Thermodynamic Properties of Mozzarella Phase Transitions During Pizza Baking

1. Introduction

The moment at which a slice of pizza is lifted from the pie and the cheese stretches into a visually satisfying filament — colloquially the "cheese pull" — is among the most culturally significant and scientifically undercharacterized events in food science. Social media analyses have identified the cheese pull as the most-filmed aspect of pizza consumption, generating an estimated 847 million views annually on short-form video platforms (data on file). Yet despite this profound cultural prominence, the thermodynamic basis of the cheese pull remains poorly described in the peer-reviewed literature.

Mozzarella undergoes a complex sequence of phase transitions during baking that governs its final mechanical behavior. Below approximately 40°C, fresh mozzarella exists in a semi-solid gel state characterized by a cross-linked casein network with dispersed fat globules. Between 40–55°C, fat melting initiates, reducing viscoelastic modulus. The critical transition, which we term the Filament Formation Window (FFW), occurs between 55–70°C and involves the partial denaturation of casein proteins, fat coalescence, and the emergence of the characteristic stringy viscoelastic behavior that enables cheese pull. Above 80°C, oxidative browning and moisture loss degrade pull quality, eventually producing what consumers describe, with notable accuracy, as "rubbery."

Prior thermodynamic characterization of mozzarella has focused on dairy processing contexts (Schmelz et al., 2018) rather than pizza baking, where the presence of sauce moisture, direct oven radiation, and convective heat transfer from above and below create a thermal environment qualitatively different from the cheese vat. We address this gap using differential scanning calorimetry (DSC) and in-situ optical coherence tomography to characterize mozzarella phase behavior under authentic pizza baking conditions.

2. Materials & Methods

Mozzarella Samples. Three commercial mozzarella formulations were evaluated: fresh fior di latte (moisture content: 58.3 ± 0.4%), part-skim low-moisture (moisture: 47.1 ± 0.6%), and a premium artisanal buffalo mozzarella sourced from a specialty importer at considerable institutional expense (moisture: 62.1 ± 0.3%). All cheeses were sliced to uniform 4mm thickness using a PRI Cheese Precision Slicer (Model CPS-2, calibrated per PRI Instrumentation Protocol ICP-004).

Differential Scanning Calorimetry. DSC measurements were performed using a TA Instruments DSC 2500 at a heating rate of 5°C/min from 20°C to 100°C, in hermetically sealed aluminum pans. Samples were prepared both in isolation and layered on a 3mm tomato sauce substrate to assess sauce moisture contributions to thermal behavior. At least six replicate runs were performed per condition.

In-Situ Optical Coherence Tomography. A custom pizza baking chamber was constructed to allow real-time OCT imaging of the cheese surface during baking. The chamber incorporated an IR-transparent sapphire window in the oven ceiling through which a Thorlabs Telesto-II OCT system acquired cross-sectional images at 14Hz. This apparatus, which took approximately seven months to build and which collapsed once during commissioning, allowed direct visualization of cheese microstructure evolution at temperatures between 20°C and 95°C.

Cheese Pull Assay. Following baking, cheese pull performance was evaluated using the PRI Standardized Pull Protocol (SPP-2022): a 1-inch-wide cheese strip was gripped at the distal end and lifted at a controlled velocity of 2 cm/s using a texture analyzer (TA.XT Plus, Stable Micro Systems). Maximum filament length before fracture was recorded as the primary pull outcome. A panel of 24 trained evaluators also assigned subjective "pull satisfaction" ratings using a 7-point scale developed for this study.

This protocol was approved under PRI IRB #IRB-2022-PZZ-019. The study team acknowledges that designing a laboratory procedure for evaluating cheese pull required an unusually high number of pilot sessions.

3. Results

DSC thermograms revealed that the onset of the Filament Formation Window (FFW) occurred at 55.2 ± 1.4°C (fior di latte), 57.8 ± 1.1°C (part-skim), and 53.1 ± 1.9°C (buffalo), with the FFW terminating at 69.4 ± 1.8°C, 72.1 ± 1.3°C, and 67.8 ± 2.2°C respectively. The total enthalpy change across the FFW (ΔH_FFW) was −18.4 ± 0.8 J/g for fior di latte, −14.2 ± 0.6 J/g for part-skim, and −21.7 ± 1.1 J/g for buffalo. In the presence of the sauce substrate, FFW onset shifted to lower temperatures by a mean of 2.3°C (p = 0.008), consistent with moisture-mediated acceleration of casein network hydration.

OCT imaging revealed that fat channel coalescence — the structural correlate of filament formation capacity — occurs within a narrow time window of 18 ± 4 seconds following FFW onset. Crusts removed from the oven during this window demonstrated maximum filament lengths of 34.2 ± 4.1 cm (fior di latte), compared to 21.8 ± 3.6 cm for pizzas baked 30 seconds beyond the window. The decay in pull performance beyond the FFW was well-described by a first-order exponential model (r² = 0.94).

Subjective pull satisfaction ratings correlated strongly with measured filament length (r = 0.89, p < 0.0001), validating filament length as a mechanistic surrogate for the more labor-intensive subjective evaluation. Notably, the 24-member evaluator panel reached consensus that filaments between 28–36 cm were "ideal," filaments above 40 cm were "showboating," and filaments below 15 cm were "not really a pull, more of a separation."

4. Discussion

The Filament Formation Window concept provides the first thermodynamically grounded framework for optimizing pizza baking protocols relative to cheese pull performance. The narrow 14°C FFW, and the rapid post-window deterioration in pull quality, explain why precise oven timing is essential for pizzaioli seeking to maximize cheese behavior — and why the 18-second window for removal represents an operational challenge that merit formal training protocols.

The superiority of buffalo mozzarella in filament formation (lowest FFW onset, longest filaments) may explain its historical preference in traditional Neapolitan pizza, a choice previously attributed entirely to cultural tradition and proximity to water buffalo. We do not suggest that thermodynamic optimization was consciously practiced by 18th-century Neapolitan pizza-makers; we merely note that they appear, intuitively, to have gotten it right.

The finding that subjective pull satisfaction peaks at filament lengths of 28–36 cm imposes a practical upper bound on baking optimization: maximum filament length is not the target, but rather optimal filament length. This distinction is philosophically important and has practical implications for quality control. We intend to develop a standardized Pull Index incorporating both length and the panel's qualitative assessment in a subsequent paper.

5. Conclusion

We have characterized the thermodynamic properties of mozzarella phase transitions relevant to pizza baking and defined the Filament Formation Window (55–70°C) as the critical temperature regime for cheese pull optimization. Buffalo mozzarella demonstrates the most favorable FFW characteristics. Baking protocols should be calibrated to remove pizza from the oven during the FFW, ideally within 18 seconds of its onset. The in-situ OCT approach introduced here represents a significant advance in real-time cheese characterization during cooking.

Acknowledgments

The authors thank Dr. Nina Schmelz (ETH Zürich) for consultation on DSC methodology and for the loan of reference mozzarella standards used in method validation. The OCT system construction was supported by a PRI Infrastructure Grant (PRI-IG-2021-01) and required the combined efforts of six people and one particularly patient optical engineer. The evaluator panel is thanked for their precision, their patience, and their restraint in not eating study specimens before official evaluation was complete.

References

1. [1]

   Cheeseberg, M., & Crustworthy, F. (2024). Cheese pull mechanics: A viscoelastic analysis of mozzarella filament formation. Journal of Applied Pizzology, 23(1), 89–104. https://doi.org/10.1883/jap.2024.006

2. [2]

   Napolitano, F., Cheeseberg, M., & Romano, G. (2024). Optimal cheese-to-sauce ratios in Neapolitan-style pizza. Journal of Applied Pizzology, 23(1), 1–18. https://doi.org/10.1883/jap.2024.001

3. [3]

   Romano, G., & Saucington, A. (2023). Structural integrity of pizza crust under variable topping load conditions. Journal of Applied Pizzology, 22(3), 114–131. https://doi.org/10.1883/jap.2023.017

4. [4]

   Saucington, A. (2021). Sourdough fermentation duration and its effect on pizza dough extensibility. Journal of Applied Pizzology, 20(1), 1–19. https://doi.org/10.1883/jap.2021.003

5. [5]

   Schmelz, N., de la Salsa, R., & Umamida, H. (2018). Casein network denaturation in fresh mozzarella during thermal processing. Food Structure Quarterly, 14(2), 88–102.

6. [6]

   Schmelz, N. (2020). Comparative fat globule architecture in bovine and buffalo mozzarella: Implications for melt behavior. Rheology of Food Systems, 6(1), 14–29.