Choose your reading route
Begin at the puzzle or jump to the worked threshold. All twelve chapters remain open; expandable index groups are navigation only.
Two mixtures of long molecules may appear similar when described by an average molar mass, yet one can form a connected network and the other can remain a set of separate chains. The missing information is connectivity.
This guide makes the difference testable with a small branching model, two explicit tables, a changed-conversion example, practical checks and original scientific sources.
Chapter index
Find the distinction · Chapters 1–2
1. A longer chain is not necessarily a gel
2. Name the units and reactive sites
State the model · Chapters 3–5
3. Build a mixture before calculating
4. First and second factorial moments
5. Follow a branch after entering a unit
Calculate and check · Chapters 6–7
6. Find the ideal critical conversion
7. Change conversion and compare linear chains
Change the conditions · Chapters 8–10
8. Why finite averages do not describe architecture
9. Loops can use bonds without building a network
10. A mechanical crossover is not automatically the gel point
Practise and continue · Chapters 11–12
11. Run an independent check and answer FAQs
12. Continue through chemistry and the primary literature
A collection of independent long necklaces is not the same object as a fishnet. Giving both the same average number of beads would not make their networks equivalent.
Polymer molar mass describes an important aspect of molecules. Network formation asks a different question: whether bonds create continuing pathways through the material. Linear growth, branching and crosslinking should not be collapsed into one word.
The gel point is related to the emergence of large-scale connected structure. In actual materials its operational definition also requires attention to rheological and chemical evidence.
Let f be a unit’s nominal number of reactive functional sites. A difunctional unit (f=2) can extend a chain. A tetrafunctional unit (f=4) may form a junction with several paths away from it.
Let p denote an idealised independent bond-occupation probability for an available site. E[f] describes the average number of sites per building unit; E[f(f−1)] counts possible directed departures after entering one of those sites.
Our first model is an ideal locally tree-like random network with equally reactive sites. It ignores loop formation and other chemical constraints. We are using a mathematical lens to understand the branching mechanism, not claiming a universal chemical reaction rate law.
Take a teaching population of 200 building units: 150 difunctional and 50 tetrafunctional. This is a count fraction of 75 percent and 25 percent, not a mass fraction.
There are 150×2 + 50×4 = 500 nominal reactive sites. If p=0.60 and each connection consumes two occupied sites, 300 occupied sites correspond to 150 connections under the simplified paired-bond count.
The count by itself does not determine whether those 150 bonds create separate clusters or a sample-spanning component. We also need the distribution of junction opportunities.
The mean functionality is E[f] = 0.75×2 + 0.25×4 = 2.5. The branch-sensitive moment is E[f(f−1)] = 0.75×2 + 0.25×12 = 4.5.
The second quantity assigns greater weight to high-functionality units because they offer many more ways to continue after entering along a reacted bond. A single average would conceal those opportunities.
| Unit | Count share | f | E[f] contribution | E[f(f−1)] contribution |
|---|---|---|---|---|
| Difunctional | 0.75 | 2 | 1.50 | 1.50 |
| Tetrafunctional | 0.25 | 4 | 1.00 | 3.00 |
| Total | 1.00 | — | 2.50 | 4.50 |
Follow a randomly selected occupied connection into a unit. Once there, the unit offers f−1 other potential connections. Higher-functionality units are also more likely to be reached along an edge, so the correct average divides by E[f].
Under the graph assumptions, the expected number of continuing occupied branches is R = p E[f(f−1)] / E[f] = 1.8p for our mixture.
When R is below one, an exploration usually dies out in the infinite-population approximation; when it is above one, an extensive connected component becomes possible. This is a model threshold, not a direct reading from a rheometer.
Set R=1. The critical bond occupation is pc = E[f]/E[f(f−1)] = 2.5/4.5 = 5/9 ≈ 0.5556.
At p=0.50, R=0.90: the branching process is subcritical. At p=0.60, R=1.08: it is supercritical in the ideal large-system graph model.
A finite sample cannot have an infinite molecule. A random draw of 200 units also does not gel with certainty at an exact conversion. The threshold is a large-system reference point whose assumptions must be exposed.
CHAPTER 7 OF 12 · Calculate and check
7. Change conversion and compare linear chains
If every unit had f=2, the analogous branch factor would be R=p, which is below one for p less than one. A largely linear collection may contain long chains without branching into an extensive network at incomplete conversion.
In our mixed population, a quarter of units possess four nominal sites. Those junctions change how connections can spread; the average molecular mass alone does not identify them.
This table compares mathematical branching conditions, not experimentally proven gel points.
| Network model | p | R | Ideal interpretation |
|---|---|---|---|
| 75% f=2, 25% f=4 | 0.40 | 0.72 | Below threshold |
| 75% f=2, 25% f=4 | 0.50 | 0.90 | Below threshold |
| 75% f=2, 25% f=4 | 5/9 | 1.00 | Ideal threshold |
| 75% f=2, 25% f=4 | 0.60 | 1.08 | Above threshold |
| All f=2 | 0.60 | 0.60 | Linear-only control |
CHAPTER 8 OF 12 · Change the conditions
8. Why finite averages do not describe architecture
Number-average molar mass weights molecules by count; weight-average molar mass weights larger molecules more strongly. Two populations can agree on an average while differing in distribution, sequence and connectivity.
A soluble portion of a polymer network may have a measurable molar-mass distribution even when another portion is insoluble. Near or beyond gelation, an average of the extractable fraction need not describe the entire original sample.
Compare the existing molar-mass averages and chain populations guide. It is complementary, not a substitute for specifying junction topology.
CHAPTER 9 OF 12 · Change the conditions
9. Loops can use bonds without building a network
Intramolecular cyclisation consumes reactive groups while closing loops inside existing clusters. It may create no new connection between otherwise independent molecular pieces.
Unequal functional-group reactivity, diffusion limits, steric hindrance, concentration, temperature, phase separation, reversible bonds and unbalanced stoichiometry can all move a real material away from the ideal tree-like random graph.
That is why the toy pc must not be presented as a validated production threshold. Chemical details determine whether it is a helpful approximation or a misleading one.
CHAPTER 10 OF 12 · Change the conditions
10. A mechanical crossover is not automatically the gel point
Rheological tests measure how a material stores and loses deformation. A crossover between storage modulus G′ and loss modulus G″ at one chosen oscillation frequency is not universally identical to the gel point.
A defensible study may compare multiple test frequencies, time and temperature history, swelling and extraction, functional-group conversion and spatial homogeneity. A single measurement may be useful without being decisive.
Read the Rheology root and the Cox–Merz limitations guide to practise identifying when an instrument model is being pushed beyond its conditions.
CHAPTER 11 OF 12 · Practise and continue
11. Run an independent check and answer FAQs
Practice: keeping the mixture fixed, suppose p=0.40. Recalculate 1.8×0.40 = 0.72, below the ideal threshold. If a laboratory sample nevertheless shows elastic-like behaviour, investigate network type, frequency dependence and chemical assumptions instead of deleting the observation.
FAQ: Does R>1 mean every building unit is in the largest network? No. Does a finite average molar mass prove there is no crosslinking? No. Can physically assembled reversible systems behave like gels? Yes; the interaction model must match the chemistry.
Before accepting a gel-point claim, ask what was measured, how reactive sites were defined, whether loops and finite-size effects were considered, and what uncertainty the test carries.
CHAPTER 12 OF 12 · Practise and continue
12. Continue through chemistry and the primary literature
Use The Gold Standard of Polymer Science as the root. Related guides cover chain scission and number-average mass and step-growth imbalance. They address different mechanisms and are best read together.
Primary scientific context includes Chemical Society Reviews on covalent adaptable networks and gel points, network-formation research assessing Flory–Stockmayer ideality and loops and research extending the classical treatment with cyclisation and mobility constraints.
The transferable habit is to ask whether a summary of component size can possibly establish connectivity. Usually it cannot.
Continue with Polymer Science, The Gold Standard of Education and How X Works.
Sources checked 22 October 2026. All calculations are original teaching illustrations with limitations stated.