← Back

Notes · exotic hadrons

Di-J/ψ Structure Without Tetraquarks

The bumps and dips in the LHC's J/ψ-pair spectrum are usually read as new four-quark particles. Here they come out of ordinary charmonium pairs bouncing off each other after they are made.

arXiv:2610.15248 · 6 November 2026

1The puzzle

In 2020 LHCb measured the mass spectrum of J/ψ pairs produced in proton–proton collisions. Above threshold it found a broad enhancement, a dip near 6.75 GeV, a narrow peak near 6.9 GeV and a hint of more structure near 7.2 GeV. ATLAS and CMS confirmed the pattern.

CMS describes its spectrum with three interfering resonances, \(X(6600)\), \(X(6900)\) and \(X(7100)\), and reads them as excitations of a \(cc\bar c\bar c\) tetraquark. Most theoretical work also assigns the structures to states.

2The idea: rescattering

Two gluons make a compact \(cc\bar c\bar c\) lump. Before it is detected as two J/ψ, the four quarks can rearrange into other charmonium pairs, such as J/ψ ψ(2S) or \(\chi_{cJ}\chi_{cJ'}\), and back again. Each time a new pair channel opens, it pulls flux out of J/ψ J/ψ, which leaves a dip, and the interference between these paths leaves peaks between the dips.

To compare with data, the model supplies only the distortion \(E(M)\) of the J/ψ J/ψ rate. The undistorted continuum shapes come from LHCb's templates for single-parton (SPS) and double-parton (DPS) scattering:

\[ \begin{equation} \frac{dN}{dM} = \big[a\,E(M) + c\big]\,N_{\rm SPS}(M)\,e^{-\lambda (M - 6.2\,{\rm GeV})} + N_{\rm DPS}(M). \label{eq:fit} \end{equation} \]

That leaves three background parameters, \(a\), \(c\) and \(\lambda\). The positions and sizes of the features are not adjustable.

Figure 1. The model built up one step at a time against LHCb (black) and CMS (green, scaled) data. It starts with the smooth continuum, adds double-parton scattering (grey band), then lets the J/ψ pair rescatter through more and more channels. Dashed vertical lines mark the channel thresholds. Watch the dip appear at the J/ψ ψ(2S) threshold, and the 6.9 GeV peak rise with it. The χ² in the corner falls from about 1000 to 61 for 50 degrees of freedom. 23 s.
rescattering.mp4 · 0.2 MB
Stage in Figure 1χ² / 50
Single-parton continuum only1018
+ double-parton scattering109
+ rescattering, J/ψ J/ψ only156
+ J/ψ ψ(2S)117
+ \(\chi_{cJ}\chi_{cJ'}\) and \(h_c h_c\)91
+ J/ψ ψ(3S)71
+ ψ(2S) ψ(2S)65
+ J/ψ ψ(4S): all ten channels61
Table 1. Fit quality to the LHCb spectrum (6.2–7.7 GeV) at each stage of the movie, with \(a\), \(c\), \(\lambda\) held at their final values and model thresholds mapped onto the physical ones (56/50 with unshifted model masses). The J/ψ J/ψ-only stage is worse than none because, with the background fixed, the elastic bump overshoots the dip region until J/ψ ψ(2S) opens.

3What it gets right, and wrong

Fitting only the three background parameters gives χ²/ndf = 56/50 for LHCb and 117/55 for CMS (which gets its own background). The model produces:

It falls short in places: the 6.9 GeV peak is 15–20% low, CMS's dips are deeper than the model's, there is no enhancement near 7.2 GeV, and the model has a small bump at the ψ(2S)ψ(2S) threshold (7.4 GeV) that the data do not show.

The amplitude does have poles, near 6.81 and 7.10 GeV. They come from the shape of the quark-exchange force (repulsive at low momentum, attractive at high), not from bound four-quark states, so they should not be read as tetraquarks.

The di-J/ψ structures can be understood as rescattering of charmonium pairs made by a compact source, with no tetraquark resonances required.

4A prediction for Υ pairs

The same mechanism should shape the spectrum of Υ(1S) pairs. Refitting the quark model to thirteen bottomonium levels and rerunning the ten analogous channels gives the curve in Figure 2. The Υ-pair spectrum shape has not yet been measured.

Predicted Upsilon-pair mass spectrum from 18.9 to 20.3 GeV with dips at the Upsilon Upsilon(2S), (3S), (4S) and Upsilon(2S) Upsilon(2S) thresholds
Figure 2. Predicted Υ(1S)Υ(1S) spectrum (arbitrary units): no rescattering (dashed), with 2⁺⁺ rescattering (solid), and smeared by a 50 MeV resolution (dotted). Expect broad bumps near 19.3 and 19.7 GeV, a dip near 19.5 GeV, and a sharp dip near 20.05 GeV.

5Where it sits in the literature

Figure 3 maps the paper's 47 references by topic and date. A line joins two references when one cites the other (data from INSPIRE). Hover over a dot to see what it cites and what cites it. Click or tap a dot to go one layer further back: that paper's own references open in a lane below the map. The dark square at the top right shows this paper's links to all 47.

Figure 3. The references, one lane per topic, placed by date of first appearance (arXiv month where there is one; the time axis is stretched after 1990 and again after 2016). Dot area grows with total citations on INSPIRE. Of the 241 citation links among the references, only 2 cross between the quark-exchange and scattering-theory references (bottom two lanes) and the di-J/ψ literature (top six lanes): this paper joins two bodies of work that otherwise barely cite each other. Two 1948–51 scattering papers are not on INSPIRE and have no links. The second layer holds 1031 further papers; the darker ones are cited by five or more of the 47 references, the shared foundations of the field.
List of references by topic

6The paper

When hadrons collide: di-J/ψ structure in the final state

E. S. Swanson, University of Pittsburgh · arXiv:2610.15248 [hep-ph] (2026) · PDF