Relativistic Heavy Ion Group


Lisbon Jet Workshop 2014

Doga presents a study of dijet events in pPb collisions, making good use of the analog pointer.

The third Heavy Ion Jet Workshop was held recently in Lisbon, sponsored by MIT and Instituto Superior Tecnico. Over three days, more than twenty five theoretical and experimental physicists gathered and discussed recent results in jet physics and the modifications observed when jets propagate through the strongly coupled medium created in the high energy collisions of nuclear matter, or Quark Gluon Plasma (QGP). Of particular interest at this workshop were new experimental results recently presented at the Quark Matter conference in Darmstadt, new background subtraction techniques for a many particle environment, and the introduction of several Monte Carlo generators modeling heavy ion collisions.

Jets are of particular interest for study of the QGP, as in dijet and gamma-jet events they make excellent probes of medium properties. Jets correspond to high momentum partons scattered perpendicular to the beam pipeline, producing a highly collimated spray of particles which can be reconstructed with our detector and then clustered into the final object. As opposed to the vacuum case studied by our colleagues in particle physics, a loss of jet energy in the QGP has been observed. This energy loss is dependent on the size of the produced medium, as an increasing number of nuclear participants in the collision corresponds to greater energy loss. New experimental results and Monte Carlo generators seek to characterize this effect, and thus develop our understanding of the produced medium.

Chris presents recent results in PbPb collisions. Much like background, a blurred photo can obscure the signal.

Doga Gulhan, Yen-Jie Lee, and Chris McGinn presented recent results on behalf of the MIT Heavy Ions Group, showing respectively results from dijet events in pPb collisions, an exploration of nPDF sensitive observables at the CMS detector, and a missing momentum study of dijet events in PbPb collisions. Additionally, Doga Gulhan presented a generator level study of dijet system eta distributions in pPb events, showing that the effect can be reproduced under the assumption that the proton is largely driving the shift. These experimental results in heavy ion jet physics can constrain the space of current explanations for observed jet quenching, and perhaps even inspire new ideas for further measurement.

An illustration of the new powerful HF/Voronoi background subtraction. After subtraction, momentum in eta-phi space is largely consistent with zero.

To study jets and the impact of the QGP, a rigorous background subtraction method specific to the high particle multiplicity environment of heavy ion collisions is needed. The large numbers of particles produced in collisions of nuclear matter can obscure a signal from a jet, or enhance said signal if found on an upward fluctuation of the background. Thus, a jet must be corrected to more accurately reflect the true momentum and energy. To this end MIT postdoc Yue-Shi Lai presented the new HF/Voronoi subtraction technique. The panel here from his slides shows a map of momentum in eta-phi space before and after applying HF/Voronoi background subtraction.

The workshop concluded with some discussion of standardizing Monte Carlo output such that these new generators can be more readily incorporated into existing software frameworks. With standardization in place, cross comparisons between different generators and data will hopefully teach us more about the properties of jets in the QGP.

Quark Matter 2014

Here we see relative to proton-proton systems, dijet imbalance is recoved in central PbPb systems by low momentum particles distributed through large angles.

The Quark Matter conference returned to its birthplace, Darmstadt, in its 24th edition. Results from proton-Lead collisions at the LHC continued to generate interest, with contributions from MIT and CMS at the forefront. The most intriguing results from the previous Quark Matter conference concerned energy loss in the hot dense medium formed in heavy ion collisions. This conference continued in that vein, and MIT presented exciting new results exploring the origin of this energy loss in proton-Lead and Lead-Lead collisions.

In PbPb collisions, Doga Gulhan presented an extensive study of parton energy loss traversing the quark gluon plasma in dijet events as part of a parallel talk. In dijet events of high momentum imbalance, the missing momentum is recovered by summing over tracks scattered through large angles in the medium. By projecting the momentum of tracks onto an axis symmetric with respect to the dijet system, we are able to study the angular distribution of particles in both the leading and subleading jet hemispheres. The accompanying slide of particle angular distribution is taken directly from Doga’s presentation.

Doga giving a talk on the interesting dijet imbalance result in PbPb collisions.

Yet another way to investigate this energy loss is to use heavy-flavored particles. By looking at the spectra of B-mesons, we can try to understand whether or not the energy lost by a particle in the medium depends on the flavor of the quarks which constitute it. Gian Michele’s poster, discussing B-meson reconstruction and spectra in proton-Lead collisions was one of 5 posters selected (from nearly 400 entries) for a five minute flash talk on the final day of the conference.

While dijet and heavy-flavor systems can offer insight into the state of the probe after interaction with the quark gluon plasma, it is difficult to disentangle what the state of the probe was before traversing the medium. One solution is to use a much rarer event topology using photons. These photon-jet events allow an unbiased characterization of the energy both before and after interaction with the medium. Alex Barbieri presented updated results of our previous photon-jet paper confirming the earlier result (significant quenching in PbPb) with higher statistics pp reference data. In addition, no modification of the probe was observed in the new pPb data.

Alex presenting dijet and photon-jet results in pPb and PbPb collisions.

Using the new pPb data collected in 2013, a very sensitive measurement of the change of parton distribution functions (PDFs) due to nuclear effects was conducted using the distribution of the average eta of dijet systems as the ruler. Alex presented this result as part of his parallel talk, and there was significant interest from the theoretical community for this result, as it is one of the only measurements sensitive enough to differentiate between multiple competing models of the nuclear PDFs.

Another highlight of the conference was the ATLAS confirmation of a baffling result first presented by CMS at the Hard Probes 2013 conference which shows that there is an excess of high transverse momentum charged particles in proton-Lead collisions as compared to proton-proton collisions. We hope to explore in further detail these, and other interesting phenomena in heavy ion collisions in the next few months!

June 4, 2014: Fragmentation function measurement in PbPb and pp collisions

Figure 1

Figure 1 in the paper: Fragmentation functions (top) and PbPb-to-pp ratio (bottom) vs. ξ for 100-300 GeV/c jets and charged track pT > 1 GeV/c.

Our measurement of the fragmentation function in PbPb and pp collsions is submitted to arXiv:1406.0932 and for publication in Physical Review C. This paper is the second PbPb/pp fragmentation function measurement from CMS, and the first paper to observe a significant modification in PbPb collisions when compared to pp. The evolution of this difference is further explored as function of both the the collision centrality, and for the first time, jet momentum.

January 20, 2014: Dijet measurement in pPb collisions

EPS09 comparison

A comparison between dijet eta distributions in data and in NLO calculations are shown.

Our paper on first event activity dependent jet measurements in pPb collisions is submitted to arXiv:1401.4433. The dijet transverse momentum ratio is found to be independent of event activity in pPb collisions. This confirms that the dijet transverse momentum imbalance enhancement in central PbPb collisions is not originating from initial-state effects. It also shows that any final state effect on jets in pPb collisions is small, justifying the usefulness of jets in studying initial state effects. Initial state effects due to nuclear parton distribution functions(nPDFs) are quantified by the measurement of dijet pseudorapidity in minimum bias events. The measurement is agreement with EPS09 nPDF, while it disfavors the calculation with CT10 proton PDF. The observation of a large modification in dijet pseudorapidity and the shift in its mean towards the lead beam direction as a function of event activity sets a new puzzle for the field. Some clues to understand the shift are included in the paper by the measurement of dijet pseudorapidity as a function of activity in the lead direction for fixed the activity in the proton direction.

July 1-4, 2013: Paris Jet Workshop

Proton-lead collision

An event display of a heavy ion collision containing two reconstructed jets with unequal energies.

At the start of July, MIT and Universite Pierre and Marie Curie jointly organized a workshop on Jets which took place in Paris. About twenty high energy and heavy ion theoretical and experimental physicists were in attendance discussing the challenges and new opportunities of doing jet measurements and predictions in heavy ion collisions. A jet is the experimental signature of a quark or gluon that is scattered off at very high momentum perpendicular to the direction of motion of the colliding particles, the beampipe, and into our detector. However a single quark or gluon can not freely propagate through the vacuum to reach our detector but will transform into more quarks and gluons sharing the original momentum, in a process called fragmentation. It will also pull particles from the vacuum to create a collimated spray of color neutral particles that exist for timescales observable by our detectors, in a process called hadronization. This spray of particles is what we call a jet and it carries exactly the energy and momentum of the initial parton, so by reconstructing the jet in our detector we can infer these kinematic properties of the partons that we can't directly observe.

Jets are also the link between the theory and experiment of high energy quarks and gluons. One challenge is for experimentalists to make sure their measurements of jets are calculable theoretically, as well as theorists making jet predictions that are measurable. For this we have jet algorithms that both theorists and experimentalists agree upon and use, the anti-kt algorithm is most commonly used by our group to reconstruct jets. At this workshop we discussed how to expand the ways in which theorists and experimentalists can compare data and theory in more effective ways.

Physicists clustered by the anti-kt algorithm.

Physicists clustered by the anti-kt algorithm.

In the figure on the left are displayed two jets reconstructed with this algorithm in a collision recorded by the CMS detector. One of the jets here has almost three times the energy of the of the second one but is completely back-to-back , pi units apart in azimuth. This means the jets coming from a pair of particles scattering off one another conserve momentum in direction but not in magnitude, which implies the quark gluon plasma seems to slow down high energy quarks and gluons but not significantly change their direction.

However alongside the spray of particles originating from the jet, in heavy ion collisions there is also a large spray of particles everywhere in the detector from the expanding quark gluon plasma. This makes reconstructing jets to accurately reflect the properties of the original parton much more difficult in heavy ion collisions since it becomes ambigous which particles around the jet come from the quark gluon plasma and which come from the parton. At this workshop we discussed various techniques and new ideas to subtract these background effects and understand the nature of jet reconstruction in a dense environment.

May 13, 2013: Frank Ma defends thesis

Congratulations to Frank on his successful thesis defense on "Detailed Characterization of Jets in Heavy Ion Collisions Using Jet Fragmentation Functions". In the fall Frank will continue his studies in Boston at the Gordon Conwell Theological Seminary.

March 28, 2013: Yongsun Kim defends thesis

Congratulations to Yongsun for succesfully defending his thesis on "Study of jet quenching using gamma-jet events in Heavy Ion Collisions at 2.76TeV". Yongsun will soon go back to Korea where he will continue to work on CMS as a postdoc at Korea University.

February 28, 2013: Yetkin Yilmaz defends thesis

Congratulations to Yetkin for succesfully defending his thesis on "Jet quenching in heavy-ion collisions at LHC with CMS detector". Yetkin will now be continuing to work on the CMS experiment as a postdoc in the heavy ion group at LLR.

Spring 2013: Colliding different particle species: the LHC run 1 ends with proton-lead collisions

Dijet in pPb collision

A dijet event in a proton-lead collision as seen by the CMS experiment.

The new year brought a new type of collision at the LHC: the accelerator smashed protons and lead nuclei together. Although we already caught a glimpse of these asymmetric proton-lead (pPb) collisions during a pilot run last September, the data collected early this year was the first sustained pPb run at LHC energies.

The high luminosity pp data taking aimed at precision measurements addressing the properties of the recently discovered higgs boson ended in mid december 2012 with a total integrated luminosity of 23fb-1. After three weeks of downtime the recommissioning of the accelerator and the CMS detector for pPb data taking started on January 4th during the annual CERN Christmas break. Commissioning the detector involves making sure all of the subdetectors of the CMS experiment are operating within their expected parameters. On 7 January, when everyone returned to CERN, cooling and power were up and running. Since the collision rate in the pPb run was lower than that in the pp one, two forward subdetectors that couldn't withstand the radiation under the harsher conditions of proton collisions have been reinstalled in CMS: CASTOR and ZDC. These subdetectors will help measure the collision remnants that travel very close to the beampipe, crucial to the study of pPb data.

The biggest challenge in preparing the CMS experiment for pPb collisions was to configure the trigger system that selects the collisions to be recorded to mass storage. Compared to proton proton or lead-lead collisions we are looking at different physics processes happening at different beam intensities. This requires dedicated trigger menus to be developed by the CMS Heavy Ion group, which tell the data aquisition system how many events of which type (e.g events containing high pT jets or pairs of muons) to keep and which events to reject. When colliding lead nuclei, there are around 4000 interactions each second, of which around 200 are selected to be recorded to tape. The proton-lead collisions, on the other hand, happened at around 2,000,000 interactions each second, and CMS recorded around 1000 of these. This of course requires a carefull selection of events to ensure all future physics analysis will have a sufficient number of events to work with, while making sure not to exceed the available output bandwidth.

For this run, CMS also joined forces with the TOTEM experiment to cover a greater range of collision data. The two are essentially separate entities -- independent experiments that use different analysis software -- and they are fully complementary. CMS measures in the central region and TOTEM exclusively measures in the very forward region. Combining information from both allows us to perform a lot of physics studies that previously were impossible to do by correlating proton remnants seen in TOTEM with objects such as jets and Upsilon particles observed in the central part of CMS.

Proton-lead collision

A proton-lead collision as seen by CMS from a side view, here the protons come from the right and lead nuclei come from the left.

During the 2013 Ion beam period the LHC delivered 31.3 nb-1 of pPb collisions to CMS and also provided a short run of pp collisions at the center of mass energy of the PbPb collisions collected in 2011. The pPb data sample corresponds to about 60 billion collisions sampled by the CMS experiment and will now serve as a reference to the PbPb collision data.

In PbPb collisions a hot and dense system of strongly interacting matter is expected to be produced, which in turn we study by means of hard interactions produced inside the hot medium. Fast partons produced in these hard interactions traverse the hot medium and are modified by it, e.g. they lose energy due to the interaction with the medium. To perform precision studies to quantify these medium modifications it is essential to disentangle the effects on the production of these probes due to the initial state, i.e the parton distribution function of a highly Lorentz contracted lead nucleus, and the final state effect due to the presence of the medium. At this stage the pPb data comes into play, since colliding a proton with a lead ion will expose all initial state effects while the system size of the interaction in expected to be too small to form an extened volume of a strongly interacting medium. Apart from serving as a reference sample to PbPb collisions the analysis of pPb data is already starting to yield results that also make the study of these collisions a very interesting topic in its own right, like the recent analysis of two particle correlations that shows an unexpectedly large ridge like structure.

Going into the first long shutdown period of the LHC lasting until end of 2014 the CMS Heavy Ion group is now well equipped with a large PbPb collision data set and matching statistics pPb and pp reference samples. The quiet time during the LHC shutdown is eagerly awaited by all group members to dig into the wealth of data, produce exciting new physis results and further our understanding of nuclear matter at extreme conditions.

[Note about all event displays: The red and blue boxes show energy deposits from particles produced in the collisions in the electromagnetic and hadronic calorimeters, respectively. As the inner tracker was not activated during this run, charged particles could not be reconstructed as charged tracks.]

Fall 2012: Discovery in the proton lead test run

Ridge in high multiplicity pPb collisions.

The biggest physics surprise of the 2012 pPb test run.

Shortly after the August Quark Matter conference, the Large Hadron Collider geared up for a test run of proton on lead (pPb) collisions. Proton nucleus collisions are the intermediate stage between the short length scale proton proton physics and finite sized physics of lead lead collisions. These types of collisions are going to be the main focus of the January running period and this short test was meant to get us ready for what's to come. One of the test fills, a few hour continuous running period from midnight to 6am on September 13, gave us about two million good proton lead collisions.

It was very exciting as this is the first look at these collisions at such a high energy, 5TeV center of mass, producing never-before-seen particle densities in proton nucleus collisions. We were curious not only to see what reference measurements we can obtain from these collisions, but also if there is any new undiscovered physics lurking around. Indeed last time we saw unprecedented particle densities in proton proton collisions when the LHC turned on two years ago our group discovered a surprise in the form of a "ridge" in two particle correlations. This "ridge" indicates that something is driving particles produced from these collisions which have very little time to talk to each other, i.e. they are causally disconnected shortly after the collision, to have some slight preference in which plane they are produced. To see this effect in proton proton we had to look at extreme rare collisions in which high enough particle densities were produced, picking out the top 300K of over 150 billion events.

So while we expected to see some kind of ridge since it's there both in pp and PbPb collisions we didn't quite expect to see a strong signal in just the two million events we recorded. However when we looked at pPb collisions with similar particle densities as those which showed a ridge in pp collisions, we saw a huge ridge correlation! The effect is the same as described above for the pp ridge. We have something driving produced particles to align to some sort of plane, except in a manner much stronger than observed in the high multiplicity pp collisions, so strong it resembles PbPb collisions more than pp. This generated quite a bit of excitement, the unveiling of the result at the Hot Quarks conference was met with applause. CMS posted the following article on this result, which was the top headline in the physicsworld magazine a few days ago.

This exciting news may be just the tip of the iceberg of what's to come from the full proton lead run in January. The entire statistics from the test run that yielded this result, two million good collisions, is just one second of running time in 2013!

Quark Matter 2012

Gunther at Quark Matter

Gunther presenting an overview of all the results from CMS heavy ions on the opening day of Quark Matter. See more photos here.

The 2011 PbPb run was the run for rare physics and pushing the boundaries of what can be measured and high precision. With an arsenal of dedicated triggers we showed a whole range of interesting new results at this years Quark Matter conference.

By looking at the top 0.01% of collisions which deposited the highest amount of energy in our detector we selected almost fully overlapping nuclear geometries. Measuring the fourier spectra of these collisions with the tight geometric control we provide some of the best constraints to the viscosity of the hot quark gluon plasma produced at the LHC.

By selecting jets and high momentum tracks right as the collisions occur in the machine, we have aquired and analyzed a dataset that paints a comprehensive picture of jet quenching and parton energy loss from many angles. We showed the azimuthal anisotropy of charged particles of momentum above 20 GeV/c up to 60 GeV/c which probes the path length dependence of energy loss in the quark gluon plasma. We showed a complete picture of particle suppresion in PbPb compared to proton proton collisions for both charged particles and jets, as well as bottom quarks through j/psi's and tagged jets, which all show significant suppression. However bosons which don't interact strongly with the medium have been measured to have no significant suppression compared to proton proton, in the isolated photon, and W and Z channels.

We have shown the direct energy loss of colored partons traversing the medium by precisely measuring the initial energy of the outgoing photon in gamma-jet events. In addition we have studied the detailed structure the constituents of jets both in momentum and angular redistributions compared to similar jets found in proton proton collisions. These measurements further probe in what ways the quark gluon plasma responds to high energy particles.

May 5, 2012

Our measurement of medium induced energy loss in QGP using photon-jet events was finally submitted to arXiv:1205.0206 and for publication in Physics Letters B (PLB). This paper has been long awaited since the conception of the heavy ion program at CMS more than 12 years ago (See early CMS Note, p122). It is the first time in the heavy ion field where the medium induced energy loss can be directly studied with a fully unbiased probe (photon).

PhotonJet energy loss vs npart

Figure 3 in the paper. The measured distribution of fraction (x) of remaining jet energy after losing energy to the QGP, plotted in 4 bins of increasing amount of medium interaction.

Also today our dihadron correlations paper got accepted into EPJC!

April 2012

EPJC front cover

Our paper featured on the cover of European Physical Journal C.

Our paper on the measurement of high transverse momentum charged particle suppression, accepted by the European Physical Journal C (EPJC) in February 2012, was recently published and featured in the cover of the new EPJC issue (March 2012). The figure that was featured in the cover is the main result of the paper showing the suppression of the charged particles over a large transverse momentum range, where a number of theoretical predictions show large variation, in particular at high transverse momentum regime, exhibiting large theoretical uncertainty. Together with other jet quenching related quantities that we have measured, this measurement should help elucidate the mechanism of jet quenching and the properties of the medium produced in heavy-ion collisions at collider energies.

March 2012

We have three new papers about to be published that mark some of the most important Heavy Ion measurements to date, addressing how colored particles lose energy as they traverse the quark gluon plasma created in our relativistic heavy ion collisions. The first measurement we made probes how high energy quarks and gluons fragment as they pass through the hot colored medium and into the vacuum. The second clarifies how these high energy particles lose energy depending of how much medium they pass through. The final paper uses back-to-back pair of photon and quark or gluon to directly probe the absolute amount of energy loss.

February 2012

2011 / 2010 Data Taking

We recorded 7 inverse microbarns in all of 2010, which was surpassed in 1 day of 2011 data taking.

For the month of November the LHC switched to colliding 2.76 TeV lead nuclei. The heavy ion group was heavily involved in the data taking which resulted in a very rich, high statistics dataset. This year's recorded luminosity is 20 times higher than last year's data giving us access to more precise measurements of known phenomena as well as new measurements and probes available for the first time ever in heavy ion physics.

In a little over two months after the end of data taking we published an updated dijet imbalance paper taking advantage of the full 2011 high statistics dataset. In addition to analyzing new 2011 data we finalized the measurements first shown at Quark Matter and published those in three new papers. The first paper of 2012 is the measurement of isolated photon production in pp and PbPb collisions at 2.76 TeV , where the transverse energy distributions are found to be in good agreement with next-to-leading-order perturbative QCD predictions and the ratio of PbPb to pp isolated photon ET-differential yields is consistent with unity for all PbPb centralities.

Next came the study of dihardon correlations and azimuthal anisotropy harmonics where we show the evolution of short (jet) and long range range correlations with increasing centrality and transverse momentum, as well as the single particle azimuthal harmonics up to fifth order from a Fourier analysis.

The most recent completed paper is the measurement of high transverse momentum charged particle suppression. We find the charged particle yield is suppresed by a factor of 5 compared to pp collisions in the transverse momentum range of 5-10 GeV/c and rises to a factor of 2 in the 40-100 GeV/c range.

Quark Matter 2011

Yenjie at Quark Matter

Yen-Jie giving a talk in front of the full quark matter audience shortly after completing his PhD.

After barely a year of pp and a month of PbPb collisions, several exciting new results emerged from analyses led by our group. From the first 7 TeV pp collisions two summers ago novel correlations were seen in collisions producing the highest number of particles. These were never before seen in pp collisions nor were they predicted by any of the existing Monte Carlo models. This was the first manifestation of unexpected physics at the LHC which has gathered significant interest in both the media and the pp and heavy ion community.

From the start of the PbPb run we saw striking evidence in some of the first event displays of a phenomenon called jet quenching, where a colored quark or gluon passing through the QGP loses a significant fraction of its energy by interacting with the medium. While jet quenching has been observed before, the LHC was the first time you could really see it "with the naked eye". A closer investigation of these types of collisions revealed that we may need to rethink how colored partons really interact with the QGP, since it turns out a quenched jet looks no different than a lower energy jet produced in a vacuum.

These are just a small fraction of the new and exciting discoveries and measurements being performed by the MIT group at the LHC. In less than two months a new PbPb run is coming up with an order of magnitude more collisions than last year, giving access to more precise measurements, rarer events, and potentially new discoveries.