NRA 97–OSS–13

NRA Title:	Detector Definition and Instrument Assessment for an Advanced Cosmic-Ray Composition Experiment on the Space Station (ACCESS)
Principal Investigator:	Thomas Alfred Parnell ES84. Astrophysics Branch NASA-Marshall Space Flight Center (MSFC) Huntsville, AL 35812, USA (205) 544–7690 FAX (205) 544–7754 E-Mail: Thomas.A.Parnell@msfc.nasa.gov
Proposal Title:	An Imaging Calorimeter for ACCESS (ICA)
Co-Investigators:
Paul L. Hink	Washington University	hink@wuphys.wustl.edu
Robert W. Binns	Washington University	wrb@wuphys.wustl.edu
Martin H. Israel	Washington University	mhisrael@wuphys.wustl.edu
John C. Gregory	University of Alabama in Huntsville	jcgregory@matsci.uah.edu
Geoffrey N. Pendleton	University of Alabama in Huntsville	pendlgn@sslab.msfc.nasa.gov
Yoshiyuki Takahashi	University of Alabama in Huntsville	takahashi @ssl.msfc.nasa.gov
James H. Adams	Naval Research Laboratory	adams@crs2.nrl.navy.mil
M.J. Christl	NASA/MSFC	mark.christl@msfc.nasa.gov
J.H. Derrickson	NASA/MSFC	jim.derrickson@msfc.nasa.gov
W.F. Fountain	NASA/MSFC	walt.fountain@msfc.nasa.gov
J.W. Watts Jr.	NASA/MSFC	john.watts@msfc.nasa.gov
Simulation Consultants:
Leonard W. Howell	NASA/MSFC	leonard.howell@msfc.nasa.gov
Jeongin Lee		jeongin.lee@msfc.nasa.gov
K. Asakimori	Kobe Women's Junior College	asakimori@kobe-wu.ac.jp
A. Chikanian
M. Fuki	Kochi University
H. Yokomi		yokomi@tezukayama-u.ac.jp
R. Munroe	University of Mobile
Authorizing Official: Dr. Gregory S Wilson /ES01/MSFC/NASA

This version has minor editing to correct typographical errors, emphasize some points, and reduce redundancy. Any differences from the submitted version ( other than omissions) are indicated by italics.

Principal Investigator: Dr. Thomas A. Parnell

Proposal Title: An Imaging Calorimeter for ACCESS (ICA)

The baseline Advanced Cosmic-Ray Composition Experiment for the Space Station (ACCESS) instrument includes a thin ionization calorimeter of Bismuth Germanate (BGO) scintillators which is the principal instrument for measuring the high energy spectra of hydrogen and helium, and providing supplementary energy measurements for C–Fe. Principal science objectives are to test the SN shock acceleration models and propagation models by supplying reliable spectra to energies approaching the "knee" in the all-particle spectrum. We propose to study an alternative design, a thin imaging calorimeter composed of lead (Pb) sheets interleaved with layers of small scintillating optical fibers. The advantages of the proposed Imaging Calorimeter for ACCESS (ICA) include:

1) It can be made thinner without sacrificing energy resolution and has greater collecting power for the same weight. (2) It will provide accurate information on the trajectory of the primary cosmic ray, reducing the effects of back-scattered particles on measurement of the primary’s charge. (3) It will provide more definition of the longitudinal profile of each calorimeter shower, with better separation of the first-interaction shower from successive ones, enabling uniform energy resolution over a wide range. (4) The segmentation of the target in ICA with scintillating fibers allows more definitive comparison with simulations for data analysis and allows better diagnostics for mitigating the effects of Landau-Pomeranchuck-Migdal (LPM) effect and the p° decay Lorentz time dilation. (5) ICA is likely to be less expensive than the BGO baseline.

The ICA carbon target is ³ 1.0 m² in area, and £ 1.0 proton interaction lengths thick. The departure from the NRA baseline is that it is sampled by x, y pairs of square scintillating fibers. These fibers are 2 mm thick and provide the approximate position of the interaction. The ³ 1m² calorimeter has » 25 rl of lead and is sampled with 29 x,y pairs of 500 mm square scintillating fibers. The upper 3.0 rl is sampled each 0.5 rl, and the remainder each rl. The total weight of target + calorimeter is ~2,600 kg.

With simulations performed for this proposal, we describe the characteristics of the ICA calorimeter configuration. We describe the simulations that we plan for this investigation to quantify the advantages of the ICA approach and arrive at an optimum design for the ACCESS calorimeter. We will participate in the full ACCESS instrument assessment as appropriate. We also plan to perform sufficient instrument and engineering definition to develop a conservative design and cost estimate and to support the ACCESS instrument engineering study at GSFC.

REFERENCES:
Asakimori, et al, 25^th ICRC, V4, 1 (1997); submitted to Ap. J. Aug. (1997)
Christl, M.J. et al, SPIE Proceedings, V2806, 155 (1996);

Proposal Abstract	Page 2
Table of Contents	Page 3
I. INTRODUCTION	Page 4
II. THIN IONIZATION CALORIMETRY	Page 5
III. THE IMAGING CALORIMETER FOR ACCESS	Page 6
1. Configuration of ICA 2. Numbers of Events
IV. ADVANTAGES OF ICA/CHARACTERISTICS TO BE STUDIED	Page 8
1. Advantages and Characteristics of Thin Imaging Calorimetry to be Studied A. Core of Shower Peaks Earlier B. Greater Collecting Power for Fixed Weight C. Accurate Projection of Shower to Charge Detectors D. Optimize the Measurement of the First Interaction Cascade E. 3-Dimensional "Image" May Confer Advantages F. Secondary Interactions in Target G. Maximum Practical Energy Range H. LPM Effect I. Lorentz time-dilation of p° Decay J. Dilution/Transition Effect K. Knock-on Electrons; Spallation in fibers 2. Relevant Calorimetry Development and Experience A. SOFCAL B. SIFTER C. ACE/CRIS D. TIC
V. FIBERMAN ICA INSTRUMENT DETAILS	Page 14
1. Optical Read-Out System 2. Dynamic Range Issues and Solutions 3. Data System 4. Probable Costs (Image System)
VI. CALIBRATION METHODS	Page 16
VII. THE PROPOSED STUDY: SIMULATIONS AND TRADE-OFF	Page 16
1. Science Performance Studies 2. Simulations and Trade Studies 3. Simulation Techniques Currently Available 4. Improvement in Simulation Techniques to be Developed 5. Instrument Studies/Engineering Support
VIII. REFERENCES	Page 18

I. INTRODUCTION

The baseline Advanced Cosmic-Ray Composition Experiment for the Space Station (ACCESS) instrument includes a thin calorimeter of Bismuth Germanate (BGO) scintillators for measuring the energy spectra of hydrogen, helium, and some nuclei heavier than helium. We propose to study an alternative design, a thin imaging calorimeter composed of lead (Pb) sheets interleaved with layers of small scintillating optical fibers. The advantages of the proposed Imaging Calorimeter for ACCESS (ICA) include:

(1) It can be made thinner without sacrificing energy resolution. Consequently, it has greater collecting power for the same weight.

(2) It will provide more accurate information on the trajectory of the primary cosmic ray, reducing the circle of confusion in the charge-determining detectors, and thus reducing the effects of back-scattered particles on measurement of the primary’s charge.(Important for discriminating protons and helium.)

(3) It will provide more definition of the longitudinal profile of each calorimeter shower, with better separation of the first-interaction shower from successive ones. This confers several advantages, the principal one being more uniform energy resolution over a wide range.

(4) The segmentation of the target in ICA with scintillating fibers allows more definitive comparison with simulations (required for design and data analysis) and also allows better diagnostics and compensation for two effects of concern to thin calorimeters at high energies, the Landau-Pomeranchuck-Migdal (LPM) effect and the p° decay Lorentz time dilation (section IV).

(5) ICA is likely to be less expensive than the BGO baseline.

The ACCESS instrument team previously has developed performance and engineering criteria for meeting the scientific objectives and for accommodation to the Space Station (including transportation to it) [ACCESS Ref.]. The ICA will meet or exceed these criteria. In section III the collecting power described in the baseline document is compared to our Monte-Carlo results for two "strawman" configurations of the ICA.

The baseline ACCESS instrument contains three separate kinds of detectors to investigate a range of questions concerning the cosmic ray source(s), galactic acceleration mechanism(s), and propagation containment volumes and parameters. For some of the scientific objectives the instruments make complementary measurements. The primary role of the ionization calorimeter is to measure the spectra of H and He over a very broad energy range from ~ 100 GeV to > 10¹⁵ eV. The calorimeter would also measure the light elements and C-Fe nuclei below the threshold of the Transition Radiation Detectors (TRD), to the extent allowed by the overlying material.

The principal scientific objectives of the ionization calorimeter measurements concern the identification of the bend in the cosmic ray spectrum at ~ 3 ´ 10¹⁵ eV as either the end of the galactic acceleration mechanisms [Lagage , 1983], a feature that is produced or enhanced by leakage of cosmic rays from the galaxy (propagation), an artifact of change from "direct" measurements to air shower measurements, a change in composition or interaction characteristics; or a combination of effects. A definitive measurement of the H/He (and heavier nuclei) spectra will also allow the type I and type II SN remnant acceleration process to be tested [Stanev 1993, Bierman 1994]. The Li/Be/B to CNO abundance as a function of energy will allow containment volumes to be tested [Takahashi, 1997]. In order to contribute to the resolution of these questions, it is necessary to have a calorimeter that responds over a large energy range since it is known that the H and He (and heavier nuclei) spectra differ from below 100 GeV to above 10 TeV. An adequate exposure factor is necessary for testing the predicted limit to galactic SN shock wave acceleration at energies of ~Z ´ 100 TeV [Lagage, 1983], for which the highest energy direct measurements [Asakimori, 1997] have no clear evidence. The calorimeter must also have adequate energy resolution to indicate a relatively sharp bend in the elemental spectra. The configuration we have chosen has these characteristics.

This proposal addresses two categories in the solicitation: 1. Detector development, and 2. Instrument assessment. We describe in more detail in section IV the advantages of the ICA and thin imaging calorimetry characteristics we plan to study. Section V is a description of the scintillating fiber detectors, a candidate imaging system and read-out data system, and variations of this baseline that will be studied. In section VII we discuss simulation techniques we have available to initiate the category 1 studies. We also discuss improvements to be made in the simulation techniques. As needed, we will apply the simulations to the complete ACCESS instrument assessment. We will also support the engineering studies at GSFC and JSC.

II. THIN IONIZATION CALORIMETRY

A frequently applied method to measure the energy of a high energy cosmic ray is to cause it to interact in a thick target and reveal its kinetic energy through the creation of a shower composed of a large number of secondary particles. The secondary particles are predominantly relativistic and singly charged, so their collective signal is proportional to their number. The principal advantage of this method is that it works over a very broad energy range. If the calorimeter is deep enough to contain the "total" shower, an accurate measurement of the cosmic ray energy can be made. The weight of such a deep calorimeter is prohibitive for a space experiment in the ACCESS energy region.

A different approach must be used, one which relies on the fact that hadronic cascades develop an accompanying electronic cascade through the creation of gamma rays from p° meson decay. Because the interaction mean free path for hadrons and mesons is very much longer than the radiation length (rl) in high-atomic number (Z) material such as lead (30:1), one can usually develop the electromagnetic cascade before the residual primary or the highest energy secondary mesons or hadrons have a chance to interact again. For this reason, in a deep lead calorimeter, one sees the cascade develop as a series of "bursts" of energy deposition following the successive interactions of the most energetic hadrons. The objective of the thinnest calorimetry technique is to distinguish the first burst while delaying subsequent large interactions and associated bursts.

In addition to the advantages of a high-Z absorber, the shower maximum occurs at a shallower depth when one measures the shower within a small radius around its axis, which further separates the first interaction energy,å E_g , from successive "bursts". The principle of the radial dependence of the shower development can be found in three dimensional cascade theory [Nishimura,1967], in which the similarity law is given:

,

where K denotes the scattering constant (= 19.6 MeV in lead). N is number of electrons, E₀ is primary energy, r is the sampling radius, t is depth in radiation lengths. This means that the shower curve for measurements within a small radius is similar to the one-dimensional shower curve (r-> infinity) for a smaller shower energy. Correspondingly, the depth of the shower maximum will be reduced approximately by ln (r₀/r), where r₀ is a function of the Moliere length. Although the similarity law strictly holds for radii less than one Moliere length, the depth of the shower maximum is still reduced for larger radii as well. The similarity law applies to single gamma ray cascades, but the general trends also apply to the overlapped cascades from high energy interactions.

The similarity law was experimentally proven with electron beams up to 300 GeV at Fermilab [Hotta, 1980], and has been frequently applied in the emulsion chamber method. The emulsion chamber was first used by Kaplan and Ritson in 1952 for high energy shower measurements with small radius sampling, followed by systematic measurements with balloon experiments [Minokawa, 1958]. The JACEE experiments use this method with radii from 50 to 150 microns [Burnett, 1986]. Scintillating fibers are currently available down to 50 mm diameters, but these are not practical for large calorimeters. Nevertheless, the 500 mm fibers for ICA give a significant decrease in shower depth (section IV).

The energy measured in a thin calorimeter is a mixture of the shower energy from the interaction in the target å E_g (1) and those of secondaries in the calorimeter å (S _g (2)). The principle of the thinnest calorimetry is based on measuring the shower from the first interaction by using a thin target and a thin sampling calorimeter. If the energy resolution is constant over the range of measurement, the following relation holds:

,

where J_oE_o^{-b -1}dE_o denotes the primary energy spectrum and f(k_g ) is the inelasticity distribution for p^o production [Burnett,1986]. f(k_g ) is derived from accelerator experiments and simulations above accelerator energies. Preliminary simulations for this proposal indicate that the ICA energy resolution in å E_g may be » constant toå E_g =275 TeV (IV D.). An objective of this study is to accurately quantify the resolution and determine the best deconvolution technique to derive the primary spectrum.

Thick calorimeters (³ 1 mfp) have to take account of the large contribution of second, third, etc. interactions. These will appear in , where the second term fluctuates from event to event as the inelasticity distribution of protons is almost flat from 0 to 1.0. A frequently sampled calorimeter observes these fluctuations and can measure å E_g (1) reliably. The starting point for our study uses 0.5 rl sampling for the first 3.0 rl and 1 rl sampling thereafter. Preliminary simulations (in IV) indicate this may be adequate.

For thick targets, primary interactions near the top of the target will develop nuclear cascades in the target, and their secondary interactions cause the measurement of the first interaction å E_g (1) to be more contaminated by the fluctuating secondary contributions. The ICA design study includes a range of target depths for energy measurements of protons. The final choice of target thickness and sampling will be made by simulating the å E_g spectrum and subsequently recovering the input primary spectrum.

Thin active calorimeters have been flown in space [Grigorov,1963,1989] and on balloons for measurement of cosmic rays [Jones,1977], [Adams,1996]. A thin, fully active calorimeter is under development [Guzik, Seo, 1996] for cosmic ray spectrum measurements. Thin active imaging calorimeters for cosmic rays are currently being flown on balloons [Christl, 1996] or under development for high energy gamma-ray studies [Pendleton,1996]. Hybrid imaging techniques have also been used on balloons for cosmic ray electron measurements [Torii,1996]. Several authors of this proposal have been active participants in the development of cosmic ray calorimetry techniques for the past 20 years.

The concept we propose to develop for ACCESS is based on the JACEE [Burnett,1986,Parnell 1989] and SOFCAL [Christl,1996] experience with thin imaging ionization calorimeters. It utilizes the precise tracking properties of scintillating optical fibers to take advantage of the shower structure and provide an accurate prediction of the incident particle’s trajectory. It also incorporates high-Z absorber material for the calorimeter which optimizes the å E_g (1) measurement.

The configuration that we show in the next section is a starting point for our simulations and studies, and is chosen to accentuate the features we believe the study should explore. Trade-offs between target and calorimeter depth, fiber sampling size and frequency, total collecting power, energy resolution, reliable energy range, etc. must be made. The table below lists the key parameters of the ICA Strawman (Fiberman) and the NRA Baseline.

**Principal variables for proposed ICA study

III. THE IMAGING CALORIMETER FOR ACCESS

We describe the two "fiberman" configurations of the ICA to illustrate several aspects of thin calorimetry that must be explored to optimize the imaging calorimeter. We first discuss the physical parameters of these configurations and compare their calculated counting statistics with the Baseline BGO calorimeter as described in the NRA and referenced documents.

1. The ICA Fiberman configuration is illustrated in Figure 1. Fiberman 1 uses a carbon target. The departure from the NRA baseline is that it is sampled each 10 cm (1/4 interaction length) by x, y pairs of square scintillating fibers. These fibers are effectively 2 mm thick and provide the approximate depth of the interaction. The calorimeter has 25 rl of lead and is sampled with 29 x,y pairs of 500 mm square scintillating fibers. The upper 3.0 rl is sampled each 0.5 rl, and the remainder each rl. Fiberman 2 is similar, but the target is only 20 cm (0.5 l _p) thick, sampled with scintillating fibers each 5 cm (1/8 l _p). The target weight reduction allows a modest (32%) area increase The two configurations are summarized in Table III.1. The imaging system weighs 150 kg. (section V.).

†Allows greater depth, if necessary *No. 2 is within the NRA baseline weight.

Figure 1. A schematic of the Fiberman configuration (target and calorimeter). The target for Fiberman #1 uses 10 cm thick graphite slabs while Fiberman #2 uses 5 cm thick slabs.

2. The number of events for a three year mission have been calculated for these two configurations and compared with those for the NRA baseline [ACCESS REF]. They are listed in table III.2.

The NRA Baseline assumes that the particles must pass through the silicon charge detectors on the UH module. This condition is designated the "primary" aperture in Table III.2. Despite particle back-scatter problems close to the calorimeter , it may be possible to develop charge detectors that cover the wider aperture offered by ICA. We will explore the possibility of an auxiliary charge detector (figure 1) during the study phase. The second set of numbers, designated the "full" aperture, are the events should this extra charge module be included.

Table III.2. Proton and Helium Events: Expected number of events in a 3-year exposure for the NRA baseline design and the ICA fiberman designs above a total energy E_total³ 500 TeV. The events are tabulated according to where they first interact (UH + TRD, Target, Calorimeter). The primary spectra incorporates a spectral change of -0.30 at Z´ 100 TeV (Values in parentheses are for spectra extended from low energy data without spectral breaks).

Nuclei Events: Expected number of events in a 3-year exposure for the NRA baseline design and ICA fiberman#1 design (fiberman #2 has similar event numbers). The appropriate interaction cross sections were used for the CNO and Fe spectra, however the proton cross section was used for the All-particle spectrum calculation. (Note: all three ACCESS baseline instruments are sensitive to primary particles with Z ³ 6).

IV. ADVANTAGES OF ICA TO BE CONFIRMED AND OTHER CALORIMETRY EFFECTS TO BE STUDIED.

1. The ICA configuration has significant differences from the NRA baseline. We discuss below some obvious advantages of the ICA, possible advantages that need to be confirmed in the study, and a number of other factors and physical effects that must be addressed.

A. The shower reaches a maximum at a shallower depth the smaller the sampling radius. Figure 2 shows simulation results assuming circular apertures about the shower axis for a lead calorimeter similar to the fiberman.

Figure 2. A simulation of the average of ten cascades in a calorimeter similar to the fiberman design from a 500 TeV proton interacting at the top of the calorimeter. The simulation method is MCM + ECM as described in section VII. The number of electrons in rings of various radii are shown, showing the decreased depth of shower maximum for smaller sampling radii.

Figure 3 shows simulations in which 500 mm square scintillating fibers are used in the calorimeter and energy deposition in the fibers is calculated. The initial event is a proton with 500 TeV total energy. We note that the fibers sample a strip through the cascade and their fine-sampling characteristics must differ somewhat from circular apertures (as used in emulsion chambers). In figure 3., the energy deposition from the total shower and also the central two fibers (1 mm) are shown. Since the central two fibers also sample the outer wings of the shower, the result of a method to approximate three dimensional sampling (» 3D) is shown in the third curve. The energy deposition in the second two fibers from the center are subtracted from the first two. This » 3D cascade curve peaks at ~20% smaller depth than the total shower, which has about the same fractional improvement as reducing the sampling radius from 500-12.5 mm (in figure 2). This approximate method may be improved with three-dimensional shower profile reconstruction(E. below).

Figure 3. The simulation of energy deposition in 500 mm square scintillating fibers in a calorimeter similar to ICA. Cascade 1 gives the total energy deposition (´ 0.1) in a plane of fibers. Cascade 2 is the energy deposition in the two fibers nearest the shower core. Cascade 3 is the central two fiber signals minus the next adjacent ones. The simulation method is MCM +EGS3* as described in section VII. The shower is from a proton at 500TeV total energy.

B. The reduced depth requirement can be traded off between weight reduction or increased geometry factor.

C. The location of the shower core within ~500mm (at many depths in the calorimeter) allows accurate projection of the cascade up to the charge detection elements. This has an advantage in minimizing the effects of back-scatter particles on charge measurement in the Primary Identification Detectors (PID).

D. The calorimeter configurations we have chosen enhance the use of the electromagnetic cascade from the first interaction of the primary for the energy analysis. This confers the advantage that the fluctuations of the first interaction cascade energies are relatively stable with energy, over the range to be measured with ACCESS. The energy resolution of this method as determined by emulsion techniques at lower energy is about 25% in å E_g for protons [Burnett,1986]. We have performed a preliminary simulation at å E_g =275 TeV. The results are displayed in Figure 4 and Figure 5. This is an example of the calculations that must be performed to demonstrate the behavior of the energy resolution over the ACCESS range. Note that simulations performed for this proposal are vertical events, the worst case for calorimeter depth considerations.

Figure 4. The distribution of åE_g derived from simulation (MCM in section VIII) which first samples events from a hypothetical proton spectrum of E-2.7 from 500 to 5000 TeV.

Figure 5. A simulation in the ICA calorimeter geometry of 20 showers from the 250-300 TeV åE_g bin in figure 4. The shower maxima value in an experiment would be one measure to determine åE_g . For this small sample of events, the estimated dispersion is s~20%. The simulation method was EMC and the sampling radius was 500 mm.

E. The fine-sampling of the calorimeter contains information about the three-dimensional development of the shower, the application of which is to be explored during the study. As the energy increases the shower becomes symmetric, statistically "smooth", and centrally peaked. This allows for reconstruction of the three-dimensional cascade from the fiber data to reduce the core depth measurement (1 above), and to compensate for saturation of fiber signals by large showers when in a high gain mode (for calibration or measuring the low energy part of the spectrum). Figure 6 shows the intensity profile along a plane of fibers for a 500 TeV proton event.

Figure 6. The energy deposition in 500 mm square fibers across the plane near shower maximum for a 500 TeV proton at locations before and after shower maximum.

F. A reduced target thickness (Table III.1) and the scintillating fiber detectors within it will aid in reducing the effects on energy analysis of secondary interactions in the target. These secondary interactions contribute to fluctuations in the electromagnetic cascade maximum from the first primary interaction. Measuring the approximate depth of the interaction with fiber signals allows more detailed application of the simulations to quantify the effects of the secondary interaction fluctuations, and to compensate for them in the analysis should this be necessary.

G. A critical question for ACCESS is the maximum practical energy for the calorimeter technique chosen, subject to physical constraints (weight). A method that depends on the first interaction cascade must demonstrate that such cascades can reliably be recognized and measured in the calorimeter. We have with the techniques currently available, and within the computer time limits available for the proposal, simulated some very high energy cascades in the ICA candidate calorimeter configurations. Figure 7 shows 10 of these cascades at vertical incidence for a primary proton energy of 5,000 TeV. It is obvious that most of the cascades have a clear maximum at less than 25 rl depth at this energy.

Figure 7. A simulation of ten vertical showers from primary 5000 TeV protons in the "approximate ICA" geometry. The sampling radius was 500 mm and the method was MCM + ECM (section VII), which contains the LPM and p^o decay dilation. The average of these shower profiles is shown by the solid circles.

These cascades are started at the bottom of target and both the p^o decay effect (I. below) and the LPM effect (H. below) contribute to the depth of the maxima. These effects must be thoroughly studied using the interaction height and the zenith angle distribution as some of the input parameters.

H. The Landau-Pomeranchuk-Migdal (LPM) effect [Landau 1953, 1965] describes a change in the energy loss of electrons and photons in electromagnetic cascades that becomes significant above primary energies of » 10¹⁴ eV. The effect is due to the increase in the laboratory-frame energies of the virtual photons surrounding the highest energy electrons in a cascade, and the suppression of the "low energy" radiative processes due to successive disturbances (collisions) of the cascade particles. The LPM effect suppresses bremsstrahlung and pair production by the highest energy electrons and photons [Nishimura,1997]. This elongates both the shower core and the total (linear) shower. The effect has been studied for air showers [Stanev,1985] and lead calorimeters [Fuki,1993].

Previous calculations for lead-emulsion calorimeters indicate that for primary protons at 10¹⁵ eV, the average shower maximum is delayed by about 2 radiation lengths, and for 10¹⁶ eV protons about 5-6 radiation lengths. The preliminary simulations for this proposal indicate that 25 rl should be adequate to contain this effect for vertical showers at å E_g =500 TeV.

I. The p° mesons produced in interactions up to about 100 TeV decay within 1 cm of the interaction, but above several hundred TeV primary energy the Lorentz time-dilation of the p° decay becomes a significant factor for thin calorimetry [Fuki,1993]. The p° decay mean path length L is:

The effect on the shower profile is dependent on the energy distribution of the produced p° s. In addition to increasing the depth of the first interaction cascade, it also contributes to significant longitudinal and peak fluctuations. Since it affects longitudinal development, a physically deeper target and calorimeter is favored. Because of the large longitudinal fluctuations, a finely segmented calorimeter to recognize the successive cascades (or "bursts") is also favored.

The effect of the p° time dilation on the thin calorimeters of the JACEE experiments (8 rl of lead) has been calculated [Fuki,1993] and the results are shown in figure 8 a and b. The influence of the LPM effect is also displayed. The solid line is the assumed proton energy spectrum. The points are the simulated å E_g spectrum. This shows that for the 8 rl calorimeter, for interactions that occur 5 cm above the calorimeter, the measured å E_g spectrum has an artifact indicating a bend in the spectrum around 1000 TeV. The division of the target into thinner slabs (and zenith angles) would allow selection of events to mitigate these problem for the highest energy events. The imaging calorimeter will also allow identification of "exceptional" events from p° decay dilation (see figure 9). Some of our simulation programs (section VII) contain the LPM and p° delay effects, and they may be incorporated in the others.

Figure 8(a)	Figure 8(b)

Figure 8. (a) A simulation of the å E_g spectrum generated in an emulsion chamber 8 rl deep exposed to a E^-2.7 proton spectrum. The protons are assumed to interact in a target 5 cm above the calorimeter. A simulation method similar to MCM + ECM in chapter VII was used. The effects of the LPM and p° decay dilation are obvious. (b) Same as (a) except proton spectrum assumed to have a bend at 50 TeV. An isotropic flux was assumed.

Figure 9(a)		Figure 9(b)

Figure 9(a) Two cascades from vertical protons at 10¹⁶ eV total energy indicating p° decay dilation. The simulation was MCM+ECM.  Figure 9(b) Integral shower curves for the two events in 9(a).

J. The "transition" effect refers to a change in a shower when it passes from the absorber (e.g., Pb) into a detector material (e.g., scintillator). The "dilution" effect occurs when a space is introduced between the absorber layers. Both effects lower the electron density in the shower core. The transition effect depends on the critical energies (E_c) of the absorber and detector, and it has been studied for various kinds of calorimeters [Pinkau,1968]. With thin scintillators the effect is moderated by back-scatter of electrons from the next absorber layer. Simulations will predict these effects, and their accuracy will be checked in the study phase.

K. Knock-on electrons with significant energies and ranges are produced by hadrons and mesons at the rate of a few percent and will add to the scintillator signals. This effect will be evaluated. The spallation of nuclei in the scintillator or lead surface would produce "stars" depositing ~100 MeV in the fiber. If found significant, shower reconstruction should remove it.

2. Relevant Calorimeter Development Experience

A. The Scintillating Optical Fiber Calorimeter (SOFCAL) Instrument

The SOFCAL balloon-borne instrument is a hybrid detector system that uses a passive nuclear emulsion chamber in tandem with a scintillation fiber sampling calorimeter [Christl,1996]. It was designed to study the low energy cosmic ray proton and helium spectra below and overlapping the energy threshold of an emulsion chamber (~0.1-10 TeV). Passive calorimeters use X-ray films for event triggers, trajectory information and energy measurements [Burnett,1986]. However, the energy threshold of these x-ray films is ~ 1 TeV in å E_g (~10 TeV primary energy for protons). The scintillating fibers serve a similar purpose (threshold detection and cascade energy measurement) as the x-ray films but are sensitive to much lower energy events. The 7 rl fiber calorimeter (0.25 m² in area) comprises ten 4 mm (0.7 rl) lead plates and 10 xy pairs of 500 mm square fibers. We describe here a few features that are relevant to ICA.

The raw images of a shower in the fiber calorimeter initiated by a proton interacting in the emulsion chamber are displayed in figure 10. These orthogonal views of the shower allow the trajectory of the event to be reconstructed using the center of gravity and least squares fitting techniques. The analysis of events in the emulsion chamber require predicting their location within <1 mm to find these small cascades with microscope scans of the emulsions. This has been accomplished for showers that are well below the energy threshold of the x-ray films.

Figure 11. Longitudinal shower profile from SOFCAL event in figure 10. The depth of the shower maximum is reduced as the number of fibers used for the signal is reduced.

The transition curve for the same event is displayed in figure 11. This data has not been corrected for optical coupling, imaging detector variations, or fiber position. Four different sampling widths along the fiber planes are shown. These widths correspond to summing the signals from 2, 10, 20 , and 60 fibers about the event centerline. The shower maximum and its depth is determined from fitting the data to calculated transition curves.

The total ensemble of showers recorded are used for in-flight monitoring, calibrations, and position and optical variation corrections. The heavy nuclei that pass through the instrument are used to check intensity calibrations of the fibers. The SOFCAL active imaging calorimeter was an inexpensive training tool which serves its function in this hybrid instrument very well. Several of its performance features will be considerably improved in the ICA. Another type of image intensifier will eliminate the optical cross-talk or blurring of the images. The read-out system will have a much wider dynamic range. The fiber to intensifier coupling will have a more uniform fiber-to-fiber output.

B. SIFTER The Scintillating Fiber Telescope for Energetic (Gamma) Radiation (SIFTER) is an instrument under development to measure the direction and energy of gamma rays from ~20 MeV to ~10 GeV and is part of the Gamma Ray Large Area Space Telescope (GLAST) technology study. The high energy part of the SIFTER measurement will be made with a calorimeter with features similar to ICA. SIFTER will use a lead-scintillating fiber calorimeter with fibers approximately the same cross-section and length (~500 mm and ~ 1m) as ICA and a similar image-intensifier-CCD. These techniques have been under development for about a year and test components (fibers, II’s) are being assembled for an accelerator test in 1998. ICA will benefit from the technological developments of SIFTER.

C. ACE/CRIS The Cosmic Ray Isotope Spectrometer (CRIS) [Hink, 1997] experiment which was launched on Aug. 25, 1997, utilizes a scintillating fiber hodoscope with an image intensified CCD camera system very similar to that proposed for ICA. The CRIS fiber system is working very well and has demonstrated that such a system can be successfully launched and operated in space.

D. TIC The Thin Ionization Calorimeter (TIC) is a simple sampling iron-scintillator calorimeter. It measures the total energy deposited in order to measure the all-particle spectrum. The TIC has had one 76 hour balloon flight and has been calibrated with 2.3 TeV Au ions at BNL and 0.8 TeV protons at FNAL [Adams 1997].

V. FIBERMAN ICA INSTRUMENT DETAILS

The Fiberman #2 ICA system consists of three sections: the Target, Upper Calorimeter, and Lower Calorimeter. The Target section consists of four slabs of graphite, each 5 cm thick, for a total of 0.5 proton interaction lengths. These graphite targets are separated with five planes of scintillating fibers. Each fiber plane is composed of eight layers of 0.5 mm square scintillating fibers having a polystyrene core and a bonded multi-clad sheath. The multi-cladding consists of an acrylic inner clad of 25 mm, and a fluorinated acrylic outer cladding of ~15 mm. The fiber plane is configured with four layers of fibers orthogonal to the other four layers, i.e., 4-X and 4-Y fiber layers per plane. With the readout system described below, the X, Y detection efficiency for minimum ionizing particles (MIP) in the target section is 92%. The active area of the target section is 115 cm x 115 cm. The scintillating fibers are coupled to non-scintillating fiber light pipes [Bossert 1993] at the edge of the target active region to allow the light signal to be conveniently routed and coupled to the readout system. This minimizes contamination of the signal due to out-of-geometry particles, both primaries and secondaries. The target fibers could be 2 mm square fibers, however ease of routing to the readout system favors the use of multiple layers of thinner fibers. Part of the proposed study will include an optimization of the fiber size, total thickness of scintillator per plane, fiber readout segmentation, and fiber readout system.

The Upper Calorimeter (UC) consists of six plates of 0.5 rl Pb (or other high-Z material) with one x,y plane of 500 mm square scintillating fibers directly below the plate. The Lower Calorimeter (LC) consists of 22 plates of 1 rl Pb with one x,y plane of 500 mm square scintillating fibers directly below the plates. The active area of both the UC and LC is 115 cm x 115 cm. The fibers are readout through non-scintillating fiber light pipes. Although the x,y detection efficiency for a MIP in the UC and LC fiber planes using the baseline readout system is only ~55%, the energy reconstruction will be sufficient given the number of MIPs in the central portion of the shower. These fibers will be readout on each end with gains chosen to yield an overall dynamic range of about 10⁵. Part of the study will include ways to increase the MIP detection efficiency of the fiber planes, including alternative readouts, increased fiber thickness (i.e., additional layers), etc. Additional planes of fibers may be added to ICA and readout with PMT’s for use in forming the event trigger.

The baseline ICA readout system is based on the image intensified CCD (II-CCD) camera readout being tested for the SIFTER instrument. This system (figure 12) was chosen since it uses currently available technology and satisfies the baseline requirements of the ICA. In the calorimeter both ends of the fibers will be readout with a high or low gain II-CCD. The fibers in the target will be silvered on one end and coupled to a high gain II-CCD on the other end. Both the high gain and the

low gain systems have a common first stage, a "Generation 1" electrostatic reducing image intensifier. This intensifier has fiber optic input and output windows, a bi-alkali photocathode, a photon gain of ~50, a "fast" phosphor such as P46 (100 ns decay and peak emission of 440 nm), and reduces an 80 mm input image to 25 mm at the output. We currently are testing an image intensifier of this type, manufactured by Photek for the SIFTER program.

Figure 12. ICA fiber imaging system.

The output of the first image intensifier is coupled to an II-CCD camera system similar to one developed by Washington University for the Cosmic Ray Isotope Spectrometer (CRIS). For ICA this camera system utilizes a Photek 25 mm diameter microchannel plate (MCP) image intensifier having a S-20 or bi-alkali photocathode, fast phosphor, ms gating, and fiber-optic input and output windows. For the high gain system a dual MCP plate device would be used, with a single MCP plate device for the low-gain system. The high-gain system would be capable of detecting single photoelectron signals.

We will also develop a method of obtaining a signal proportional to the total light pulse viewed by the image intensifier for use in forming the event trigger. This has been demonstrated for a similar system [Charon,1991], although not for use as an event trigger. We propose to demonstrate the operation and performance of this anode trigger as part of this study.

The output of the 25 mm image intensifier is coupled to a Thomson 7866 CCD array using a glass fiber optic reducer. The CCD is a 244 x 550 pixel frame transfer array with both an anti-blooming structure and quick-clear capabilities. The anti-blooming structure prevents "bleed-over" of signal from one pixel to adjacent pixels when a signal exceeds the pixel well depth of about 2.5 x 10⁵ electrons. This is critical for obtaining a high quality profile of the shower core for the highest energy events. The quick-clear feature uses the anti-blooming structure to remove charge from the CCD. The quick clear time is about 3 ms, and will be performed once every several hundred ms to remove dark current and background signals from the CCD. Triggers will be inhibited during this quick-clear time. Immediately following an event trigger, the MCP image-intensifier is gated off, the CCD information is transferred to the image zone, and then digitized using a 10–12 bit flash ADC. Pixels exceeding a digital threshold are passed onto the data system for further processing. The total readout time is 16.6 ms. We will evaluate other CCD devices for use on ICA, including several devices having readout times of ~1 ms.

The total readout system will consist of 24 high gain and 14 low gain image intensified CCD cameras. PMT’s may also be part of the system for use in forming the event trigger, coupled to non-imaged fiber planes as described above. The master trigger will be formed by requiring that at least one x,y pair of anode pulses (or PMTs) be above a threshold equivalent to the interaction of ~100 GeV proton. The formatting of the image intensifier systems will be optimized based on results of simulations performed as part of this study. Currently, we plan on grouping all target fibers together, in horizontal sections of 23 cm. The calorimeter fibers will be divided into seven horizontal sections of 17 cm. The addition of several planes of scintillators readout by PMT’s would allow for a flexible definition of an event, and reject most background events. The expected event rate will be ~5 cps with a 100 GeV threshold. Allowing for out-of-geometry events and other forms of background, we estimate a trigger rate of 15 cps. With a readout time of 16.6 ms this yields a livetime of about 75%. We plan to increase the threshold to a few TeV after several weeks of operation since event statistics will be sufficient for the 100 GeV - 10 TeV region. With this higher threshold, the event rate will be <5 events per minute, and CCD deadtime is negligible.

The data system will consist of multiple CPU’s acquiring and compressing the data collected by each camera system. Processing will include summing pixel data into fiber intensities using on-board fiber-to-pixel maps. A study of compression methods will be performed, using results of event simulations, to optimize the on-board processing and telemetry requirements of the ICA readout system.

VI. CALIBRATION METHODS

The ICA detector ideally would be calibrated by accelerator beams. However, neither Fermilab (800 GeV) nor the CERN LHC (10 TeV) have plans for external beams.

1. Since no direct calibrations above 10 TeV will be possible, the response will depend upon simulations and a reference for the scintillation light output compared to 1 MIP. This is easy for the 2 mm fibers in the target, but the ~1 meter length (500 µm) fibers in the calorimeter will not have 100% detection efficiency for 1 MIP throughout its length (optical attenuation length » 2.0 m). However, fiducial signals for He and for heavier nuclei at lesser calorimeter depths can be used. Fiducial light pulse sources and a particle sources on fibers have been used in the SOFCAL experiment to monitor gain changes and both µ mesons and pass-through He and heavier nuclei have been used to determine the one MIP reference.

2. For a significant number of heavy nuclei, Z = 6 and above, the particles that transit the TRD in its sensitive range will supply a calibration of the energy response of the calorimeter. The comparisons with pre-flight å E_g calculations will supply a check that the heavy nucleus interaction models we use are valid. Since new data should be available from RHIC soon, these models can be updated, but the å E_g fixed target measurements we make with ICA will include center of mass angles not easily accessible with colliding beam experiments. The number of events that can be used (within the TRD dynamic range) are: 1700 CNO (E_total 30-300 TeV), and 70 Fe (E_total 100-1000 TeV).

VII. THE PROPOSED STUDY: SIMULATIONS AND TRADE-OFFS

1. Science Performance Studies

The two-year study must arrive at an optimum configuration for the target and calorimeter within the ACCESS constraints (weight, power, funding, etc.). Due to the many factors involved "optimum" translates to the best compromise. Although small laboratory studies on instrument topics are planned, most of the proposed study will involve simulations. The major issues are counting statistics (geometry factor), energy resolution, and reliable performance at the energy range limit of the statistics, balanced with instrument mass and practical accommodations to the spacecraft and other instruments (UH and TRD).

2. Some of the numerous simulations and trade studies to be performed are summarized here:

A. Geometry—Efficiency Factor Calculations with all "strawman" configurations including interactions in the UH and TRD instruments, interaction rates at various depths the in target, and usable interactions in the calorimeters. (e.g., those with ³ 25 rl remaining after interaction). These calculations should also generate some of the factors to correct the detected flux to the primary flux.

B. Hadronic—Electromagnetic cascade calculations of the generic types shown in figures

2–9. A statistically significant number of events simulated at representative energies from ~100 GeV to ~10¹⁶ eV primary energy. These must be performed with a variety of interaction depths in the target and calorimeter and at various "zenith" angles from 0° to ~70° . These simulations will be used to determine the minimum practical calorimeter depth, the necessary sampling intervals, and the definitive dynamic range required of the read-out. The result of these simulations will allow the selection of a final configuration for detailed analysis.

C. After selecting the "optimum" configuration, many of the above calculations will need repeating in more detail to establish the definitive energy resolution. The main product of this study is the best method of determining the primary spectrum from the å E_g spectrum. Both the thinnest calorimetry method [å E_g (1)] and thin calorimetry (DE) methods will be applied, along with associated deconvolution methods to derive the primary spectrum.

D. The number of particles scattered backwards in the calorimeter changes significantly over the energy range covered here. We must determine their effects on the primary identification detectors (PID), the sampling fibers in the target, and optical components of the ICA. We can also provide valuable input to the UH and TRD definition teams in this regard.

3. Simulation Techniques Currently Available

There are several simulation codes that are available to initiate the study.

A. Geometry—Efficiency Factor Methods. The geometry factor for different target and calorimeter configurations will be examined using a Monte Carlo code based on GEMFAC. This code traces CR trajectories through the proposed configuration and accepts some events based on hitting the detectors, interacting and developing a shower past shower maximum before exiting the calorimeter. It uses a formula for the mean distance from the first interaction to shower max that is derived from fitting cascade simulations. Several variants of this method are available in the ICA team.

B. Multi-chain Interaction Model plus EMC Electromagnetic Cascade Model [MCM Ref.][EMC Ref.]

These simulations employ essentially the same models for hadron-hadron and nucleus-nucleus interactions as those used previously for some emulsion chamber experiments. In this model, the position of the nucleons in the target nucleus and in the projectile nucleus are sampled in the center of mass system according to the nuclear density function of each nucleus. After the nucleon positions and impact parameter have been determined, the nucleon-nucleon cross-sections are used to determine collision pairs of nucleons. A nucleon-nucleon collision is simulated for each pair. Successive collisions are allowed for incident proton or nucleons, but produced secondary hadrons do not interact with nucleons in the target. Both projectile and target "fragmentation" effects are included. Following a nucleon-nucleon interaction, the produced particles consist of pions, kaons, nucleon-antinucleon pairs, and eta particles in the approximate ratios 85%, 6%, 1%, and 8%, respectively. The fluctuation in the produced particle numbers is taken into account and they are sampled randomly around mean values according to the binomial distribution. The result of the MCM is a list of produced particles, their energies, and their laboratory production angles.

The EMC [EMC Ref.] electromagnetic cascade code starts with the gamma rays from the MCM list and generates each photon-electron cascade. As presently configured the ultimate output is the position of each cascade electron on a series of target planes, and the number of electrons within a circle of given radius (e.g., 100 mm, 500 mm). This output configuration is matched to X-ray film and emulsion measurements in emulsion chambers, but must be modified for general application to fiber calorimeters. The EMC method includes the LPM effect in the code. For p° ’s a routine exists in the code to simulate the Lorentz time dilation of the p° decay.

The codes used for this proposal also transport all hadrons and mesons from the first interaction and generate secondary interactions according to their mean-free paths. Further electromagnetic cascades are generated from p° produced in these secondary interactions.

C. Multichain Model (MCM) and EGS-3.*

For simulating energy deposition and electron numbers in the fibers, the MCM as described above is linked to EGS-3*. This code is used to generate electron numbers and energy depositions in the linear fibers. This combination of codes has the shortcoming that the LPM effect is not currently included in the codes. However, EGS–3* has been modified in to include the Lorentz time-dilation of the p° decay.

D. GEANT is in general use by the particle physics community to model accelerator experiments. It is equipped to handle complex geometries, and it correctly describes all the known physical processes affecting particle transport through matter and fields up to the highest accelerator energies. It includes time-dilation for the p° decay (Lorentz time dilation), but the electromagnetic cascade codes do not include the LPM effect. We will use GEANT to examine the backscatter signal in the PID plane. We will use the GCALOR routine in GEANT. This routine is based on the Oak Ridge codes, HETC and MCNP, which incorporate the details of neutron transport which are important in this calculation since the target and the instrumentation above the calorimeter are composed of light nuclei that moderate the fast neutrons from the nuclear interactions.

4. Improvements in the Above Simulation Methods for this Study

A version of the GEANT code developed at NRL includes a wounded-nucleon model for heavy nuclei that will be use in the ICA study. The EGS 3 code that has been used with GEANT does not contain the LPM effect or p° decay dilation effect. The latter has been installed as a subroutine in the EGS 3* and we will examine the feasibility of using this with GEANT. The LPM effect may be taken into account by empirical methods.

Our simulations team will apply cross-checks between simulation methods, including the latest developments in shower simulations [Knapp,1996].

5. Instrument Studies/Engineering Support

Most of the instrument components (fibers, optical imaging systems, data systems, and mechanical arrangements) are under active study for the SIFTER experiment. SIFTER studies include the potential for thinner fibers and alternate read-out devices such as avalanche photo-diodes. The ICA has some separate issues that will be examined through calculations or laboratory studies. These include optimizing methods for the large dynamic range, alternative imaging devices, and methods of energy calibration (absolute fiber light output). A study will be performed of the methods of mechanical fabrication of the ICA assembly, and the effects of launch dynamics. The instrument assessment study at GSFC will be supported. A conservative mechanical design concept and reliable cost estimate are goals of the study.

REFERENCES

[ACCESS Ref: Instrument Team http://hep.uchicago.edu/~swordy/access.html and Team X reports]
Adams, Jr. J. et al, Proc 25th ICRC, Vol. 3, pp 401 (1997)
Asakimori, K. et al, ApJ, Submitted (1997)
Asakimori, K. et al, 25th ICRC, Vol. 4, pp1 (1997)
Bierman, P.L. 23rd ICRC Rapporteur Papers (1994)
Bossert, et al, Conf of Scintillation, Fiber Det. Oct. 1993
Brun, R. etal, GEANT Manual, CERN Geneva (1994)
Burnett, T.H. et al, NIM A251, 583 (1986)
Burnett, et al, Ap J 349, 25 (1990)
Charon, Y. et al, NIM, A310, 379 (1991)
Christl, M.J. et al, Proceedings of SPIE, V2806, pp 155 (1996)
[EMC Ref.] An empirical-Monte Carlo method developed by Okamato, M. and Shibata, T. NIM A257, 155 (1987) with modifications by Fuki, M. and Roberts, E.
Fuki, M. et al, 23rd ICRC, Vol. pp (1993)
Grigorov, N.L. et al, JETP, V45, p 1544 (1963)
Grigorov, N. L. et al, JETP, V49, p71 (1989)
Guzik, T.G. et al, Proceedings of SPIE, V280b, pp 122 (1996)
Hink, P. et al, Proc. SPIE 2806 199 (1997)
Hotta, N. et al, Phys. Rev. D, Vol. 22, No. 1 (1980) pg. 1
Jones, W.V. et al, NIM, 140, 557 (1977)
Kaplan, M.F. et al, Phys Rev, 85, 900 (1952)
Knapp, J., et al, Forschungszentrum Karlsruhe, Institute für Kernphysik, December 1996.
Lagage, P.O. et al, A&A 118, pg 223 (1983)
L.D. Landau, The Collected Papers of L.D. Landau, Pergamon Press 1965.
L.D. Landau and I.J. Pomeranchuk, Dokl. Akad. Nauk USSR, 92, 535 (1953).
[MCM ref] Kinoshita et al, Prog. Theor. Phys 61, 1,165) 1979, Zeitshrift fur Phys. C8, 205 1981, with additions by M. Fuki (Thesis)
Migdal, A.B., Phys Rev. 103, 1811 (1956)
Minokawa, D., et al, Nuovo Cimento, Suppl 8, 761 (1958)
Nishimura, J., Handbook der Physik, V62, 1 (Springer-Verlag, Berlin) 1967
Nishimura, J. et al, 25th ICIG, Vol. pp (1997)
Parnell, T.A. et al, Adv. Space Res. 9(12), pp 45 (1989)
Pendleton, G.W. et al, Proceedings of SPIE, V2806, pp 164 (1996)
Pinkau, K, Phys. Rev. B, 139, 1548 (1968)
Seo, E-S et al, Proceedings of SPIE, V2806, pp 134 (1996)
Stanev T. et al, Astron. and Astrophys. 274, 902-908 (1993)
Takahashi Y. et al, Nuc Phys A, in press (1997)
Torii, S. et al, Proceedings of SPIE, V2806, pp 145 (1996)