By Dr. Julien Huen and Frida Köning, ttz Bremerhaven
Why does bread go stale even in airtight packaging? Why do some margarines crack during lamination while
others fold like silk? And why does a perfectly good frozen dough sometimes collapse the moment it hits the oven? These three questions seem to belong to entirely different corners of the bakery, spanning texture science, fat technology and frozen logistics, yet they share a single, surprisingly elegant root cause: crystallization.
Baked goods are mosaics of components that can exist in crystalline or amorphous states, including starch, sugar, salt, fats and, in frozen products, water in the form of ice. How these components crystallize, and how quickly, determines not only how raw materials behave on the production line but also how a product looks, feels and tastes on the shelf weeks later. Far from being a topic confined to physics textbooks, crystallization is one of the most practically powerful, and most underused, levers available to anyone trying to control quality in baked goods. This article explores the thermodynamic principles behind it, and shows how understanding them can turn an invisible molecular process into a concrete tool for recipe and process design.
A simple case to start: water and ice
To understand why crystallization behaves the way it does, it helps to start with the simplest possible system: pure water. Cooled in a differential scanning calorimeter (DSC), distilled water does not freeze the moment it crosses 0°C. Instead, it keeps cooling, often down to around minus 22°C, before crystallization finally kicks in. (Figure 1) This phenomenon, known as supercooling, illustrates a principle that holds for every crystalline component in a baked product: crystallization is not instantaneous. Molecules need time to find their way into an ordered lattice, and close to the melting point, both the liquid and the solid state can coexist for a while before the more stable, lower energy state wins out. (Coupland, 2014)
Figure 1: DSC curve of distilled water. Cooling from 20°C to -25°C. Effect of supercooling visualized by the crystallization peak around -22°C. © Stefanie Kempel, Julien Huen, ttz Bremerhaven
The rate of cooling determines what the resulting crystal structure looks like. Cool quickly, and strong supercooling produces many crystal nuclei and a fine network of small crystals. Cool slowly, and fewer nuclei form, allowing existing crystals to grow large. (Coupland, 2014) (Figure 2) This is precisely why shock freezing, using cold, moving air at around minus 40°C or cryogenic gases such as liquid nitrogen, produces the fine ice crystal structure needed to protect product texture, while slow or fluctuating storage temperatures let crystals coarsen over time through recrystallization, as small crystals dissolve in favour of larger ones, gradually changing mouthfeel and structure. The practical takeaway is simple but easy to overlook: store frozen products as cold and as steadily as possible.
Figure 2: Polarisation microscopy images visualising the crystals in palm oil under different temperature conditions over time. Above: fast cooling creates many small crystals. Below: slowed cooling creates big crystals. The temperature profile over time determines the crystal structure, which in turn affects the rheological properties of the product. | © Julien Huen, ttz Bremerhaven
Fat: a much messier story
Where water crystallizes as a single, simple molecule, fats are a different challenge altogether. A typical fat blend contains a large number of different triglycerides, varying in chain length and degree of saturation (Gee, 2007). To complicate things further, each triglyceride can adopt several different crystal forms, known as polymorphs, each with its own melting point (Bayés-García, 2021). The result is a fat crystal network that builds up across multiple length scales, from individual crystal platelets to larger aggregates, ultimately determining whether a fat feels soft and spreadable or firm and brittle (Marangoni, 2020). (Figure 3)
Figure 3: Schematic overview of the stages of crystal network formation | graphic is AI generated illustration
This complexity matters enormously for laminating margarines used in croissants and puff pastry. During storage, crystallization continues. The crystalline fraction increases, crystals recrystallize into rounder and larger forms, and the fat network becomes increasingly interconnected, so firmness rises and the fat can eventually turn brittle (Zhang, 2014) (Figure 2). Tempering studies on industrial sheet margarine make the practical consequence visible: margarine sheets tempered at the wrong temperature crack and tear during lamination, while correctly tempered sheets fold smoothly into uniform layers (Gao, 2022). The manufacturing process for margarine moves the product through a sequence of carefully controlled temperature zones, from blending and emulsification at 50 to 80 °C down to crystallization, kneading and storage around 15 to 25°C, and every one of those steps offers a lever. Choosing fast crystallizing fat blends, applying mechanical work, allowing sufficient time for crystallization to complete, and keeping storage temperature stable all help avoid the cracking and reduced plasticity that plague lamination in practice.
“Because crystallization happens at a molecular and microstructural scale, most of it is invisible to the naked eye, which is why a combination of complementary analytical techniques is needed to capture both the structure and the energetics of the process.”
Dr. Julien Huen, ttz Bremerhaven
Sugar: crystal, glass or rubber?
Sucrose adds yet another twist. In its pure, anhydrous form it melts at a high 185°C and, unlike water, barely shows a crystallization peak on cooling in the DSC at all. Cooled quickly, molten sucrose does not recrystallize; it solidifies into an amorphous glass or rubber-like state
instead. (Figure 4) Once dissolved in water, the picture changes again. As a two-component sugar water system cools slowly, ice crystallizes first, concentrating the
remaining sugar in the unfrozen liquid until it becomes saturated and sugar itself starts to crystallize out. Cool the same system rapidly, and everything solidifies as an amorphous glass before crystallization has a chance to begin, which is exactly the principle exploited in shock freezing of sugar-containing products to avoid grainy, gritty textures from sugar crystallisation. (Sablani, 2010)
Figure 4: DSC curve of sucrose. Heating from 20 °C to 220 °C (yellow line). Sucrose melt sover 185°C. Cooling from 220°C to 20°C (blue line) of the same sample. No crystallization peaks as sugar solidifies into an amorphous glass or rubber-like state when cooled quickly.
© Stefanie Kempel, Julien Huen, ttz Bremerhaven
Starch: the slow road to staleness
The most familiar crystallization story in baking is also the most economically important: starch retrogradation, the molecular process behind staling. Starch granules consist of two polymers, amylose and amylopectin, arranged in alternating crystalline and amorphous
regions. (Wang, 2015) During baking, heat and water disrupt this order: amylopectin crystallites melt, amorphous regions hydrate, some amylose leaches out into the surrounding matrix, and amylose-lipid complexes melt as well. The granule swells, while partially retaining its structure. (Figure 5)
The reverse process begins the moment the product starts to cool. Amylose, freed during baking, recrystallizes quickly into a three-dimensional network, within hours. Amylopectin recrystallizes far more slowly, over days, but it is this slow process that is largely responsible for the gradual firming and going stale that bakers know all too well, visible under the microscope as a progressive densification of the starch matrix between a freshly baked loaf and one that has sat for a week. (Hug-Iten,1999)
Figure 5: Fluorescence microscopy images of native starch (above) and gelatinised starch (below) | © Ainhoa Ramos, ttz Bremerhaven
Keeping bread soft for longer
Because retrogradation is a kinetic, time-dependent process rather than an instant event, it can be slowed, though not stopped, through several complementary strategies. These include avoiding storage temperatures around 4°C, where recrystallization is fastest, steering clear of a roughly one to one starch to water ratio that favors rapid network formation, adding hydrocolloids, polysaccharides or fibres that physically slow the diffusion of amylose and amylopectin molecules, using enzymes such as alpha amylase or maltogenic amylase to modify the starch chains themselves, and including emulsifiers and lipids that bind amylose into V type amylose lipid complexes, which interfere with crystalline network formation. Each of these is, in effect, a different way of buying time against the same underlying thermodynamic drive towards order.
Making the invisible visible: how crystallization is studied
None of the mechanisms described above can be managed on instinct alone. They need to be measured. Because crystallization happens at a molecular and microstructural scale, most of it is invisible to the naked eye, which is why a combination of complementary analytical techniques is needed to capture both the structure and the energetics of the process.
Fluorescence and polarization microscopy make crystalline and amorphous regions visible directly, image by image. Under polarized light, crystalline structures such as starch granules or fat crystals appear bright against a dark, amorphous background, because their ordered molecular arrangement rotates polarised light in a way that disordered regions do not. (Figure 2). Staining with fluorescent dyes can further highlight specific components, for example distinguishing intact starch granules from gelatinised, hydrated regions. (Figure 5) This makes it possible to literally watch a structure change, for instance to follow how a starch granule’s crystalline core erodes during baking, or how a fat crystal network coarsens during storage.
Differential scanning calorimetry (DSC) takes a different, complementary approach. Rather than imaging structure, it measures the heat absorbed or released as a sample melts or crystallises. The size, shape and position of the resulting thermal peaks reveal not just whether crystallisation or melting is occurring, but how much material is involved, at what temperature and how readily, turning an otherwise invisible phase transition into a quantifiable curve (Figure 1, Figure 4).
Texture analysis closes the loop between microstructure and the property that ultimately matters to the customer: how a product feels. Because crystallisation directly governs the firmness, brittleness or elasticity of fats, starches and frozen structures, mechanical measurements of hardness, spreadability or fracture behavior provide an indirect but highly practical readout of how far a crystallization or recrystallisation process has progressed, and how it will be perceived at the bakery counter or kitchen table.
Used together, these three techniques (visual, thermal and mechanical) provide a fairly complete picture of crystallization behavior: what is forming, how much, how fast, and what it ultimately means for product quality. This combination has, in fact, been the analytical backbone of 15 years of applied research into crystallization phenomena in baked goods and related food systems at ttz Bremerhaven, spanning projects on frozen dough, laminating fats and starch-based products alike.
Microscopic view of sucrose crystals in polarized light | picture: © Pawel Burgiel – stock.adobe.com
One set of rules, many applications
What unites ice, fat, sugar and starch crystallization is that they all obey the same handful of thermodynamic rules: nucleation and growth, the trade-off between cooling rate and crystal size, and the long, slow march toward the most stable, and often least desirable, structural state. Once this is understood, crystallization stops being a background nuisance and becomes a parameter that can be designed for. It can be accelerated where a product needs to set quickly, slowed where shelf life depends on staying soft or workable, or suppressed altogether where an amorphous, glassy structure is the goal. Whether through recipe formulation, selecting raw materials with the right crystallization behavior, or through process control, meaning temperature management, seeding and mechanical working, the lever is the same one, applied to a different ingredient each time.
“Baked goods are mosaics of components that can astarch, sugar, salt, fats and, in frozen products, water in the form of ice. How these components crystallize, and how quickly, determines not only how raw materials behave on the production line but also how a product looks, feels and tastes on the shelf weeks later.”
Frida Köning, ttz Bremerhaven
For bakeries and ingredient suppliers, this shared logic is good news. A quality issue that looks unique to one product, be it a stale loaf, a cracking margarine sheet or a frozen dough that spreads too much in the oven, is very often a variation on a problem that might has already been solved elsewhere in the industry. Recognizing crystallization as the common thread, and having the right combination of microscopy, calorimetry and texture analysis to track it, is what turns recurring quality complaints into a manageable, predictable part of process design. This is exactly the kind of cross-product expertise that institutes such as ttz Bremerhaven have built up over more than a decade of applied research into crystallization behavior across the baking and food industry.
References
Bayés-García, L., Sato, K. and Ueno, S. (2021). Polymorphism of Triacylglycerols and Natural Fats. In Bailey’s Industrial Oil and Fat Products, F. Shahidi (Ed.).
Coupland, J. N. (2014). An introduction to the physical chemistry of food. Springer.
Gao, H., Gao, W., Yang, X., Liu, Y., & Wang, Z. (2022). Effects of different tempering temperatures on the properties of industrial sheet margarine. RSC Advances, 12(36), 23311–23321.
Gee, P. T. (2007). Analytical characteristics of crude and refined palm oil and fractions.
European journal of lipid science and technology, 109(4), 373-379.
Hug-Iten, S., Handschin, S., Conde-Petit, B., & Escher, F. (1999). Changes in starch microstructure on baking and staling of wheat bread. LWT-Food Science and Technology, 32(5), 255-260.
Marangoni, A. G., Van Duynhoven, J. P., Acevedo, N. C., Nicholson, R. A., & Patel, A. R. (2020). Advances in our understanding of the structure and functionality of edible fats and fat mimetics. Soft Matter, 16(2), 289-306.
Sablani, S. S., Syamaladevi, R. M., & Swanson, B. G. (2010). A Review of Methods, Data and Applications of State Diagrams of Food Systems. Food Engineering Reviews, 2(3), 168–203.
Wang, S., Li, C., Niu, Q., & Wang, S. (2015). Starch Retrogradation: A Comprehensive Review. Comprehensive Reviews in Food Science and Food Safety, 14(5), 511–680.
Zhang, X., Li, L., Xie, H., Liang, Z., Su, J., Liu, G., & Li, B. (2014). Effect of temperature on the crystalline form and fat crystal network of two model palm oil-based shortenings during storage. Food and bioprocess technology, 7(3), 887-900.