding effort Which energy resolution for 1 Ge. V/c electrons will I get with a “basic” PWO calorimeter? Sep, 3 2013 A. Kiselev

Calorimeter “designer” tools § As long as the following is true: n n your dream calorimeter is a logical 2 D matrix … … composed of “long cells” as elementary units, all the game is based on (known) light output per energy deposit, energy resolution after “ideal” digitization suffices as a result § … one can with a moderate effort (99% of which is writing a ROOT C macro with geometry and mapping description) build custom Eic. Root-friendly calorimeter which can be used for both standalone resolution studies and/or as an optional EIC device (and internal cell structure does not matter) -> see tutorials/designer/calorimetry directory for details Sep, 3 2013 A. Kiselev

Calorimeter R&D (by-product for PHENIX upgrade) Sep, 3 2013 A. Kiselev

PHENIX upgrade calorimeter setup n n tungsten scintillating fiber epoxy sandwich default configuration: n n 1 mm fibers; 7 fiber layers per 20 x 20 mm 2 “tower” sampling fraction for e/m showers ~2. 3% Pure tungsten metal sheet (r ~ 19. 3 g/cm 3) Thickness ~ 0. 5 -1. 0 mm 10 cm Tungsten powder epoxy (r ~ 10 -11 g/cm 3) Scintillating fibers ~ 0. 5 – 1. 0 mm Sep, 3 2013 A. Kiselev

PHENIX calorimeter energy resolution 1 mm fibers + ~2 mm thick tungsten plates -> poor energy resolution at small incident angles! Sep, 3 2013 A. Kiselev

PHENIX calorimeter energy resolution 1 mm fibers + ~1 mm thick tungsten plates -> with 1 mm plates energy resolution becomes acceptable Sep, 3 2013 A. Kiselev

Summary Sep, 3 2013 A. Kiselev

n While smearing generator is still the main physics simulation tool for the group, Eic. Root development is in a good progress and code is already suitable for several types of studies: n n n detector acceptance tracking resolutions e/m calorimeter design optimization First physics simulations in Eic. Root framework should become possible really soon Other immediate code development goals include n n implementation of IR (material and fields) implementation of PID algorithms (RICH, TPC d. E/dx, …) Sep, 3 2013 A. Kiselev