It has been an intense few weeks: After the submission of ILC detector letters of intent at the end of March, see for example the ILC newsline article for more details, the committee evaluating those documents came back with some questions at the end of April. The one that concerned me was the question of how you calibrate such a finely segmented calorimeter like we are planning for an ILC detector. That is certainly a reasonable question to ask, considering for example the ATLAS hadronic barrel calorimeter has 10 000 electronic channels, while the barrel hadron calorimeter for the ILD detector, the detector concept I am working on, will have about 4 million channels, so 400 times as many channels as ATLAS! In order for the detector to work, the signal that each of these channels puts out when a certain amount of energy gets deposited in it has to be known, since this will be different for each of these cells, just due to variations in the components in the detector. So with the calibration procedure, we’ll have to get a scale factor for each of the 4 million cells that tells us how it responds.
A report that outlines our calibration plan is due today, so over the last few weeks a calibration task force of the CALICE Analog HCAL group, with people from DESY, Heidelberg and my group here in Munich, has looked at all sorts of aspects of this problem. Here, the wealth of test beam data we have is both a blessing and a curse: Our concept of the hadronic calorimeter is the only one that has been tested in particle beams so far, so we can actually test all of our ideas on real data, not only on simulations. That is a massive advantage, but also a whole lot of additional work.

Particle tracks identified within a hadronic shower in the CALICE hadron calorimeter: The right stuff for calibration!
We looked at detailed simulations to determine how good the calibration actually has to be, and it turns out that you don’t have to know each individual cell very precisely, just because so many of them contribute and the statistical averages save you in the end. That is very important, since you need a reasonable amount of information, for example from real events or from cosmic particles, for each cell that you want to calibrate. And the more precise you want to be, the more data you need. And at 4 million cells total, each single one is not getting a whole lot. Luckily, we now know that already from the construction and from bench tests we can get the precision we need for individual channels. But, as expected, the overall scale in larger units, such as the layers in the detector, requires a much higher precision, something we can only do once everything is put together. But also for this we have a plan. The high granularity of our detector here actually helps: We can identify individual particle tracks, so called minimum-ionizing particles, within the hadronic showers. The nice thing about those: We know how much energy they loose in our detector, so they are exactly what is needed for calibration. And, as you can see from the picture on the right, we already know from our beam data that this works indeed!
OK, we know how precise we need to be, and we know how we are going to do it, but we went even one step further: Try the whole thing on real data! So, we took a calibration for our calorimeter that was recorded at Fermilab, and applied it to data that was taken at CERN a year earlier. In between the calibration run and the data run, the detector was completely disassembled, electronics were exchanged, things got broken and got fixed again, and everything was moved by truck, rail and ship from Geneva to Hamburg and on to Chicago. So, things are not getting harder than this. If we can use the calibration from one continent, and apply it to data taken on another continent for our test detector, then we’ll definitely be able to do it for the final detector modules, which get specifically built to allow this.
And indeed, it worked! At first, there was still a systematic offset in the energy between the different calibrations, but using the tracks in hadronic showers, we were able to adjust the calibration of our detector layer by layer, getting back the correct response. This is the result:

Energy reconstruction in the CALICE hadron calorimeter with three different calibrations. Left: Calibrations from CERN and Fermilab for CERN data, Right: Calibrations from CERN and Fermilab for CERN data, layer by layer calibration with hadronic tracks.
The figures show the deviation of the reconstructed energy to the known beam energy. Perfect agreement is not expected, since only the data in the hadron calorimeter are considered here. The left plot shows data taken at CERN, reconstructed with three different calibrations. The black points are reconstructed with the calibration taken at CERN, the other two points are two different versions of the calibration derived from Fermilab data. The deviations are clearly apparent, with about 7% difference in the total reconstructed energy. After a recalibration of each detector layer using our track method, the agreement is much improved, demonstrating that the calibration scheme planned for a future ILC hadron calorimeter indeed works.
The document describing all this is now being integrated into the updated ILD letter of intent, and I’m looking forward to a weekend with a lot of sun and not too much work for a change.