At the 2025 Ig Nobel Prizes, a team of physicists answered a question every home cook has wrestled with: how do you make Cacio e Pepe (just pasta, pecorino cheese and pepper) without the sauce going lumpy?
They tested how much starch, how much water and which helpers keep the sauce smooth as it heats up. The result is a map of the sauce’s “phases”, including the dreaded Mozzarella Phase, when the cheese gives up and turns into one big rubbery clump.

Cheese mixed with plain water (top), normal pasta water (middle) and extra-starchy pasta water (bottom), then slowly heated. The more starch, the fewer and smaller the lumps. With plain water, the cheese collapses into the Mozzarella Phase from about 70 °C.
01
Too little starch and the cheese melts into one big rubbery lump.
Starch from the pasta water stops cheese proteins sticking to each other.
Equal parts cheese and water is the easiest mix to break. A little more or less water is safer.
A pinch of this salt stops lumps completely, but it isn’t traditional.
Each square is the sauce at one temperature. Dark means it has split into one giant lump.
02
They described the sauce with one equation from physics, the binary mixture free-energy model. It treats the sauce as just two ingredients, cheese proteins and “everything else” (water, starch, salt and fat), and scores whether that mix would rather stay blended or separate. Three things pull against each other:
“ideas want to spread out” → proteins and water naturally mix.
“friendship or rivalry” → proteins either cooperate with water (smooth sauce) or gang up on each other (clumps).
“energy level of the party” → higher heat makes proteins more likely to denature and aggregate.
Pick how much water goes in, as a percentage of the cheese weight. The dot shows roughly the temperature where lumps start. Equal parts (100 %) is the touchiest mix.
Equal weights of cheese and water: the bottom of the U, so this is the mix most likely to go lumpy.
03
The paper ends with a “scientific recipe”. These are the numbers that matter.
Starch, compared with the cheese weight. The sweet spot for a smooth sauce.
Water, compared with the cheese weight. A little less than equal parts is safer.
Around here the cheese starts to clump, so let the pasta cool for a moment before mixing.
A properly stabilised sauce can be reheated this hot in a pan without splitting.
Starch, % of cheese weight
Trisodium citrate, % of cheese weight
Tradition check
Citrate works even better than starch, but it isn’t in the traditional recipe, and the authors found it slightly dulls the cheese flavour. Use it in a Roman kitchen at your own risk.
Sauce for two
04
While the idea of exploring particular dish might seem kind of excessive, but from it comes a serious exploration of phase separation, emulsions, and protein aggregation; concepts that apply not only to food but also to materials science, biochemistry, and even engineering. Understanding how mixtures behave under heat and concentration changes helps. Plus, it’s a reminder that cooking is chemistry in action.
05
Think of the pasta sauce as a dance floor...
06
I love how this paper turns a kitchen challenge into a physics experiment. As a former international student, there were times when I spent an evening trying to experiment different ways to cook pasta. It’s a perfect example of how engineering curiosity meets everyday life. The “Mozzarella Phase” is not just a funny name — it’s a vivid reminder that even simple recipes hide complex thermodynamics. It confirmed something I'd felt intuitively — that a little starch changes the whole texture of a sauce. While the flavour of the food itself is more of personal taste, this paper gives an idea of what a true preparation to which we could do for our creation. While the application of this particular equation and model is designed for Cacio e Pepe, I believe some dish in Asia and Africa could also benefit in this, such as Tsukemen of which it uses thick broth as dipping sauce to enrich its flavour. For me, this resonates with the idea that science and music share a rhythm: both are about finding harmony in systems, whether it’s molecules in sauce or notes in a melody.
← Back to all talks
Discussion
Register with your email to join the discussion
No comments yet. Be the first to share your thoughts.
At the 2025 Ig Nobel Prizes, a team of physicists answered a question every home cook has wrestled with: how do you make Cacio e Pepe (just pasta, pecorino cheese and pepper) without the sauce going lumpy?
They tested how much starch, how much water and which helpers keep the sauce smooth as it heats up. The result is a map of the sauce’s “phases”, including the dreaded Mozzarella Phase, when the cheese gives up and turns into one big rubbery clump.

Cheese mixed with plain water (top), normal pasta water (middle) and extra-starchy pasta water (bottom), then slowly heated. The more starch, the fewer and smaller the lumps. With plain water, the cheese collapses into the Mozzarella Phase from about 70 °C.
Normal pasta water here is 100 g of pasta cooked in 1 litre of water. The extra-starchy version is pasta water “risottata”: boiled down in a pan to a third of its weight so the starch concentrates. The cheese and liquid were heated from 50 to 95 °C and photographed at each step. Higher starch means aggregates are smaller and appear at higher temperatures.
01
Too little starch and the cheese melts into one big rubbery lump.
Starch from the pasta water stops cheese proteins sticking to each other.
Equal parts cheese and water is the easiest mix to break. A little more or less water is safer.
A pinch of this salt stops lumps completely, but it isn’t traditional.
Each square is the sauce at one temperature. Dark means it has split into one giant lump.
How they measured it: each sample was spooned onto a backlit Petri dish and photographed. Software picked out the dark cheese clumps, fitted each one with an ellipse and averaged their lengths into a mean aggregate size in millimetres. That number is the colour of every square in the paper’s phase diagram below. Starch is given as a percentage of the water, with cheese and water mixed 1 : 1; citrate as a percentage of the cheese weight.
“Sauce holds” in Chart A still allows small lumps: with starch, little aggregates remain even when the Mozzarella Phase is gone, while 2 % citrate leaves no visible aggregates at all. Chart A is read from the paper’s Figs. 2(b) and 5.

Paper Fig. 2. (a) Phase diagram of the sauce state against starch percentage (relative to the water) and temperature. Each box is a photo of the sample, outlined in a colour that shows its mean aggregate size. (b) The same diagram smoothed with a Gaussian filter. (c) A continuous map from kernel regression. The grey region is the Mozzarella Phase, where the clump is as big as the whole sample.
02
They described the sauce with one equation from physics, the binary mixture free-energy model. It treats the sauce as just two ingredients, cheese proteins and “everything else” (water, starch, salt and fat), and scores whether that mix would rather stay blended or separate. Three things pull against each other:
“ideas want to spread out” → proteins and water naturally mix.
“friendship or rivalry” → proteins either cooperate with water (smooth sauce) or gang up on each other (clumps).
“energy level of the party” → higher heat makes proteins more likely to denature and aggregate.
Pick how much water goes in, as a percentage of the cheese weight. The dot shows roughly the temperature where lumps start. Equal parts (100 %) is the touchiest mix.
{{bin.note}}
The U-shaped line is the binodal: the boundary between a well-mixed sauce and one where protein aggregates form, measured with starch fixed at 1 %. Plotted against the protein mass fraction φ (here {{bin.phi}} for your choice), its lowest point sits just above 60 °C. Physicists call that a lower critical solution temperature: unusually, this mixture separates when heated rather than cooled.
The team fitted the equation above to this boundary. As the sauce heats, both χ (how sticky the proteins are) and n (their effective size) rise in a straight line, because heat unfolds the whey proteins and lets them latch onto each other and onto casein micelles. With those two lines, the model redraws the measured boundary almost exactly.

Paper Fig. 4. (a) Cheese is mostly casein, packed in micelles with calcium phosphate, plus a little whey protein. Heat denatures the whey, which then aggregates with itself and with the micelles, and the micelles aggregate too. (b) The interaction χ and relative size n obtained from the experiments at each temperature. (c) Theory (crosses) against the experimental binodal (line).
About Chart B: the temperatures come from a parabola fitted to the binodal in paper Figs. 3(c) and 4(c), using the reported φ = 0.134 for a 1 : 1 mix to convert water percentage into φ. Dashed ends are extrapolated beyond the tested range, so treat them as rough guides.
03
The paper ends with a “scientific recipe”. These are the numbers that matter.
Starch, compared with the cheese weight. The sweet spot for a smooth sauce.
Water, compared with the cheese weight. A little less than equal parts is safer.
Around here the cheese starts to clump, so let the pasta cool for a moment before mixing.
A properly stabilised sauce can be reheated this hot in a pan without splitting.
Starch, % of cheese weight
Trisodium citrate, % of cheese weight
Starch and citrate stop lumps in different ways. Gelatinised starch gets in between the cheese proteins, so lumps shrink gradually as you add more. Trisodium citrate grabs the calcium ions that hold casein proteins together; below 2 % there isn’t enough of it and the sauce still collapses (just later), from 2 % it stays completely smooth. The paper tested citrate at 1 %, 2 % and 3 % with cheese and water mixed 1 : 1.

Paper Fig. 5. The sauce with 0–3 % trisodium citrate (relative to cheese weight), heated from 50 to 95 °C.
Tradition check
Citrate works even better than starch, but it isn’t in the traditional recipe, and the authors found it slightly dulls the cheese flavour. Use it in a Roman kitchen at your own risk.
Sauce for two
04
While the idea of exploring particular dish might seem kind of excessive, but from it comes a serious exploration of phase separation, emulsions, and protein aggregation; concepts that apply not only to food but also to materials science, biochemistry, and even engineering. Understanding how mixtures behave under heat and concentration changes helps. Plus, it’s a reminder that cooking is chemistry in action.
05
Think of the pasta sauce as a dance floor...
06
I love how this paper turns a kitchen challenge into a physics experiment. As a former international student, there were times when I spent an evening trying to experiment different ways to cook pasta. It’s a perfect example of how engineering curiosity meets everyday life. The “Mozzarella Phase” is not just a funny name — it’s a vivid reminder that even simple recipes hide complex thermodynamics. It confirmed something I'd felt intuitively — that a little starch changes the whole texture of a sauce. While the flavour of the food itself is more of personal taste, this paper gives an idea of what a true preparation to which we could do for our creation. While the application of this particular equation and model is designed for Cacio e Pepe, I believe some dish in Asia and Africa could also benefit in this, such as Tsukemen of which it uses thick broth as dipping sauce to enrich its flavour. For me, this resonates with the idea that science and music share a rhythm: both are about finding harmony in systems, whether it’s molecules in sauce or notes in a melody.
← Back to all talks
Discussion
Register with your email to join the discussion
{{cm.err}}
{{cm.err}}
No comments yet. Be the first to share your thoughts.
{{c.text}}