The proton mass is 938 MeV. People often claim that

(A) The proton is a bound state of two up quarks and one down quark, with the three quarks contributing a total rest mass of $2 \times (2.2 \text{ MeV}) + (4.7 \text{ MeV}) = 9.1 \text{ MeV}$ (about 1% of the proton mass), and the other 99% coming from "QCD binding energy".

People further claim that

(B) The "QCD binding energy"'s contribution to the proton mass is independent of the Higgs field. Therefore, only 1% of the proton mass comes from the Higgs field.

Claim (A) is perhaps defensible as a heuristic explanation, although it sweeps a lot of subtleties under the rug. (For example, in a non-QCD context, binding energy (as usually defined) always decreases a bound state's energy below that of its constituent parts - the opposite of what happens in hadrons.) But as I understand it, claim (B) is simply incorrect.

As explained here, the proton ground state is best thought of not as a bound state of exactly three quarks, but instead as a superposition of many different huge collections of quarks and antiquarks, each of which has total up quark number 2 (i.e. two more up quarks than up antiquarks) and total down quark number 1 (i.e. one more down quark than down antiquark).

This is sometimes described as "two 'valence' up quarks and one 'valence' down quark, surrounded by a huge 'sea' of quark-antiquark pairs". In my opinion, this description is also misleading, because it implies that there are three specific "valence" quarks that are "real", while the other "sea" quarks are just "virtual" and physically distinct from the valence quarks. In fact, every individual quark is physically on the exact same footing. In particular, I believe that all of the quarks gain a mass contribution from the Higgs mechanism.

As Prof. Strassler explains in the link above, the proton mass is best throught of as arising from the sum of three contributing terms:

  1. The sum of the rest energies of all of the (many!) quarks and antiquarks
  2. The kinetic energy of the quarks, antiquarks, and gluons
  3. The binding energy stored in the gluon fields, which is negative and actually decreases the proton mass relative to what it would be if the quarks did not interact.

I believe that the Higgs mechanism is responsible for the contribution (1), which is much larger than the naive 9.1 MeV from just three quarks, and therefore much more than 1% of the proton mass. Is this correct?

Also, do we have any kind of quantitative estimate of the contribution of each of these three terms? The number of up quarks and the number of up antiquarks are individually indefinite, but can we numerically estimate their ground-state expectation values in order to estimate (1)? I know that QCD is strongly enough interacting that it's not particularly useful to think in terms of individual quarks, but I'm curious whether anyone's attempted that calculation.


3 Answers 3


Yes, the standard model of a hadron as "three bound quarks" and therefore computing the contribution of the Higgs mechanism to the hadron mass solely from these three bound quarks is overly simplistic.

The name for the model of the proton that Strassler describes in the post you link is effectively the parton model, which was initially proposed without any knowledge of what the partons were, while we now identify them as quarks and gluons. The parton distribution functions describe the probability to detect a particular parton with a particular momentum at a particular overall energy scale.

And there's the first hindrance to resolve the question of "how much" the Higgs mechanism contributes to the mass: The parton distribution is not fixed, and depends on the overall energy scale (the energy scale is a measure for how "finely" we can "resolve" the substructure of the proton). There is no claim of a "true" parton distribution, it's all in flux and depends on the resolution power of the observer.

Secondly, we don't really know how to compute parton distribution functions, at least not when I last checked. We have to fit them to collider data - perturbative approaches are utterly worthless in this kind of strongly-bound QCD, and lattice calculations hit other roadblocks. I know that there are lattice models that in some situations predict cross sections from which you can extrapolate reasonable parton distributions, but I don't know that any such numerical computation is agreed upon to be "the" way to do it. A cursory search finds e.g. this paper about ab initio calculation of parton structure on the lattice, but someone with more expertise in the numerical aspects of the field will have to evaluate the merits of approaches like this.

Therefore, if we accept the parton model, I'd argue that we are forced to concede that the question of "how much" of the proton's mass is coming from where doesn't really make a lot of sense. In full QCD, an ab initio computation of hadronic masses (as in "Ab-initio Determination of Light Hadron Masses" by Durr et al.) is a bunch of really complicated lattice computation rather than something that would separate neatly into contributions to the mass of the three types you are looking for. Sure, the lattice model takes into account these three factors, but you don't get separate values for their contributions, the lattice model simply spits out a net hadron mass and that's it.

However, as the ab initio mass calculation I just linked shows, the sea quark contributions are highly relevant - the authors refer multiple times to how "light sea quark contributions" are necessary to make the result match the actual experimental masses. But I caution against trying to simply compute the difference between the lattice models with and without these contributions and assign that as "the sea quark contribution to mass", since the relation between lattice models and experiment is often much more subtle.

In "Proton Mass Decomposition from the QCD Energy Momentum Tensor" by Yang et al. it turns out that, while my pessimistic claims above are true in a naive sense, one can use lattice models that indeed do produce four distinct contributions to the hadron mass we naively expect. Note that the way in which they retrieve the actual contributions from the numerical results is rather convoluted and requires several intermediate quantities, passing through a parton distribution function in the form of the "momentum fractions" on the way. This lattice computation shows the following contributions:

  • Quark condensate: 9%
  • Quark energy: 32%
  • Gluon energy: 36%
  • Trace anomaly: 23%

We can note two things here:

First, if all we're looking for is "how much of this is the Higgs", then the straightforward and false answer is that it is the quark condensate, because it's the only contribution where the masses of the quarks play a role. The gluon energy term includes the $\beta$-function for the QCD coupling, and as knzhou's answer rightly points out the (non-)existence of the Higgs massively affects the running coupling. So even after all this trouble we still haven't a "true" figure for how much of this is the Higgs, although we could certainly combine this with the 60% from knzhou's answer to get yet another value (see below for why that's not straightforward either).

Second, about one fourth of the mass comes from a place that the parton model doesn't even know about, namely the trace anomaly. The cause of this is the "instanton-filled vacuum" of QCD - the QCD vacuum is not a simple perturbative vacuum but instead a $\theta$-vacuum, which is the superposition of many instantonic vacua, and so fermions and the bosons of the gauge field carry a global anomaly term that contributes to the action, and hence the energy.

The contribution of the Higgs to this part is murky because the paper does not compute the split contributions of the fermions and the gluons to this, but the $\beta$-function that depends on the Higgs only directly affects the gluonic anomaly. I'll leave to the reader to decide if this exercise actually brought us any closer to being able to talk about the Higgs' contribution to the hadron masses.

  • 1
    $\begingroup$ I am in full agreement with your deconstructive sense the question is a fool's errand. Ideally, you might try to compare a lattice estimate of the proton's mass with massive and almost massless quarks (hard !) and reassure yourself the corresponding difference for the lightest glueball estimate (really hard !) is negligible, or else subtract it, by fiat, from the former estimate. This is such a theological exercise I'm not sure what one could ever expect to learn from it. $\endgroup$ Apr 21, 2019 at 14:33
  • $\begingroup$ My intuition would be that as a binding energy, the gluon energy contribution would be negative. Indeed, Prof. Strassler claims that it is "often" negative, although I don't know exactly what he means by "often". Any insight on how to reconcile this with Yang et. al.'s claim that it's positive? $\endgroup$
    – tparker
    Apr 21, 2019 at 22:53
  • 2
    $\begingroup$ @tparker The "gluon energy" there is not "binding energy", it is just the expectation value of the gluonic Hamiltonian for the hadron state. Since gluons are confined, it seems non-sensical to me to talk about a "binding energy" for them, as the very notion of binding energy requires that there is a finite amount of energy that will separate the components into free particles. $\endgroup$
    – ACuriousMind
    Apr 21, 2019 at 23:23
  • $\begingroup$ Fair point. But we can still certainly have individual terms in the Hamiltonian having negative expectation values (assuming that the full QCD Hamiltonian is offset so that the ground state has zero energy), and Prof. Strassler seems to imply that this is the case for the gluonic Hamiltonian in the one-proton state. Any idea how to reconcile this claim with Yang et. al.'s results? $\endgroup$
    – tparker
    Apr 21, 2019 at 23:51
  • $\begingroup$ Actually, I guess there are two different ways to quantify the contribution of a term in the Hamiltonian to the proton's mass: (a) its expectation value with respect to the one-proton state, or (b) the difference between its expectation values w.r.t. the 1-proton and the vacuum states. Maybe the resolution of this apparent contraction is that the expectation value of the gluonic Hamiltonian is negative w.r.t both states (explaining Strassler's claim of a negative contribution), but higher for the one-proton state (explaining Yang's claim of a positive contribution). $\endgroup$
    – tparker
    Apr 22, 2019 at 0:15

I don't think it's valid to add up the rest masses of all the "sea" quarks and antiquarks -- a proton is a complex, strongly-coupled bound state of QCD, and I doubt one can just treat it as a bag of particles. The idea of a definite number of sea quarks probably doesn't make sense at all.

Here's a simpler argument that gets an intermediate estimate. Essentially by dimensional analysis, the mass of the proton is around $\Lambda_{\text{QCD}}$, which is around the GeV scale; we would expect to get a contribution of about this size whether the quarks were massive or not.

The Higgs field affects $\Lambda_{\text{QCD}}$ because it affects the running of the strong coupling constant; recall that $\Lambda_{\text{QCD}}$ is defined as the point where this coupling becomes large. With the Higgs field on, heavy quarks "freeze out" when the RG scale passes their masses, causing the running to accelerate, since quarks provide a charge screening effect. With the Higgs field off, no quarks ever freeze out, so the running to be slower overall, and hence lowering $\Lambda_{\text{QCD}}$. In this post it's argued that this lowers $\Lambda_{\text{QCD}}$ by 40%, so the Higgs is responsible for 60% of baryonic mass.

  • 1
    $\begingroup$ Isn't this the same sort of dimensional analysis that would predict gravity is really strong, i.e. whose failure in the gravitational case is the hierarchy problem? :P $\endgroup$
    – ACuriousMind
    Apr 21, 2019 at 10:38
  • $\begingroup$ @ACuriousMind Not really? If you trust naturalness, the hierarchy problem tells us that the unknown UV physics leads to the Higgs must be special in some way. This is not really relevant to the proton mass, because we already know exactly what the relevant UV theory is, namely QCD. $\endgroup$
    – knzhou
    Apr 21, 2019 at 10:44
  • $\begingroup$ It seems to me you don't need to argue that the mass of the proton - or any other QCD bound state - is "around $\Lambda_{QCD}$" at all. They can be as far as they like, what you need to argue is that their masses depend linearly on $\Lambda_{QCD}$ and on no other factor that varies when turning on/off the Higgs. $\endgroup$
    – ACuriousMind
    Apr 21, 2019 at 11:00
  • $\begingroup$ @ACuriousMind True! That does sound more plausible, though I'm not sure I can make the argument rigorous in either case. $\endgroup$
    – knzhou
    Apr 21, 2019 at 11:10
  • $\begingroup$ Well, I explicitly said in my question that the number of sea quarks is indefinite, but shouldn't it have a well-defined expectation value? $\endgroup$
    – tparker
    Apr 21, 2019 at 15:47

The present status is that the proton mass comes "from four sources [2] known as the quark condensate ( ∼9%), the quark energy ( ∼32%), the gluonic field strength energy ( ∼37%), and the anomalous gluonic contribution ( ~23%)". https://physics.aps.org/articles/v11/118#

Only the first contribution of 9% depends on the quarks having mass. The Higgs contribution is minor, although larger than the naive figure based on the sum of the quark masses.

  • $\begingroup$ Doesn't your 'update' contradict your original answer (which is still there saying A and B are correct)? The article you link does seem to be a direct answer to the question though, so +1 but I wish you'd remove the first paragraph and instead describe the paper. $\endgroup$
    – jacob1729
    Apr 21, 2019 at 11:18
  • 1
    $\begingroup$ Yes, it contradicts the figure of 1%. This should be replaced by 9%. I kept the previous answer for fairness. $\endgroup$
    – my2cts
    Apr 21, 2019 at 11:21
  • 1
    $\begingroup$ Am I right in thinking that this disagrees with @ACuriousMind's answer which comes to the conclusion that separating out the contributions is probably impossible? It's possible I'm missing something since I'm not knowledgeable in this area. $\endgroup$
    – jacob1729
    Apr 21, 2019 at 11:26
  • $\begingroup$ @jacob1729 I have amended my answer with a discussion of these computations, which I simply didn't know about until now. $\endgroup$
    – ACuriousMind
    Apr 21, 2019 at 16:13
  • $\begingroup$ @jacob1729 the separation is probably not exact but an estimate. $\endgroup$
    – my2cts
    Apr 21, 2019 at 16:34

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.