AGH UST scientists carry out first observation of light-by-light scattering in ATLAS experiment on LHC

Scientists and researchers who took part in the ATLAS experiment, here at the conference Quark Matter, where they presented the results of their work, source: photo archives of Iwona Grabowska-Bołd

Dr Mateusz Dyndał presenting the analysis of photon scattering at the conference Quark Matter, source: photo archives of Iwona Grabowska-Bołd

The ATLAS detector, source: cds.cern.ch

The computer farm of the trigger system, source: cds.cern.ch

A visualisation of photon-photon scattering, source: atlas.web.cern.ch

Within the framework of research conducted in the ATLAS experiment on CERN’s Large Hadron Collider in Geneva, the AGH UST scientists have made the first observation of light-by-light scattering. Until now, the phenomenon had never been observed directly, therefore the conducted research is of a pioneering character, and in the near future its results may open up doors to the so-called “new physics”. 

The observation of light-by-light scattering, and more precisely photon-by-photon scattering (because the term “light” is restricted to a particular range of wavelength), was performed by an international team comprising the researchers of the AGH UST Faculty of Physics and Applied Computer Science, German scientists from the national research centre DESY in Hamburg and Johannes Gutenberg University in Mainz, as well as the physicists of the Institute of Nuclear Physics of the Polish Academy of Sciences in Krakow. An important role in preparations for the experiment was also played by theoreticians, for example, from the Institute of Nuclear Physics of the Polish Academy of Sciences, who came up with the idea of performing an observation on the Large Hadron Collider (LHC). 

We had to wait 80 years for the experimental confirmation of theory 

“This phenomenon is not possible if we take into account the classical theories in physics. Classical electrodynamics does not permit the occurrence of this type of processes. In the classical theory, two photons do not interact with each other – if these photons could interact easily, we would not be able to see one another and, for example, looking at the sky at night, we would not see the stars. Thanks to the fact that we can see them, we have an indirect proof that if the phenomenon of interaction between two light beams exists, it happens extremely rarely. Only in the 1930s, when quantum theory was at its early stage, Werner Heisenberg and his doctoral student Hans Heinrich Euler suggested within the framework of quantum theory that two light beams, or two photons, as it was then when the term photon as a single quantum of light came into being, could interact with each other. It means that when they meet, they can get scattered, i.e. they change the directions of their movement – in the same way as is the case of two billiard balls hitting one another. It turns out that the phenomenon is extremely rare, and that was why we had to wait 80 years for the theory to be confirmed. In the meantime, scientists attempted to measure the phenomenon, for example, by means of indirect measurements, but until now, a direct observation of photon-photon scattering had not been performed. But at the time when our team was preparing for data collection in 2015, two independent groups of theoreticians made calculations for the Large Hadron Collider, and it turned out that with the amount of data that we could gather at the end of 2015, for the first time it would be possible to directly observe the phenomenon of photon-by-photon scattering,” says professor Iwona Grabowska-Bołd of the Department of Particle Interactions and Detection Techniques Department of Particle Interactions and Detection Techniques at the Faculty of Physics and Applied Computer Science, who took part in the measurements. 

Lead nuclei as the source of photons 

The idea came into being alongside another experiment. For most of the year, LHC is used for colliding protons, however one month is reserved for colliding heavier objects, such as the nuclei of lead. In the easiest way, a nucleus of lead can be presented as a sphere made of neutrons and protons, and thanks to the latter – carrying an extremely large electric charge. Physicists examine what happens when these two spheres collide with each other, for example, among the collision products they are looking for the famous Higgs boson. However, in the conducted experiment of photon scattering, the scientists were trying to obtain such a situation in which lead nuclei would not collide with one another, but they would pass one another in very close proximity – so closely that they would feel one another’s presence through the electromagnetic field. 

“The theoreticians postulated that in this situation the nuclei of lead would become good sources of photons. Passing one another with a speed close to the speed of light, each of them would emit photons which, by means of occasional colliding with each other, would get scattered. Therefore, we began with this concept. However, it was a challenge to measure these scattered photons. For this purpose, we decided to use the ATLAS detector. We wanted to find a simple signature of two photons which would not be accompanied by any other activity on the detector, thus apart from the observed photons, the entire detector would be “empty”. Also, in our preparations for the experiment, we had to take into account the fact that the phenomenon was extremely rare. It meant that among billions of cases recorded in the experiment, only a small number of them would be a result of the process that was of our interest. At the same time, we had at our disposal the predictions of the two groups of theoreticians. To our surprise, it turned out that they presented substantially different results. One of them was more optimistic, the other one less so. Therefore, if everything went well, we would expect between several and several dozen such cases where we could see a signal from two scattered photons in the detector,” continues professor Grabowska. 

Trigger – looking for a needle in a haystack   

The experiment had to be planned very well. In the case of research conducted by the group in 2015, when lead nuclei were collided, at the time of beams crossing one another there were up to 200 thousand collisions per second. From among such a large number of collisions only about 700 could be selected for analysis, so 996 cases out of 1,000 had to be rejected. Therefore, how to choose the most interesting cases? For this purpose, the researchers use a system called the “trigger system”. 

“As the AGH UST team, for many years we have been heavily involved in preparing the trigger system and its software for the collisions of lead nuclei. I also need to emphasise the fact that the idea of carrying out the measurement was not accidental. In 2015, I had the pleasure to coordinate work connected with the preparations of the trigger system for lead collisions. At one of the team meetings, where we were discussing the details of the settings of the trigger system and its implementation, there was a researcher who had a completely new idea. The group had been planning to search for and record the collisions of electrons, muons and other charged particles, but he came to the meeting to convince us that we had to build a trigger for two photons. We understood that it was worthwhile doing it. If he had not come to the meeting, the cases with two photons would not have been recorded for further analysis,” explains the AGH UST researcher. 

Trigger in the ATLAS experiment is a two-stage system of filtering data, whose task is to select and record for further analysis a predetermined number of collisions (for example, the ATLAS experiment in 2015 selected 700 out of 200 thousand collisions). Its first stage, specifically hardware-related, is the electronics, which, on the basis of a signal in the detectors, registers in a given area the detection of a signal that is consistent with the activity of the particle created at the point of the collision of lead nuclei. Subsequently, the given case can be passed on to the second stage, which can process 100 thousand collisions per second. The second stage is a server farm composed of about 40 thousand cores (comparable with the most powerful Polish supercomputer Prometheus of the AGH UST Academic Computer Centre CYFRONET), where filtering algorithms are performed. On the basis of the signals, these algorithms provide information that, for example, in a given area of the detector the signals patterned in a straight line, which corresponds to a trace created by an electron of a given charge. A very important feature of the particles obtained by means of collision is that when their trajectories are extended, they all meet at the point of interaction, at which the collision of two lead nuclei took place. Therefore, the trigger system recognises whether the particles come from a common point, the so-called “interaction point”, or if their directions are random. It is not possible to record all 200 thousand cases per second, as we do not physically have at our disposal so much disk capacity. Recording each case means a given number of bytes, which need to be stored on a disk in several copies, processed and analysed, which is very expensive. 

A great challenge: to design a trigger 

“When two nuclei of lead collide with one another, they produce a great number of particles, and it is the most frequent signature that we can observe in the detector, that is, in other words, most parts of the detector shine, which for us is an indicator that a collision has taken place. Therefore, our task in the experiment with photons consisted in designing a trigger that would look for these two photons in the detector, and nothing else. And I need to say that it was a great challenge! However, we were quite lucky, as prior to the beginning of data collection a researcher of Johannes Gutenberg University in Mainz approached us and said that the competitive experiment CMS were also preparing to look for a signature of photon-by-photon scattering, as it was the first time that a possibility of performing such a measurement in the collisions of lead nuclei had ever occurred. As a team working at the ATLAS experiment, we began preparations, which was not easy, as we did not have much experience in looking for cases with two photons. But we were extremely motivated, and we succeeded. Between November and December 2015, we recorded hundreds of millions of cases, in which we had planned to look for a dozen or several dozen cases that would be of our interest. Please remember that we had started off with two predictions, of which one claimed that we would find several dozen cases with two photons, the other one that we would find only a few… At that stage, we did not know why these two predictions were so different. But what is important, our trigger recorded the cases. After completing data collection, we faced another challenge. We had recorded a large amount of data, about 700 interesting cases per second, but it turned out that we had another problem, as the amount of data was so large that we were not able to reconstruct it, i.e. to transpose detector signals into particles (for example, the result of the collision were also electrons, muons, and other interesting particles, e.g. the Higgs boson). The reconstruction experts said that if we did not do something about it, the reconstruction would take 8 months, and only after this time our data would be ready for further analysis. We could not accept it, and so we began work on reconstruction optimisation. We were lucky, however, as we had recorded the interesting cases with two photons in a dedicated stream, which was so small that we could reconstruct it very quickly. The data was ready for analysis quite soon. I began an initial analysis, noticing interesting cases, including ones that were a possible proof that the experiment had observed photon scattering. This is how the analysis started. When the other scientists realised that we could see something interesting in the experiment, they joined us. In this way, a small group of scientists and researchers, among other, from DESY and Mainz, was created, and we began a joint analysis of data,” explains professor Iwona Grabowska-Bołd. 

The ATLAS detector like the Notre-Dame Cathedral 

The measurement was carried out within the framework of the ATLAS experiment, which is one of the four detectors collecting data on LHC. Alongside CMS, it is an experiment of the so-called “general purpose”, which has been designed to examine precisely the elementary particles and their interactions, described by the theory of the Standard Model. The aim of the programme of the physical experiment is also looking for signals beyond the Standard Model, which would answer the questions related to the asymmetry of matter and antimatter in the Universe, and similar ones. The ATLAS detector is 45 metres in length, 25 metres in height, and it weighs 7,000 tons. Its size is comparable to a half of Notre-Dame de Paris, and it weighs as much as the Eiffel Tower. It is a detector with millions of read-out channels, and millions of other parts and elements that are used to record particles. 

Measuring photon-by-photon scattering 

“It needs to be emphasised that ATLAS is a very good experiment to measure photons. It has been designed and optimised to search for the Higgs particle. The Higgs boson has a property of a short lifespan, and it breaks up into pieces shortly after it has come into being, inter alia, into two photons. And two photons is exactly the type of signal that we were looking for in our experiment. The difficulty was that our two scattered photons had much lower energies and they were more difficult to measure, as the experiment had been optimised to search for photons of high energy, being a result of the Higgs decay. Because of that, we had to calibrate the detector to measure lower energies. Photon-by-photon scattering is one of the predictions of quantum electrodynamics. But we had to wait 80 years for the theoretical prediction to be observed in the collisions of lead nuclei, which are good sources of photons. Exactly as in the case of the Higgs boson, although less time was needed here, as the boson was observed in 2012, and the theoretical predictions were made in the 1950s. Eventually, we observed 13 cases in which we found two scattered photons. Physicists announce a discovery when they have obtained a result of at least “5 sigma” (5 standard deviations above probable background fluctuation), for which the probability of an error is only 1 in 3.5 million. Due to a limited number of cases, we estimated our discovery at “4.4 sigma”. Because of that, we would like to repeat the experiment in the future in order to see even more pairs of scattered photons and to reconfirm our discovery. Taking into account the CERN’s timetable, the subsequent experiments are planned for November and December 2018. Moreover, this phenomenon – photon-by-photon scattering – from the theoretical point of view opens up interesting possibilities, for example, searching for the signals of the so-called “new physics,” says the AGH UST researcher. 

The members of the team that conducted the experiment are the employees of the AGH UST Faculty of Physics and Applied Computer Science: professor Iwona Grabowska-Bołd and Marcin Guzik, DSc, as well as Mateusz Dyndał, DSc, an AGH UST graduate, currently working at DESY, plus three researchers of Johannes Gutenberg University in Mainz. 

The results of the experiment were presented at two conferences: in September 2016 (as a preliminary result, without publication), and the final results were presented at the beginning of February 2017, at the most important conference in the field of the physics of heavy-ion collisions “Quark Matter” in Chicago. They were received with great enthusiasm by the scientific environment.  

More information in the articles: