Part 2: Diffusion

Coffee fermentation can be deconstructed into three parallel processes: fermentation, diffusion and “pathways”.  In part 1 we examined the role of the namesake process, fermentation, identifying it as a separate process responsible for the generation of aroma and flavour compounds through the microbial metabolism of the coffee’s mucilage externally to the bean. 

In this article we will tackle the important role of diffusion in coffee fermentation.  Although, as we shall see in part 3, the bean itself can change its own composition during the fermentation process, the main influence on cup quality comes from diffusion of the aroma and flavour precursors produced during fermentation.

Diffusion is the movement of molecules from regions of high concentration to low concentration.  The molecules don’t “know” that they are in regions of high or low concentration, it’s just that on average there are more molecules moving from high-to-low than from low-to-high.  After enough time, concentrations reach an equilibrium.  The amount of time this takes depends on the molecules (smaller ones move faster and, as we shall shortly see, very large ones hardly diffuse at all), the distance they have to travel, and the diffusivity of the medium through which they travel – a dense bean parchment is slower to move through (lower diffusivity) than a swollen mucous.

In coffee fermentation, the regions of “low” and “high” concentration are changing via two dynamic processes: the external generation of aromas and flavours through fermentation and the internal alternative reaction pathways, which will be discussed in the third article of this series.  These dynamics lead to continuous fluctuations in the concentrations of potentially valuable flavour precursors both inside and outside the bean, adding to the challenge of controlling the eventual flavour components inside the bean.  

Adding to the complexity, the dynamic exchange of aroma and flavour components occurs across multiple external layers of the bean, each with their own thicknesses and diffusivities.  To the best of the author's knowledge, these bean properties are often neglected in the literature.  Given that these bean barrier properties significantly impact the quality of fermented coffees, we will now provide a summary of the current understanding on this topic.

Let us start by confronting a belief within the coffee community that microbial enzymes are capable of penetrating the green bean and that these enzymes are responsible for elevating coffee’s flavour during fermentation.  For the vast majority of microbes used in coffee fermentation this is untrue [1].  Nature has engineered coffee beans to be incredibly dense.  This is likely not without intention as it serves as a natural defence mechanism against foreign enzymes that may penetrate and destroy the seed, i.e. threaten its survival.  To better illustrate this point let us consider an industrial example which has been exhaustively researched:

Asparaginase is an enzyme that weighs approximately 140-150 kDa [2] that has been successfully used in starchy foods to reduce asparagine levels.  Free asparagine within foodstuffs, including green coffee, has been related to the formation of acrylamide upon heating [3], e.g. during roasting.  As acrylamide is a “probable human carcinogen” according to the International Agency of Research on Cancer (IARC), governments regulate the levels found within foods to ensure consumers are not exposed to levels beyond their body’s natural capacity to deal with them [4], i.e. lead to detrimental health consequences.  These regulatory measures incentivized industry to innovate around the removal of asparagine from green coffee beans.  However, when asparaginase was attempted as an enzymatic approach the results were discouraging.  Concluding that, “[t]he green coffee bean is a very dense, hardly permeable raw material and additional processing steps (steam treatment and soaking in a water bath) are required for the enzyme to be effective,” (FoodDrink Europe, 2019).  In other words the enzymatic approach did not work because the enzymes were too large and could not penetrate the bean’s structure even after significant coaxing.

This raises the question: if enzymes are too large to diffuse into a green bean, how small must the molecule be to enter a green coffee bean?  

Currently, there are limited data available on this topic, however several studies into tangential topics may provide us some clues.  “In-bean” [5] experiments where green beans are initially exhaustively extracted with hot water can reveal to us the molecules that are capable of diffusing out of the bean.  If these molecules can leave, it only makes sense that they also can move in the reverse direction and go back in.  These green bean extracts routinely contain sugars, amino acids, minerals, organic acids, chlorogenic acids and coffee alkaloids (Poisson et al., 2009 [6]).  Müller and Hofmann (2005) categorized this extract into low molecular weight (<1kDa) and high molecular weight (>1kDa) portions, accounting for 98% and 2% of the yield, respectively.  This provisional maximum weight limit of around 1kDa for compounds with the capacity to diffuse out of green beans (Müller and Hofmann, 2005) is optimistic compared to the well-known rule of thumb in the pharmaceutical literature that it is hard to get molecules > 500Da to diffuse in or out of biological systems.  

This weight restriction is likely worth challenging as the extraction method used in these experiments is quite harsh, with hot water baths reaching temperatures of 95°C.  These conditions may not seem harsh, but in the context of the bean’s structure they may be.  The bean's cellular membrane begins to disrupt at >40°C during post-harvest drying [7], suggesting that the bean’s integrity may have been compromised under these experimental conditions.  If the membrane has been disrupted then this could allow larger molecules, that would have otherwise been retained, to diffuse out.  In other words I suspect that ~1 kDa is an overestimation and the pharma max of 500 Da is more plausible.  Indeed, a further rule of thumb is that speed of diffusion goes inversely as weight squared, so if it takes 1hr for a typical flavour molecule of weight 100 to diffuse, it will take 25 hours for a 500 Da molecule.  Moreover, we know from research into bean development that sucrose needs to be broken down into fructose and glucose before diffusing into the bean (the simple rule suggests that it’s 4x slower), further suggesting that the weight limit for diffusion may be significantly lower.  In other words, those within the coffee community who believe polysaccharides and peptides can seep into a green bean and enhance fermented coffee’s flavour are likely to be mistaken [8].  This all depends on what the actual weight restrictions are for green bean coffee diffusion.

The practical implications of these weight restrictions are substantial.  If we focus on the fermentation tank and were to isolate the molecules capable of diffusing into the bean, e.g. <500Da, how much of the fermented mucilage would remain [9]?  Experimentally one could assess this through filtration or dialysis of the fermented mucilage, but as far as this author is aware, this has not been done or at least publicly disclosed. 

Please feel free to guess, “hypothesize”, below what percent of fermented mucilage meets these weight restrictions and consequently has the potential to contribute to fermented coffees’ quality.

If you chose anything below 50% than what you are saying is that the majority of the fermented mucilage is waste, i.e. a by-product, and the desirable aroma and flavours produced during fermentation are minor products, revealing an inefficient process.  Inefficient processes tend to be less cost effective as the risks and rewards of the process have not been optimized, but before tackling this subject further there is one last barrier, literally, to confront.

Many studies on coffee fermentation are conducted outside of origin countries, and typically use green coffee beans, which do not reflect the actual process.  When coffee fermentation is conducted in origin countries, the green bean is enclosed within parchment, surrounded by mucilage and an outer layer of skin, all of which play a role in the fermentation process.  The diffusion of aroma and flavour precursors is then not simply through the green bean but through all or a subset of these layers, depending upon the post-harvest processing method being used.  While the outer layers may be partially removed, the parchment remains attached to the bean until the final drying stage, so at minimum the parchment layer is present.  Therefore, the microbial metabolites that need to penetrate the coffee bean during fermentation must not only diffuse into the bean, but also pass through the residual layers of the coffee cherry [10].  Making this process potentially a chemical steeple-chase.  Luckily, a recent series of publications by Hadj Salem et al. (2020, 2023) has shed light on this topic for us.

Hadj Salem and colleagues (2020, 2023) used known concentrations of microbial metabolites, refer to table below, [11] representing different categories of molecules; sugars, amino acids, aromas, organic acids . . . etc., to study their diffusion into the bean as well as past the various cherry layers.  Perhaps only a coincidence, but it’s interesting to note that none of the compounds investigated by these authors were larger than 200 Da.  

If you're curious about how Hadj Salem et al. (2020, 2023) distinguished between naturally occurring compounds, e.g. amino acids and sugars, and those intentionally introduced for the experiment, they used isotopically labelled compounds [12] throughout their experiments.  Isotope labelling is like assigning team jerseys to compounds, allowing one to tell the teams apart at a glance on the field, or in an experiment.

Their initial experiments, reflective of studies made abroad, with only the green bean, demonstrated that the contrast in concentration gradient, between the exterior and interior of the bean, gave rise to an initial greater rate of diffusion into the bean that gradually tapered off as equilibrium conditions were established.  This trend is reflective of the relative disparity between the regions; solutions and beans, concentrations, with the rate being greatest where the concentration parity is least.  All compounds assessed by these authors, except for fructose [13], demonstrated this behaviour.  A similar concentration gradient may not be present when fermentation is carried out in parallel with diffusion, e.g. fermentation tanks, and the microbial metabolites may be more gradually formed.  While these aroma and flavour precursors distinguish themselves from one another by their rate of diffusion, K1 in the table below, there is no clear pattern between the compounds’ chemical properties and their diffusion rates into the bean.  It wouldn’t be unexpected if these rates varied among different green bean varieties and growth conditions, making further research in this area recommended.  

What we do know is that equilibrium conditions were reached after approximately 48 hrs for all compounds within this study [14].  This result meant that approximately 50% of the molecules in the surrounding solution diffused into the bean once equal concentrations inside and out have been established, indicating a simple diffusion process.  In practical terms, considering the weight limit and defining the useful content of the tank as discussed earlier, in the best-case scenario where no additional layers of the coffee cherry are present, only half [15] of this content would diffuse into the bean in the fermentation tank.

Extending the diffusion trials to include the retention of the extra layers of the coffee cherry, such as parchment and mucilage, demonstrated that some compounds, like butanal and isoamyl acetate, continued to diffuse into the bean unaffected.  However, other compounds; 2-phenyl ethanol, alanine and glutamic acid, were affected by these additional layers.  These experiments also demonstrated that the compounds were mainly obstructed by the parchment layer, emphasizing its significance as a diffusion barrier towards microbial metabolites entering the bean.  The parchment resistance, represented by p, has a substantial impact on the diffusion of the compound into the bean, reducing the transfer efficiency by approximately 30 to 60%.  This lower loading capacity is naturally complimented by a lower rate of diffusion, K2 in the table below.  

Hadj Salem and colleagues (2022) noted in their study that the resistance of the parchment can increase as the diffusion progresses.  In simpler terms, the diffusion of the desirable aroma and flavour precursors becomes less efficient as time passes.  This directly affects the duration for which coffee fermentation, in its current form, remains practically feasible.  The source of this rise in diffusion resistance has yet to be discovered. The author speculates that it may come from:

• • Swelling of the parchment reducing the size of pores through which molecules diffuse.

  • Increasing the acidity of the mucilage layer would enhance pectin's ability to bind water, leading to tissue swelling. This increased interaction between water and pectin would reduce the availability of water, i.e. the mobile phase, slowing the diffusion of aroma and flavour precursors into the bean.

  • Larger molecules that are unable to transfer through the parchment may clog it instead, much like how microfilters can be clogged by larger molecules.  

  • In the presence of microbes biofilms may also form, which could also impede the transfer of microbial metabolites by increasing parchment resistance.  

Due to the practical implication of this resistance further research to determine the origins of this resistance is encouraged. 

Given that the compounds are not uniformly impacted by the presence of the coffee cherry, especially the parchment layer, it could be argued that an inherent bias in diffusion is introduced [16].  This has the potential to impact the aroma and flavour profile developed during coffee fermentation.  For example, diffusion of isoamyl acetate, which contributes to the banana-fruity aroma and flavour in certain fermented coffees, produced from specific yeast strains, remains unaffected by the coffee fruit.  This may result in a perceived prominence of these aroma notes when the coffee is fermented in the presence of its parchment layer.  On the other hand, compounds, e.g. 2-phenyl ethanol with a rose like floral aroma, where diffusion is reduced [17] would require significantly greater concentrations to elevate the green coffee to a similar quality standard [18].  This suppression may also be useful, once understood, especially if there are undesirable compounds being controlled.

Exploring the impact of diffusion in coffee fermentation uncovers further process inefficiencies beyond those identified in Part 1 of this article series, which resulted in the recommendation that separating the fermentation process from that of the bean could be advantageous.  This recommendation is further extended to the diffusion process, as its efficiency is dismal and could be much improved if treated as a separate process.  By separating these processes, several benefits can be realized:

• • • Optimization of the fermentation process through microbial selection and fermentation conditions, leading to high-value flavours and aromas that can be utilized separately.

    • Optimization by selective filtration of desired metabolites.

    • Removal of the parchment layer to enhance infusion efficiency.

    • Infusion of green beans with a targeted aroma/flavour profile, eliminating guesswork.

    • Optimization of the infusion process for time efficiency.

    • Reduction of process waste to optimize revenue streams and increase farmers' income.

While these steps require research, investment, and support for coffee origins through education programs, the perishable nature of coffee cherries suggests that fermentation should remain an origin operation.