Fouling accounts for 30–40 per cent of membrane operating cost, and cake-resistance models describe the symptom rather than the cause. The cause is adhesion between extracellular polymeric substances and the membrane surface — an interfacial problem, and one that predicts which cleaning chemistry will work.
EPS is defined by how you extract it
There is no single substance called EPS. The term covers whatever the extraction method releases, and the fractions behave quite differently in a membrane system.
| Fraction | Typical extraction | Composition | Fouling role |
|---|---|---|---|
| Soluble microbial products (SMP) | Centrifugation, then 0.45 µm filtration | Proteins, polysaccharides, humics in solution | Pore blocking; the irreversible component |
| Loosely bound EPS | Mild chemical or thermal release | Polysaccharide-rich, protein-poor | Cake layer bulk and its compressibility |
| Tightly bound EPS | Heat extraction, ~80 °C | Protein-rich, hydrophobic amino acids | Adhesion to the surface; biofilm initiation |
Interaction energy: why DLVO is not enough
Classical DLVO theory adds van der Waals attraction to electrostatic double-layer repulsion. In water that is insufficient — it omits the polar interaction that dominates at short range in hydrogen-bonding media. The extended form adds a Lewis acid–base term:
Utotal = ULW + UEL + UABLW = Lifshitz–van der Waals (apolar); EL = electrostatic double layer; AB = Lewis acid–base (polar, hydrogen bonding)
Each surface is characterised by an apolar component and a pair of polar parameters, obtained from contact angle measurement with three probe liquids:
γ = γLW + 2√(γ+γ−)γ+ = electron-acceptor parameter; γ− = electron-donor parameter (mJ m−2)
Representative surface energies
| Material | γLW | γ+ | γ− |
|---|---|---|---|
| PVDF | 43.2 | 0.02 | 9.8 |
| PES | 44.0 | 0.05 | 11.2 |
| Ceramic (Al2O3) | 35.6 | 1.2 | 45.0 |
| PTFE | 19.8 | 0.1 | 0.5 |
Values in mJ m−2. Note the electron-donor column: the ceramic is an order of magnitude more electron-donating than the polymers, and that single difference drives the result below.
Free energy of adhesion predicts irreversible fouling
The free energy of adhesion between foulant and membrane, mediated by water, is the sum of the apolar and polar contributions evaluated for the three-phase system:
ΔGadh = γfm − γfw − γmwf = foulant, m = membrane, w = water. Negative means adhesion is spontaneous.
Evaluating this for tightly bound EPS at 25 °C gives the practically important contrast:
| Membrane | ΔGadh with TB-EPS | Interpretation |
|---|---|---|
| PVDF | ≈ −8.4 mJ m−2 | Spontaneous adhesion; irreversible fouling expected |
| Ceramic Al2O3 | ≈ +2.1 mJ m−2 | Net repulsive; fouling largely hydraulically reversible |
Critical flux follows from adhesion energy
Critical flux is the flux below which no appreciable fouling occurs. It is not an arbitrary operating rule — it is the point at which the hydrodynamic drag carrying a particle to the membrane balances the interfacial force resisting attachment. For unmodified PVDF against protein-rich tightly bound EPS, that balance lands at roughly 18 L m−2 h−1, which is consistent with operating experience on municipal MBRs.
Hydrophilic surface modification — PEGylation, plasma treatment, blended hydrophilic additives — raises the membrane electron-donor parameter, makes ΔGadh less negative, and lifts critical flux to the 25–30 LMH range. That is the mechanism behind every “low-fouling” membrane claim; ask suppliers for contact-angle data, not just flux curves.
Flux, TMP and aeration trade-offs are covered in membrane bioreactor design and ultrafiltration system design.
Matching the cleaning chemistry to the foulant
Blanket hypochlorite cleaning is the commonest error in MBR operation. Hypochlorite attacks protein; it does little to a polysaccharide cake, and it oxidises the membrane polymer while failing to solve the problem.
| Dominant mechanism | EPS fraction | Agent | Mode of action | Typical conditions |
|---|---|---|---|---|
| Pore blocking | SMP, proteins | Sodium hypochlorite | Oxidative cleavage of protein structure | 500–2,000 mg/L, ambient, 2–4 h |
| Cake layer | Loosely bound, polysaccharide | Caustic | Saponification and solubilisation | 0.5–1.0% w/v, ~35 °C, 4–6 h |
| Adhesive biofilm | Tightly bound, protein and lipid | Citric acid plus surfactant | Chelation of bridging Ca2+ and Mg2+ | ~2% citric, ~40 °C, 6–8 h |
Characterise before you specify
Four measurements on a fouled coupon settle the protocol:
- Contact angle goniometry with three probe liquids — gives the surface energy components and hence ΔGadh.
- FTIR — the protein amide I band near 1,650 cm−1 against the polysaccharide C–O band near 1,030 cm−1.
- XPS — surface N/C ratio indicates protein; O/C indicates polysaccharide.
- AFM force spectroscopy — direct measurement of adhesion force, if the interfacial calculation is being challenged.
Design rules
- Measure the protein-to-polysaccharide ratio before selecting a membrane material or a cleaning regime. It predicts fouling propensity better than MLSS, SRT or flux alone.
- Specify hydrophilic or ceramic surfaces where the mixed liquor is protein-rich. The adhesion energy, not the pore size, governs the irreversible fraction.
- Set operating flux below critical flux rather than designing to a supplier’s peak. Roughly 18 LMH for unmodified PVDF, 25–30 LMH for hydrophilically modified.
- Match CIP chemistry to the dominant fraction. Hypochlorite on a polysaccharide cake wastes reagent and consumes membrane life.
- Re-characterise after any process change — a shift in SRT, load or industrial input moves the PN/PS ratio, and the CIP protocol should move with it.
Frequently asked questions
Why do ceramic membranes foul less than polymeric ones?
Because their surfaces are strongly electron-donating, which makes the free energy of adhesion with protein-rich EPS slightly positive rather than negative. Attachment is not spontaneous, so most of the fouling stays hydraulically reversible. It is an interfacial property, not a matter of smoothness or strength.
Is hypochlorite always the right cleaning agent for an MBR?
No, and defaulting to it is the commonest operating error. It attacks protein but does little to a polysaccharide cake, while still oxidising the membrane polymer. Identify the dominant fraction by FTIR or XPS first — caustic for cake, chelant plus surfactant for adhered biofilm, hypochlorite for pore blocking.
What is a realistic critical flux for a municipal MBR?
Around 18 LMH for unmodified PVDF against protein-rich mixed liquor, rising to 25–30 LMH for hydrophilically modified surfaces. Treat supplier peak-flux figures as a ceiling that assumes favourable feed, not as a design point.
Can I predict fouling from MLSS alone?
Poorly. MLSS sets cake thickness but not adhesion. Two plants at identical MLSS can differ several-fold in irreversible fouling if one runs protein-dominated EPS and the other polysaccharide-dominated. The PN/PS ratio is the more informative measurement.