When increasing the collider energy does not pay off July 14, 2008

Posted by dorigo in physics, science.
Tags: , , ,

Run 1C at the Tevatron lasted only a few months between 1995 and 1996. The data which had been instrumental to discover the top quark and measure its mass had been already collected, and a long shutdown was awaiting to allow a massive enhancement of the accelerator complex and the upgrade and commissioning of the improved CDF and D0 detectors in view of Run II.

I was reminded of that small dataset -a handful of inverse picobarns, which fruited only a few papers and studies- while making a summary of past searches for heavy particles (W, Z, top quarks) in hadronically triggered datasets at hadron colliders. Hadronic signals of electroweak decays are hard to extract in hadron collisions, where they get swamped by strong interaction processes: the signal to noise ratio is usually smaller than a thousandth before any selection.

In 1987 the UA2 experiment used 4.7 inverse picobarns of proton-antiproton collisions collected at a center-of-mass energy of 630 GeV to show that W and Z bosons did indeed also decay to jet pairs (the signal is the small bump on the lower plot on the right). It was not a real surprise -nobody doubted that weak currents coupled to quarks as much as they coupled to leptons- but it was indeed a significant achievement, because it showed that the comparatively tiny signal could be extracted.

Then, we had to wait until 1996 for the next hadronically-triggered weak decay to emerge: it was the one of top quark pairs with a totally hadronic decay into six jets. My group in Padova isolated the signal after five years of studies, to which I participated actively from the start: it was my first achievement in CDF. We were later also able to measure the top mass from the same sample, and our measurement was surprisingly good, given the premises.

The next news of a hadronic signal was the observation of Z decays to b-quark pairs, the topic of my Ph.D. thesis, in 1998. I have written several times about my work on this topic, so I will not indulge in details here.

Nothing else of the same kind has emerged in CDF and D0 since then (I repeat: hadronically-triggered events yielding a signal of weakly-decaying heavy objects). The next possible signal could be that of supersymmetric particles showing up in multijet datasets with large missing energy, but I try to keep believing, against all odds, that it will not be the case.

Of the three signals discussed above, the top quark pair decay to hadronic jets was later also found in D0 with Run I data, and then in Run II by both CDF and D0. So was the Z decay to b-quark pairs (no published paper there by D0 though, while a publication by CDF is coming out in NIM very soon). Instead, the 1987 signal of W and Z decays to jet pairs found by UA2 was not replicated. Why, you might well ask.

The reason is that while UA2 collected those 4.7 inverse picobarns of data with a dedicated trigger collected by a very low transverse energy requirement, no such thing was attempted during Run 1A or Run 1B, because triggers then were tuned to collect events with leptons, which yielded much cleaner signals of W and Z decays; the hadronic decay of W and Z bosons was not considered very interesting any more, and the large fraction of the bandwidth required by hadronic triggers discouraged their collection.

We thus arrive at Run IC. In the fall of 1995, after some weeks of running, a wire of the central tracking chamber of CDF broke, and with it a whole sector of the tracking had to be turned off. Rather than having to cope with biases hard to deal with, CDF decided to stop collecting data based on the presence of tracks, and to give a much larger bandwidth to calorimeter-based triggers: in the menu, a very low-energy jet trigger was then inserted, to see if the UA2 signal of hadronic W and Z bosons could be replicated.

The W and Z signal was however not found in more than two inverse picobarns of data collected in the remainder of Run 1C (later a portion of the data was collected with the Tevatron running at the lowered energy of 630 GeV, to compare UA2 and CDF cross sections for QCD events; but the triggers had then changed again). The question again is, why couldn’t CDF replicate the UA2 result ?

CDF is the better detector, and the cross-section for W and Z production is three times larger at 1.8 TeV than it is at 630 GeV. But for once, the larger energy of the Tevatron is a disadvantage! In fact, while signals increase their rate threefold, the background from strong interactions yielding two jets with the same characteristics as those emitted by the weak decay of W and Z bosons grows by more than one order of magnitude.

(Above you can see the calculated and measured trend with collision energy of the W and Z cross sections. The data point correspond to the $S p \bar p S$ and Tevatron energies)

The different behavior of strong and weak interactions as a function of energy has to do with the inner structure of the processes yielding jets. In electroweak processes, one has to produce a heavy vector boson by quark-antiquark annihilation. That means finding in the proton and antiproton quarks carrying a fraction of about 15% (at the $S p \bar p S$) or 5% (at the Tevatron) of the parent hadron.

In contrast, jet pair production occurs for the largest part by gluon initial states. So, since other things are equal (strong versus electroweak coupling constant, branching ratios of W and Z to jet pairs), the question really is: how much larger is the probability to produce gluon interactions than $q \bar q$ annihilations at a total energy equal to 15% and at 5% of the total proton-antiproton energy ?

Posed that way, the larger energy of the Tevatron appears to have been derated to asking for smaller energy partons to produce the studied processes. It turns out that, as one decreases the energy fraction, gluons start dominating. In contrast, when a larger fraction of the parent energy is required, the partons one finds are increasingly likely to be quarks. Because of the gluon domination at lower energy, the background is much larger at the Tevatron than it is at the $Sp \bar p S$.

In conclusion, a larger center-of-mass energy is not necessarily the better choice in the search of rare processes! The features of parton distribution functions complicate the matter
significantly.

1. Andrea Giammanco - July 15, 2008

Yeah, the gluon fraction in the initial state is really a source of trouble in some cases. And it is very energy dependent.
But in spite of that, most of the interesting things that you would like to study at 14 TeV seem to be big losers when running at 10 TeV, as foreseen for 2008. The general rule seems to be that S/B is higher at 14 than at 10 TeV, *whatever you call S and whatever is your main B*.
Please let me know if you know exceptions to this rule 🙂

2. piscator - July 15, 2008

Question: does this also hold for the Higgs search via Higgstrahlung and the H -> bb di-jet resonance? This was the search mode at LEP, and as I understand it is is the dominant search channel at the TeVatron, but the prime LHC search mode is H -> gamma gamma due to the much larger QCD background. Indeed, how much fb^-1 does the LHC need to see the Higgs in a hadronic decay mode?

If the Tevatron needs O(10) fb^-1 to discover a 115GeV Higgs that decays hadronically, then I wonder what the L(HC)/T(evatron) ratio is, for the amount of luminosity that the LHC needs to discover the Higgs through hadronic decays.

This could even be a justification for extended running of the Tevatron: suppose for some reason the very rare H -> gamma gamma mode has extra suppression beyond the SM, and so is inaccessible to the LHC, while H-> bb is unaltered. Then maybe the Tevatron/LEP were the only machines that can discover the Higgs, and shutting down the Tevatron in 2009 will be like shutting down LEP in 2000….

piscator

3. dorigo - July 15, 2008

Hmmm, good question Andrea. As far as 10 vs 14 TeV go, one would have to find a electroweak-produced final state with energy of about 200-250 GeV ending up in jets only. How about the $p p \to ZH \to bbbb$ process ? I know, it is hopeless in the first place. But isn’t it more hopeless at 14 TeV ? I bet so.
Actually, let’s take a simpler one: $pp \to ZZ \to bbbb$. The cross section for the signal should increase by not more than 50-60%, while I bet that backgrounds (4-b jets) at least double.

Hi Piscator,

the detection of H->bb at LHC is hindered not only by the huge QCD background (for instance in the process $pp \to WH \to l \nu bb$, mimicked by W + jets production), but also by the fact that in proton-proton collisions a virtual W boson requires a quark-antiquark initial state, so one of the two protons has to provide a sea antiquark, with a reduced probability with respect to the Tevatron, where both initial state partons may be valence ones.

In any case, I totally agree that the detection of di-tau and di-gamma final states of direct Higgs production at the LHC are not on ground solid enough to make us sleep tight.

Cheers,
T.

4. Andrea Giammanco - July 15, 2008

Thanks a lot, these seem to be good counter-examples, although as you say they are not going to be studied even at 14 TeV.

5. Mehmet Yanki Yonel - July 21, 2008

Thx for nice article.

Sorry comments are closed for this entry