Deat Dr. Krizan, We are thankful to our reviewers for their time and useful comments. We addressed all questions/comments and modified the manuscript accordingly. Our answers are below. Please let me know if we need to improve (or clarify) anything else. All the best, Alex Reviewer #2 ----------- > I have read with interest the paper "Electromagnetic calorimeters based on scintillating lead tungstate crystals for experiments > at Jefferson Lab", which is appropriate for publication on Nucl. Inst. Meth. A. The paper is in general clearly and well written > and, therefore, my comments are relatively minor. > General > Uniformize the choice of the past tense used in the paper, e.g. on L 164-165 in the same sentence: "is used to monitor" …. "was read-out". Would suggest, > "is read-out", and the whole paper should be consistently checked. Corrected as suggested. We also checked (and corrected) consistency of tenses over the whole paper. > Cross-check the use of "beam photon" and "photon beam" throughout the paper. Checked > Figures > Y-axis, when appropriate, replace "Events" ==> "Events/bin width" writing the actual bin width. Corrected as suggested (Fig. 5, 8, 9, and 13) > All figures showing fits to experimental data: use the same standard of Fig. 8, showing the fit parameters. Corrected as suggested > Caption fig. 14: add elasticity definition. Added to the caption: "The elasticity is defined as the energy of two clusters in the calorimeters minus the beam energy." > Fig. 12: please check the appearance of the used colours in B/W. > Same for fig. 15. We printed out the paper in B/W. Fig. 12, Fig. 15, and other plots look legible Comments to the text --------------------- > L92 as the calibration section only comes after a few pages, > maybe add: "as discussed in Section 3.5" or similar. Added: "The detector calibration will be discussed in Section 3.5." > L106 for consistency with L97, add ref. Added as suggested: "... using optical grease (EJ-550) produced by Eljen Technology [19].". > L191 "induced by the" (add the) Corrected > L217 add missing "%": 2.63% +- 0.01%, similarly for C. N in MeV? Add. We rewrote the coefficients as S = (2.63 +- 0.01)%, N = (1.07 +- 0.09)%, and C = (0.53 +- 0.01)%. All coefficients are in %. > L244 Geant refs. missing Added > Summary > Define acronyms used in the Summary e.q. L478 Compton calorimeter (CCAL). Added > References: add missing DOIs We checked all references and added the DOI to the references where the DOI exists Reviewer #3 ============ > Reviewer #3: This paper reports about the construction and performance of an electromagnetic calorimeter made > by a matrix of ~1000 crystals of PbWO4 read by PMTs. A different arrangement of the same crystals makes possible > (or will make possible) to use the calorimeter in different experiments: PRIMEX, GLUEX and NPS at Jefferson Lab. > The authors report about the measured performance of the CCAL during the PRIMEX experiment as well as some other > on-beam tests focused to check and optimize the experimental set up to face the different conditions and requirements > of the other assemblies (GLUEX and NPS). While the report is well written and easy to read, there is a little of > confusion between what has been done for the CCAL and what is related to the other two future projects. The text will > benefit by a rearrangement of the different paragraphs with a clear separation between the three projects clarifying > what has been done and what will be done. This confusion is particularly evident in the Introduction where the three > projects are mixed making the reading difficult. The authors should also describe in more details the calibration procedure > adopted for the CCAL since the final performance strongly depends on it. > After the revision, the paper should be considered for publication since the information and procedures herby described > are relevant for anyone interested in PbWO4 em calorimeters. > Beside these general considerations, below the list of detailed comments (L=Line) General: ======== 1. We agree with the reviewer and rearranged the introduction section in order to more clearly separate the projects. The structure in the new introduction version is as follows: - Design, construction, and performance of the Compton calorimeter (CCAL) in the PrimEx experiment - Upgrade of the GlueX FCAL calorimeter. CCAL as a prototype of the FCAL lead tungstate insert. - The NPS project Rearranged paragraps and addtions are higlighted with magenta color in the pdf file of the revised manuscript. 2. We expanded the calibration section and added more details there. These additions are highlighted with blue color in the pdf file of the revised manuscript. ---------------------- Below are answers to the reviewer's comments: > Beside these general considerations, below the list of detailed comments (L=Line) > Replace everywhere the word 'module' by 'crystals': it is misleading since by module usually one considers a matrix of crystals We use the word 'module' to denote a crystal wrapped with the reflective foil and coupled to the photomultiplier tube with the divider. The calorimeter is fabricated from (consists of) modules. The module design is described in Section 3.2 ("Module design"). We also added the definition of the 'module' in the introduction to avoid confusion: "The module consists of a PbWO4 crystal wrapped with the light reflective foil and coupled to the photomultiplier tube with the divider" We checked everywhere in the text and replace the word "module" to "crystal" where is refers to the bare crystal > Abstract: forward calorimeter -> Forward CALorimeter Corrected > L3-4: add (some) references to CMS, ALICE, PANDA, HPS, CLAS and CLAS12 calorimeters Added > L5: I'd also add the radiation hardness as one of the main characteristic of PbWO4 Added: " ... allows to build high-granularity radiation hard detectors with a good spatial separation ... " > L14: add (NPS) after the name Added > L22: EIC acronym precedes the name explanation in L24 The acronym was corrected > L29: FCAL acronym is not explained (other than in the Abstract) Added > The whole Introduction is a little bit confusing: I understand that crystals were used in different detectors > and arrangements but, it is not completely clear which of these many arrangements the authors are reporting > about neither if the different detectors were built or are expected to be assembled in future. We rearranged the introduction and added more details > L38: Beam photons -> Photons Replaced > L45: lepton -> electron, otherwise can be confused with positron production Corrected > L62: reconstruction of particles -> particles reconstruction Corrected as suggested > L63: " ... crystal and positioned it about 64 6 m downstream" -> crystal. > The CCAL will be positioned ... Corrected as suggested > L64: add 'Theta' to angle Added > L79: dLY/dT should reflect the inverse proportionality of LY on T The sentence was modified: "The light yield from PbWO$_4$ crystals depends on temperature and decreases at higher temperatures with a typical coefficient of $2\% / ^\circ C$ at room temperature." > L85: I'm surprise that no insulator surrounds the active volume of the calorimeter. How the T was kept uniform in the volume? > how the PMT heat (30W) was decoupled from the cold mass? The whole calorimeter box was surrounded by the insulator foam (written in the text) and is kept as a close volume. The heat exchange is performed by means of the "copper plates with built-in-pipes to circulate a cooling liquid" and "two fans with a water-based cooling system" including radiators. The fan section of the cooling loop circulates dry cold air through the box and forces cold air over the PMT electronics. We added to the text: "... two fans with a water-based cooling system including radiators are installed ... " > L160: it would be nice to have some details about the LED signal and how it compares to the actual PbWO scintillation light We added to the text: "... which allowed to generate LED pulses with a programmable rate. The width of a signal pulse induced by the LED corresponds to about $80\%$ of the pulse width produced by the PbWO$_4$ scintillating crystal. The typical amount of light injected by the LED to the crystal is equivalent to that emitted by 500 MeV photons." > L180: specify in which time frame the 6% stability has been measured Added: " ... during about 1.5 months of the experiment ... " (the experiment length is also mentioned two lines above) > L185: specify the (energy) range of the photon beam used to calibrate the detector Added: "The tagging detectors covered the energy range of the photon beam between 2.8 GeV and 11 GeV." > L192: it is not clear what Fig.7 shows: is it the amplitude of a SINGLE crystal (seed of the shower?) or the energy of the > CLUSTER (entire shower)? how crystals by crystals were equalized? > Hpw is it possible that the energy is fully contained and measured in one-crystal/cluster? 12 GeV photons are not fully absorbed > by a 20cm long PbWO calorimeter and the linearity reported in Fig. 7 does not seem to be realistic. a) Fig. 7 represents the energy deposition in a single crystal, not a shower (the beam goes in the middle of the crystal). b) The crystals were equalized by moving them one-by-one into the beam and adjusting HVs in order to set the energy response (the ADC amplitude) induced by 11 GeV photons to 3500 ADC counts c) the energy is not fully contained in one crystal, only about 80% of the shower energy is contained in a central cell (crystal). We modified the calibration section in order to better explain the procedure, see highlighted in blue color d) In regards to non-linearity in Fig. 7: - The fraction of energy (in units of %) released in a single crystal (central cell of the shower) is defined by the transverse crystal size (2 cm x 2 cm) and does not depend on the beam energy in our energy range. The longitudinal energy leakage for 12 GeV photons in a 20 cm long (~22 R.L.) crystal is negligibly small (0.2%, predicted by Geant simulation) The transverse shower leakage is defined by the Moliere radius, which does not depend on the energy. The relative non-linearity of the single-crystal response for 2.8 GeV and 12 GeV photons (due to the spacial shower leakage) is estimated by Geant simulation to be smaller than 0.4 %. Therefore the energy released in the single-crystal is proportional (linear) to the beam energy. > L197: not clear how the energy resolution is obtained. If it refers to the energy absorbed by a single > crystal exposed to the 6 GeB photon beam, it should be dominated by transverse and longitudinal leakage. > Was it accounted and corrected for? Otherwise, if multiple crystals are involved and the transverse component > of the EM shower is summed up to the seed energy, how the later crystals have been calibrated? the calibration > procedure should be described in more details. Here we describe the resolution of a single crystal (not a shower). Leakages due to the lateral and longitudinal shower spread were not corrected. As the beam can be positioned in the middle of each crystal with a good precision (better than 200 um), the relative single-crystal energy resolution can characterize the uniformity of crystals (modules) used in the calorimeter, i.e., how modules differ from each other. We added to the text how we obtained the resolution (mode details are also added to the calibration section): "We estimated the non-uniformity of the 140 CCAL modules by measuring the relative energy resolution for each individual module exposed to the beam. As the beam can be positioned in the middle of each crystal with a good precision, better than $200\;{\rm \mu m}$, the relative single-crystal energy resolution can characterize the uniformity of the crystals used in the calorimeter. We measured the energy deposited by 6 GeV photons in a single module and determined the energy resolution from a fit of the energy distribution to a Crystal Ball function[22]. The relative energy resolution, defined as the width of the energy distribution divided to the average energy deposited in the module, obtained for all 140 CCAL modules is presented in Fig.~\ref{fig:ccal_mod_uniform}. The distribution is fit to a Gaussian function. The non-uniformity of the modules, i.e., the spread of the distribution is found to be smaller than $5\%$." > L210: not clear how a 5% non uniformity in individual crystal resolution (not clear how it is defined, see > previous comment) reflects in the overall energy resolution of the CCAL (the sums up many crystals in the EM shower) We added the definition of the resolution, see response to L197 In fact, the non-uniformity of the individual crystal resolution is expected to have a relatively small impact on the overall energy resolution of the detector. Assuming that showers are reconstructed in the calorimeter consisting of 140 modules (uniformly over the calorimeter surface), and the shower-to-shower energy spread is 5%, the overall energy spread for all reconstructed shower can be naively estimated to be 5%/sqrt(140) ~ 0.5%. > L217: the resolution strongly depends on the energy thresholds (seed and sides crystals): please report here the > value. Since the threshold (per crystal, I guess) mentioned in different part of the article is different > (L185 15 MeV, L249 30MeV) it would be good report how the resolution changes by varying the threshold. No mention of the > effect of the systematic error induced by the energy calibration on energy threshold is given. A note on it would be desirable. We added to the text: "We note, that the energy resolution depends on the energy thresholds, and is getting worse with the increase of the readout threshold. This dependence is stronger at small shower energies. For the ADC readout threshold of 15 MeV per crystal used in the analysis the expected relative degradation of the energy resolution in our energy range of interest is small and constitutes to about $1.3\%$ for 2.8 GeV showers and much smaller than a percent for the shower energies around 11 GeV. In the fit to the energy resolution, we did not account for the dependence on the readout threshold. The energy resolution decreases by $8\%$ for 2.8 GeV photons if an energy threshold of 30 MeV is used." The systematic error on the energy resolution due to the calibration-related uncertainties on the energy thresholds are expected to be small in the energy range of interest used in the resolution studies (2.8 - 11 GeV) (possible uncertainties on the energy threshold due to the non-linear energy response of the module and possible gain variations could be on the level a few percent, and would not impact the resolution even at 2.8 GeV, as discussed above) > L226: what is the result of this alternative calibration? The authors should report the final parameter obtained > with the Compton calibration if different from what quoted in L217 The most precise calibration was provided by the snake scan (beam calibration), which was performed in the beginning of the experiment. The Compton calibration was used to account for small PMT gain drifts during the production run (we added more descriptions to the calibration section). > L244: I'm curious to see how simulations reproduce trigger rate as a function of the threshold and how they compare to data. > It would be good to add it to Fig. 11. We added to Fig. 11 the rate of electromagnetic interactions (EM) predicted by MC simulation. The rate of hardonic interactions (predicted by Pythia event generator) constitutes about 20% of the EM rate. As we are not confident about the reliability of the hadronic event generator (on He target), we present the expected EM rate only. > L248: I guess 200kHz is the max rate per crystal (and not per module - see my initial comment about the confusion module/crystal). > Iy is just a coincidence that this is the same value quoted in L249 but with a different threshold? We replaced module->crystal here to make it more clear. Yes, both rates were close to 200 kHz; there were two different run conditions (calibration/production) and we evaluated the rates at slightly different thresholds. > L252: the signal sampled by the fADC seems to be ~50-60ns wide: at 2 MHz, i'd expect some pile up effect. The check on the PMT > performance at that rate are not the only possible cause of a degraded resolution. It would be good to make a comment here. We used 2 MHz rate (of the periodic pulser) only for the PMT testing. We added to the text: "The rate in the detector (200 kHz per crystal) resulted in a negligibly small pile-up probability for the typical signal pulse width of 50 ns \footnote{The pile-up effects can lead to the degradation of the energy resolution and have to be considered in the FCAL insert region closest to the beamline, where the rate will approach about 1 MHz per crystal.}." > L269: did the author tried to fit the time dependence on the reconstructed shower energy? It would be nice to see the behavior. Yes, we fit the dependence. We added to the text: "The dependence of the time resolution, $\sigma_{\rm t}$, of CCAL showers on the shower energy can be parameterized by the following function: $\sigma_{\rm t} = \frac{0.32\;{\rm ns}}{\sqrt E} \oplus 0.09\;{\rm ns}$, where $E$ is the shower energy in units of GeV, the symbol $\oplus$ denotes a quadratic sum, and the parameters were obtained from a fit to the data. The time resolution is improved with the increase of the shower energy and constitutes about 330 ps and 140 ps for 1 GeV and 9 GeV showers, respectively." We also attached a plot to the reviewer's response, which demonstrates the time resolution as a function of the shower energy (see file time_res_fit.pdf ) > L272: is the deposited energy corrected by the longitudinal leak? if so, how? any comparison to simulations? The longitudinal energy leakage is negligibly small (smaller than 0.2% for 12 GeV photons) and is not corrected for in the shower reconstruction. > L277: some more details about the accuracy of the algorithm should be provided together with the cluster multiplicity > per event. We added to the text: "The algorithm was studied using Geant detector simulation. No visible bias in the reconstructed shower energy was observed after applying a non-linear energy correction. The shower coordinates were reconstructed by combining the positions of all modules constituting the shower and using a logarithmically energy-weighted sum. The coordinates of a reconstructed shower exhibit a small bias on the level of 0.1 - 0.2 mm, which depends on the position of the incoming photon on the face of the crystal. The shower position resolution depends on the cluster energy and also correlates with the coordinate of the incident photon. The resolution is smaller near the edge of the crystal and increases at the crystal center. The typical position resolution of 4 GeV photons is 1.4 mm. The algorithm provides a good separation of overlapping showers in the calorimeter by using profiles of electromagnetic showers. Two photons with energies between 1 GeV and 5 GeV positioned at a distance of 4 cm from each other can be reconstructed with a typical efficiency larger than $80\%$. The average shower multiplicity per event in the CCAL is $\sim 1.2. We consider to use this algorithm to reconstruct showers in the FCAL insert, which will be operated at significantly larger luminosity." > L279: how Compton candidates have been selected? this seems an important part that should be described in details > since the conclusion (L289) that the calorimeter is able to reconstruct Compton events is based on a clean Compton > sample identified in some other way. We added to the text: "In the reconstruction of Compton candidates we made use of the well defined kinematics of the two-body reaction. We selected events with two reconstructed showers, one in the CCAL and another one in the FCAL, that originated from the same beam interaction. The time difference between the showers and the beam photon was required to be smaller than 2 ns, which can be compared with the beam bunch period of 4 ns. We applied energy thresholds of 0.5 GeV and 1 GeV for showers reconstructed in the FCAL and CCAL, respectively. The difference in the azimuthal angle between the reconstructed photon and electron was required to be $|\Delta \phi| < 5\sigma_\phi$, where $\sigma_\phi = 5.5^\circ$ is the angular resolution. For events that passed the selection criteria we used the elasticity distribution, which is defined as the reconstructed energy in the event minus the beam energy. The elasticity distribution for Compton candidates produced by beam photons in the energy range between 6 GeV and 7 GeV is presented in Fig.~\ref{fig:compton}. ... The small remaining background under the peak of the elasticity distribution originates from the pair production reaction. For the beam energy range of interest, the $e^+e^-$ pairs are typically produced at small polar angles. The pair production contribution was estimated using the Geant simulation to be on the level of a few percent." > L292 (authors: the line number was not correct here, should be L282, not L292): is 130 MeV resolution consistent and comparable > to the resolution quoted in Eq.1? are there any MonteCarlo studies demonstrating that the observed resolution matched with > the expected one? This is a good point, we should have added this to the article. The measured resolution is in a relatively good agreement with that predicted by the MC simulation (125 MeV measured in this energy range compared with 120 predicted by the MC). We added to the text: Changed "about 130 MeV" to "is 125 MeV " and the sentence. "The measured resolution was found to be in a good agreement, at a level of a few percent, with that predicted by the Monte Carlo simulation" > L349: this is certainly true but, on the other hand, the longitudinal component of B is more difficult to shield ... > Fig. 16 Caption: you may want to say that markers represents different HCoil generated field. Changed to: "Markers denote different values of the magnetic field generated by the Helmholtz coils." > L351 'most sensitive ...' it seems that a worl such as 'part' is missed. The whole phrase should be rewritten for > better clarity. We rewrote the sentence: "We will use a 3.5 cm long acrylic light guide in order to position the PMT in the area with the smallest magnetic field. The most sensitive to the magnetic field part of the PMT is a 4.6 cm long region between the photocathode and the last dynode. The location of this region inside the PMT shield housing is shown as a box in Fig.~\ref{fig:field_distribution}." > L400: what about the attenuation induced by the 'silicon cookie'? - We didn't measure it separately, we focused on light losses in the whole module > L434: it is good to report estimates based on current test but the ultimate performance in term saturation should be > assessed by an on-beam measurement in the final configuration. I'd suggest the author to rewrite the conclusion mentioning > explicitly that his studies are only preliminary. It's a good point. We already mentioned in L432 that we would perform more beam tests "We are planning to perform more beam tests of the FCAL insert active base using the CCAL in forthcoming GlueX runs in 2021 - 2022." We also added to the summary: " ... the CCAL provided important information for finalizing the design of FCAL PbWO4 modules and understanding the performance of PMT dividers and also served to further optimize the NPS calorimeter. We anticipate to use the CCAL in forthcoming GlueX runs in 2021 - 2022 to perform final tests of the PMT dividers for the FCAL insert." > L444: I would rephrase the statement considering the factorization proof is a little bit more complicated that > what should be tested in a single experiment and certainly, I hope, 'the entire 12 GeV Jefferson Lab semi-inclusive > deep-inelastic scattering program' does not rely on this single experiment. We rephrased the sentence as: "The E12-13-007~\cite{e12-13-007} experiment will study semi-inclusive $\pi^0$ electroproduction process and seek to improve our understanding of the factorization framework, which is important for 12 GeV Jefferson Lab semi-inclusive deep-inelastic scattering program."