Fermi confirms PAMELA

The annual sabbat of the Fermi collaboration took place last week in Rome. One of the new results presented there was the measurement of the positron fraction in the cosmic rays. This is a very hot observable given PAMELA's claim that the positron fraction at high energies is larger than the one predicted by models of cosmic ray propagation. The PAMELA excess can be interpreted as a signature of dark matter, although boring astrophysical explanations are also possible. But a skeptic could doubt the PAMELA result. Measuring the positron fraction requires discriminating between positrons and much numerous protons; one needs the proton rejection power at the level of 1 in 100 000. PAMELA claims to have that rejection power, but the possibility of an unaccounted for systematic effect did exist.

Now the PAMELA result has been confirmed by a cute measurement performed by the Fermi satellite. Fermi, unlike PAMELA, does not have a magnet to distinguish positively charged particles from negatively charged ones. That's why, until now, they were presenting only the combined flux of electrons and positrons. Nevertheless, they are able to some extent separate electrons and positrons by borrowing the magnet from the Earth. The configuration of the Earth magnetic field lines happens to be such that for certain energies and certain arrival directions only electrons or only positrons are expected, see the picture. Using this effect, Fermi was able to produce the following measurement of the positron fraction:
The black data points are from PAMELA, and the grey band is the Fermi measurement with their systematical uncertainties. The two are nicely consistent. So, we still don't know whether the positron excess is due to dark matter or pulsars or old newspapers, but at least we know for sure it is real.

Another result presented in Rome deserves some advertisement. One of the main goals of Fermi is to search for gamma rays produced by annihilation of dark matter. A good strategy is to look in the direction of one of the dwarf galaxies. These are small satellites of our galaxy that are vastly dominated by dark matter, therefore the annihilation signal can be significant while the astrophysical backgrounds are less pesky. So far, no anomalous gamma ray flux from dwarf galaxies has been detected. From that, Fermi is able to put quite stringent limits on the annihilation cross section for various hypotheses about the final state into which dark matter annihilates:
The point is that for certain hypotheses, like for light dark matter annihilating into tau leptons or b-quarks, they are already excluding the cross sections expected if the dark matter is a thermal relic. So they're really closing in on the parameter space where a signal may be lurking if the WIMP paradigm is true. No luck so far, but they'll try again next year :-)

CoGeNT observes annual modulation!

What a year... Previously I had to think hard to make up a blogging subject that would not be too boring. But these days there's hardly a week without a new discovery, a new rumor of a discovery, or a refutal of the previous week's rumor. This year the particle community was already electrified by the CDF forward-backward asymmetry, the CDF W+2j bump, the would-be Higgs decaying to photons in ATLAS, and now there is CoGeNT... The rumor that CoGeNT observes the annual modulation of the signal has been circulating for a while, but only recently it was officially announced, first at the APS April Meeting in Disneyland last Monday, and today at the symposium in STSI Baltimore.

CoGeNT is a dark matter experiment located in the Soudan mine in Minesotta. In spite of a relatively small size and limited background rejection its germanium detector has certain advantages, e.g. a low threshold (0.4 keVee, corresponding to true recoil energy of about 2 keV) and a very good energy resolution. This makes it particularly sensitive to light GeV-scale dark matter whose scattering cannot produce nuclear recoils far above keV. CoGeNT was running continuously since December 2009 until March 2011 when the power was cut off due to a fire in the Soudan mine. The new results based on 145 kg.day of data continue to show an excess of events at low recoil energies which can be interpreted as the scattering of light dark matter particles in the detector. The preferred parameter region has shrunk and now points to 7-8 GeV particle with the cross section on nucleons around 10-40 cm2. More importantly they were able to measure the annual modulation of the signal. Because the velocity of the Earth with respect to the dark matter sea changes anually due to the orbital motion around the Sun, the event rate of dark matter scattering is expected to oscillate with a peak in June and a minimum in December. And here is what CoGeNT observes.
The solid line is the expectation from dark matter, and the dashed line is the best modulation fit to the signal. The phases of the two are within 1 sigma. CoGeNT estimates that the modulation hypothesis is preferred at 2.8 sigma. The modulation is most pronounced in the 0.5-2 keV region while it is absent for surface events.

Well, I don't know what to think about it. The parameter region consistent with the CoGeNT signal is naively excluded by CDMS, Xenon10 and Xenon100. One would have to assume that these 3 experiments are terribly wrong about their energy scale in order to reconcile their limits with the CoGENT signal. Maybe CoGeNT is just wrong. On the other hand, the observed modulation is very intriguing, especially in combination with the long-standing DAMA modulation signal and the oxygen band excess in CRESST. On the third hand, maybe nobody is wrong, but dark matter is simply different than what we've expected it to be. Prepare for a new wave of dark papers on arXiv.

The video of CoGeNT's presentation is here. See also this post on Cosmic Variance.

ATLAS shrugged

A new rumor propagating via twitter, social networks and other attributes of modern civilization says that the old rumor was just a rumor. The ATLAS collaboration was working hard this last week to investigate the possible signal of Higgs decaying to photons. Observations of such a signal was claimed in an internal memo written by 4 of its members. However, after analyzing about 100 pb-1 of currently available data ATLAS concluded that the suspicious excess in the γγ invariant mass distribution is going away. C'est la vie, as the local folks would say. A parallel rumor says that CMS sees nothing unusual in the γγ channel. That would pretty much end the career of the first LHC rumor. Maybe it's for the better. Otherwise the Higgs boson would have to be renamed to the Wu boson which could lead to some confusion.

Anyway, same time tomorrow, we must have a winner one day...

Update: ATLAS officially quashes the rumor. With 131 pb-1 of data there is no peak in the γγ invariant mass spectrum near 115 GeV.

Higgs in ATLAS, maybe

Peter's blog breaks a story that will be all over the news tomorrow. The Easter Bunny dropped in his comment section the abstract of a sensational ATLAS internal note, which says

...we studied the γγ invariant mass distribution over the range of 80 to 150 GeV/c2. With 37.5 pb−1 data from 2010 and 26.0 pb−1 from 2011, we observe a γγ resonance around 115 GeV with a significance of 4σ. The event rate for this resonance is about thirty times larger than the expectation from Higgs to γγ in the standard model...

So, the note claims no less than a firm evidence for a 115 GeV Higgs boson decaying into 2 photons. This decay occurs in the standard model with the amplitude dominated by a W-boson loop. However, this decay is very rare - only about 0.2 percent of Higgses decay this way. It's absolutely impossible that ATLAS is seeing the standard model Higgs - the rate is way too small. However, one can imagine theories beyond the standard model where the production cross section of the Higgs and/or its branching fraction into 2 photons is enhanced.

Well, it's past 4am, I really need to sleep. So just a few hasty remarks

• The H to γγ rate 30 times larger than the standard model one appears to be in tension with the existing Tevatron bounds. Both CDF and D0 in this mass range place a limit of order 20 times the standard model cross section. Actually, D0's limits shown at Moriond display an intriguing 2-sigma excess around 115 GeV...
• ATLAS also showed their first H to γγ results at Moriond. At that time, their limits were about 40 times the standard model rate. Actually, a bumpy feature near 115 GeV can already be discerned, especially if one assumes that the background rate is overestimated...
  
• Enhancement of the H to γγ rate by 20-30 times sounds humongous. This is certainly not a feature in any mainstream model, and I'm not aware of a single theory paper that predicts it. A recent paper argues that in the MSSM the rate is usually suppressed; in the NMSSM an enhancement is possible, however by a factor of 2 rather than 20. One ancient paper studies this issue in a more general manner. Large enhancement of the branching fraction to 2 photons is possible if the Higgs couples only to up-type quark (h0_u), or if it does not couple to fermions at all, the so-called fermiophobic Higgs (h0_bh). But this is still not enough; one would need in addition new charged particles with a large coupling to the Higgs (4th generation?, composite fermions?) in order to pump up the production rate.
• Strictly speaking, ATLAS observes a particle decaying to 2 photons, likely a scalar, and likely produced in gluon fusion. A coupling to photons or gluons is not a defining property of the Higgs boson. To demonstrate the Higgsesness of the new particle one would have to pinpoint its coupling to electroweak gauge bosons, for example by measuring its associated production with W and Z bosons. Until that is done, alternative options are on the table.

Whatever it is, it means busy days ahead, for theorists and experimentalists alike. One thing is certain: it's not a hoax, the note and the analysis really exist. But one should keep in mind that the result has not been internally reviewed yet, thus, at this point, it is not backed by the entire ATLAS collaboration. It may well turn out to be a false alarm... or it could be the discovery of the century... stay tuned.

Update: Tommaso points to newer and more stringent CDF limits than I linked to above. For a 115 GeV Higgs the γγ rate must be less than 15 times the standard model rate, which further disfavors the ATLAS signal. On the other hand, CDF has some excess near 120 GeV...

Xenon100: Nothing

The most expected experimental result of Spring 2011 is out now. XENON100 just released the results of the dark matter search based on 100 days of data-taking with xenon target in 2010. Here is what they see:
The plot shows all events that pass the quality cuts. The x-axis corresponds to the measured recoil energy determined by counting the number of scintillation photons in the event, the so-called S1. (There is an important companion paper fixing the relation between recoil and S1 at low energies where previous experimental results have been somewhat confusing). Most of the events in the plot are due to photons scattering on electrons from the xenon atoms. The way to distinguish those from the more interesting nuclear recoils (expected when a dark matter particle scatters) is by simultaneously measuring the number of ionization electrons, the so-called S2. Nuclear recoils typically lead to a smaller ratio of S2/S1 (the grey area in the plot). Therefore one makes a cut on S2/S1 (the dashed horizontal line) defining the signal region such that most electron recoils are rejected while the bulk of nuclear recoils is retained. At the end of the day one finds 3 events in the signal window (red points) while the expected background, mostly from spillover of electron recoils, is estimated to be 1.8 ± 0.6. Once again, no signal :-( Instead, we have new limits on the dark matter - nucleon cross section
For a 100 GeV dark matter particle the limit is around 10^-44 cm2, 3 times better than the previous limits from CDMS and Edelweiss. For light dark matter the improvement seems to be even better, more than an order of magnitude, which further disfavors dark matter interpretations of the CoGeNT and DAMA signals.

Actually, the paper mentions in passing that the analysis leading to these limits was not completely blind. After opening the box, there were many events at small recoil energy of which 3 fell into the signal region, which would make 6 signal events in total. However after investigating these 3 additional events the collaboration decided they were static from the electric can opener ;-), and devised additional cuts to get rid of them.

So what do these results tell us about the WIMP dark matter? At which point should we start to worry that we're on the wrong track? Unfortunately, there is no sharp prediction for the dark matter cross section. The most appealing possibility – a weak scale dark matter particle interacting with matter via Z-boson exchange - leads to the cross section of order 10^-39 cm2 which was excluded back in the 80s by the first round of dark matter experiments. There exists another natural possibility for WIMP dark matter: a particle interacting via Higgs boson exchange. This would lead to the cross section in the 10^-42-10^-46 cm2 ballpark (depending on the Higgs mass and on the coupling of dark matter to the Higgs). This generic possibility is now getting disfavored thanks to Xenon100's efforts, unless the Higgs is heavier than we expect. Therefore, even though models predicting the cross section below 10^-44 cm2 certainly do exist, it may be a good moment to start thinking more seriously about alternatives to WIMP. In the worst case dark matter may be very weakly interacting (axions, gravitinos) or very light (keV-MeV scale dark matter), in which case the current approach to direct detection is doomed from the start.

See also Peter's and Tommaso's blogs.

Another 3 sigma from CDF

Gosh, I was damn busy all day, and by now all major blogs have already run the story. Anyway, better late than never... CDF claims a 3 sigma excess in the dijet invariant mass distribution in the lepton + neutrino + 2 jets events. This analysis was originally devised to search for the diboson WW and WZ final states, where a W boson decays leptonically and the other electroweak boson decays into 2 jets. The latter should show up as a broad peak in the dijet invariant mass spectrum, on top of a much larger background from the generic W+jets production. Indeed, both D0 and CDF could see the peak below 100 GeV, which allowed them to pinpoint the diboson production at the 5 sigma level and measure the diboson production cross section. However, CDF had also a small blip around 150 GeV which was not expected.












All in all, the excess advertised in today's CDF paper is not exactly new, and it has been widely discussed among theorists. Moreover, the analysis published today has been publicly available for some time in the form a PhD thesis. What changed with respect to the earlier CDF diboson search is that the cuts have been slightly revamped to make the bump more pronounced. The excess over the standard model prediction is estimated to be slightly above 3 sigma.

It is well known that sigmas come in varieties: there or more significant 3 sigmas, less significant 3 sigmas, and astrophysical 3 sigmas. To my taste the latest CDF claim belongs to the 2nd category. We are dealing with a small hump on top of a huge background, and a small unaccounted for systematic error could easily show up as a false positive. Furthermore, D0 does not see anything; they actually have a small deficit near 150 GeV (although with much less data and different cuts). Finally, the recent experience with papers submitted simultaneously to arXiv and to popular news outlets is not quite encouraging ;-)

In spite of these caveats the effect is definitely interesting and we cannot exclude it is real. What could it be? It is not a Higgs; anything Higgsish with 150 GeV mass would prefer decaying to a pair of W bosons or b-quarks rather than to 2 light jets. The simplest explanation, proposed in this April Fools' paper, involves a 150 GeV Z' boson. A light Zprime with a significant coupling to leptons is excluded by LEP and the Tevatron, but if the coupling to leptons is small then the limits are surprisingly weak. In particular, 150 GeV Z' with electroweak size couplings to quarks is perfectly allowed, and would have the right cross section to produce a bump observed by CDF. One should note that Z' with the mass of that order could generate a large forward backward asymmetry of the top production, as observed in another CDF analysis. But one should also note that generating the asymmetry requires a large flavor violating coupling u t Z' which in principle is not related to the coupling to the light quarks that is probed by today's CDF paper. In a month or so, when I post an update, there will surely by dozens of new models explaining the bump, and some of them may link to the top asymmetry in a more direct way.

For more comments see Flip, Tommaso, Sean, and Michael, Peter, Lubos, and again Tommaso.

Update on forward-backward asymmetry

Even though the LHC is flooding us with new results, Tevatron's forward-backward asymmetry of the top-quark pair production remains the most intriguing collider result as of today. The effect hints at an exciting new physics at or below the TeV scale which may soon be uncovered by the LHC, or maybe by further analyses at the Tevatron.

To visualize what kind of departure from the standard model we need to explain the Tevatron data, see the plot borrowed from this paper The plot shows a model independent fit to the Tevatron top results. The axes are the forward and backward production cross sections of top pairs with the high invariant mass, m_ttbar > 450 GeV. One can see that the cross sections has to be modified by some 10-20 percent, and that the best fit point corresponds to the backward cross section being smaller than the standard model one. This is possible if new physics contribution to the top pair production interferes destructively with that of QCD.

To modify the cross section one can introduce one or more particles beyond the standard model that mess up into the top production. The successful models have to satisfy 2 basic requirements:

• the new particle should not be too heavy and should couple strongly enough to both the 1st generation quarks and to the top quarks. Otherwise the effect would not show up on top of the QCD top pair production.
• the new particle should couple chirally, that is to say, with a different strength to left- and right-handed quarks. Otherwise it would not generate a forward-backward asymmetry.

Models with these properties have always been around, but their number has been going through an inflationary phase during these last 3 months. One can robustly divide them into 3 classes:

1. s-channel mediators,
2. t-channel mediators,
3. new particles decaying to tops

In the text below I dropped some links to a few interesting recent examples.

In the s-channel models the asymmetry is due to a resonance which can be produced in a q-qbar collision and decays to a pair of top quarks. An example of such a resonance is a Kaluza-Klein excitation of the gluon in Randall-Sundrum-type models, or more generally an axigluon in models where the standard model SU(3) color group is extended. This class is somewhat disfavored by experiment. Since the differential top production cross section measured at the Tevatron agrees very well with the standard model, the resonance has to be either very heavy or very wide. Furthermore, since the resonance has to couple strongly to the 1st generation quarks, it is constrained by the negative results from the LHC resonance and contact interaction searches in the dijet final state.

In the t-channel models the asymmetry is generated by an exchange of an off-shell particle that has a cubic vertex with one 1st generation and one top quark. There are many possibilities for the mediator: it could be a gauge boson or a scalar; electrically charged or neutral; a color triplet, sextet, sexist, octet, and so on. Some of these models survive all experimental constraints, although there is always some tension with the measured differential top production cross section. Another problem with this class is philosophical: the new particle has to have a large flavor violating coupling to the 1st and the 3rd generation, whereas similar couplings to the 1st and 2nd or to the 2nd and 3rd generations appear to be severely constrained by experiment. If any of these models corresponds to reality then some unexpected flavor structure is being revealed.

As for the last class mentioned above, I'm aware of only one model of that kind where a new color triplet scalar decays to a top quark plus a light invisible fermion. Sounds much like stop → top + neutralino, although the required couplings do not fit SUSY. As the new particle production cannot of course interfere with the QCD top production, this possibility is somewhat disfavored by the top cross section measurements. On the other hand, such a new particle can be very well disguised, especially when its mass is close to the top mass, and is currently poorly constrained by new physics searches.

So which model is true? If we're lucky the answer may already be in the 2010 LHC data; however very few top analyses has gone public so far. Otherwise we'll have to wait till the summer conferences. If the LHC continues at this pace we may have 1 inverse femtobarn to play with by that time.