Semi-washed

Where it is practiced

The word circulates in every market. It does not designate the same operation depending on the country that uses it.

In Brazil, it stands for pulped natural (cereja descascado, in Portuguese): the cherry is pulped, then dried with its mucilage, the layer of sweet flesh stuck to the parchment. The idea appears in the Brazilian literature in the early 1950s. It then remains without follow-up for nearly forty years. It is taken up again, independently, in the municipalities of Botelhos and Ouro Fino, in Minas Gerais, in 1989 and 1990. It then spreads to the other producing states (Silva et al., 2004).

In Indonesia, it stands for wet hulled (giling basah, in Indonesian), wet hulling. The reference trial on this route covers lots from northern Sumatra, in the districts of North Tapanuli and Toba Samosir. The trade concerned covers North Sumatra and Aceh, through the port of Medan. The huller used is a screw of Dutch colonial design, made by Kemajuan Industries in Malang, on Java. The trial was run during the 2004-2005 season for the FAO/CFC/ICO project.

In Central America, the word most often covers honey, coffee pulped and dried with all or part of its mucilage. The colors used are sales conventions, not standards:

• white,
• yellow,
• red,
• black.

Each announces a proportion of mucilage left on and a drying regime. No body sets the thresholds. Two producers therefore do not call the same thing “red honey”, and the buyer cannot deduce the handling from the name.

In Colombia, mechanical demucilaging, that is, tearing off the mucilage by friction, has existed since Cenicafé's BECOLSUB technology, published in April 1997. The coffee that comes out is not called semi-washed: Cenicafé calls it café lavado. Same machine, different name. ECOMILL, published in June 2013, does not extend this gesture. It is a mechanical washer meant for coffee whose mucilage has first been broken down in a tank, by natural fermentation or by enzymes.

In Vietnam, the expression chế biến bán ướt (half-wet processing) serves as a commercial synonym for honey on specialty canephora. This usage is documented only by Vietnamese professional sources, not by published work.

The principle

Semi-washed is not a process. It is a box, and this box has no official definition. The glossary of the International Coffee Organization, reproduced by the FAO in December 2000, knows only two states. Washed coffee is there green coffee prepared by the wet process. Unwashed coffee is obtained by the dry process, and the text specifies that it is also called natural. There is no entry for semi-washed, for pulped natural, or for wet hulled. The same glossary distinguishes only four groups of coffee in price quotations, and none corresponds to semi-washed. A contract that carries this word alone therefore commits to no verifiable parameter.

What practitioners file under this word nonetheless comes down to a single physical variable: the amount of mucilage still present on the parchment during drying, and the moment the parchment is removed. Mucilage is a pectin gel loaded with sugars. It makes up 14.85% of the fresh weight of the fruit in the Colombia variety, against 43.58% for the pulp (Cenicafé, 2013). This mass matters, because it is the substrate that microorganisms will work on.

As long as it stays in place, it is alive. Yeasts break down the pectins there by means of pectinases, enzymes that cut the pectin chains and release galacturonic acid. It is this cutting that detaches the gel from the parchment. Bacteria ferment the sugars of the gel in parallel and produce ethanol, lactic acid and acetic acid. The medium acidifies: Cenicafé measures pH values of 4.00 to 5.81 in the water in contact with demucilaged coffee kept twelve to fourteen hours. Drying stops this activity by lowering the available water, failing which it continues until it becomes a defect.

Hence the real content of the word: a semi-washed is a coffee in which this work was stopped earlier than in a natural and later than in a washed, or else it was removed at one stroke by mechanical abrasion. Two opposite gestures, a single name.

The sequence

The shared first step of the three families:

1. Harvest of ripe cherries, then pulping. The pulper removes the skin and pulp. The mucilage, the inner layer stuck to the parchment, stays in place. It is what decides what comes next.

Then three branches that carry the same name.

Brazilian branch, pulped natural:

2. No fermentation in a tank, no washing. The coffee goes to drying with its mucilage. The sugars of the gel therefore stay in contact with the bean throughout drying. 3. Pre-drying on the patio (terreiro, in Portuguese), the farm's drying area, for 1 to 2 days. It drives off the free surface water and makes the lot manageable before the dryer. 4. Finishing in a mechanical dryer. On real lots from southern Minas Gerais, the measured final moisture ranges from 9.32 to 11.69% on a wet basis. Most of these lots fall below the rate recommended by the Brazilian Ministry of Agriculture (Silva et al., 2004). Drying below this rate loses sellable weight. 5. Storage in parchment, dry hulling at the time of shipment.

Indonesian branch, wet hulled:

2. Fermentation the night after pulping, then washing the next morning. Some producers ferment under water, others dry. The FAO study notes that no published work establishes the effect of this choice on the cup. 3. Half a day of skin drying, then collection the same afternoon. The parchment travels moist through the supply chain. 4. Hulling of the still-moist parchment, at between 35 and 45% moisture. This is the gesture that defines the process: the bean loses its parchment while it is still soft and swollen with water. 5. The machine is adapted for this, because soft parchment does not split like dry parchment:

• the steel hulling rib is replaced by a rubber band,
• the motor goes from 15 to 25 horsepower,
• the rotation goes from 600 to 750 revolutions per minute,
• the throughput drops to about 800 kg of wet bean per hour, against a nominal tonne in dry parchment.

6. Drying of the bare bean in the sun to 12%. In the FAO trial, it is the rapid hulling followed by direct drying to 12% that gives the best chance of obtaining the Mandheling character.

Mechanical demucilaging branch:

2. The mucilage is torn off by friction in a demucilager. Cenicafé's DESLIM removes more than 98% of it. A rod-rotor machine tops out around 80%, which forces a return through a tank to finish the job. 3. The coffee goes directly to the dryer. No fermentation time is allowed, since the gel is no longer there to feed the microorganisms. 4. Water consumption of 0.7 to 1.0 L per kg of dry parchment coffee with BECOLSUB. A classic natural fermentation uses 4.17 to 20 L. It is this saving in water, and the organic load of the effluents, that motivated the machine. 5. For comparison, the Colombian reference practice lets pulped coffee ferment 14 to 20 h in a tank before washing. 6. ECOMILL does not belong to this branch. It mechanically washes a coffee already fermented in a tank, with 0.3 to 0.5 L of water per kg of dry parchment coffee. Cenicafé designed it for producers and buyers who require a natural fermentation.

What it gives in the cup

What is measured.

On wet hulled, a trial run for the FAO/CFC/ICO project in North Sumatra produced 42 samples of 300 g of green coffee. They come from 6 origin lots and 7 process regimes. They were cupped blind by four coffee houses well versed in North Sumatran coffees. Results:

• the Mandheling character exists, one sample having been named by 3 cuppers out of 4 as the best;
• this character could not be tied to a specific treatment;
• lots dried in parchment, without wet hulling, were judged cleaner and more acidic;
• rapid wet hulling followed by direct drying to 12% gave the best chance of obtaining the sought-after character.

The cuppers also noted that the whole set of samples was too clean and not complex enough for an ordinary Mandheling. A single sample was ranked first by one cupper and last by another. The definition of the character therefore varies from one house to another, which rules out making it a contract criterion.

On mechanical demucilaging, Cenicafé had its panel cup three after-machine regimes. Immediate drying gives 100% of cups scored above 5 on its overall impression scale. Coffee kept fourteen hours without water falls to 95%, coffee kept under water to 75%. No cup shows the fermented defect or the dirty defect (fermento and sucio, in Spanish). This trial therefore compares ways of handling demucilaged coffee, not demucilaging against natural fermentation.

What is claimed without an accessible trial.

Commercial discourse credits semi-washed with a natural's sweetness with the cleanness of a washed, a body superior to the washed and a rounder acidity. I found no published, readable blind comparison that establishes this effect for all the processes filed under this word. The effect is plausible, because the mucilage left in place holds sugars and organic acids in contact with the bean. It is not demonstrated at this level of generality.

What goes wrong

The mucilage left on the parchment is a culture medium. Dried too slowly, in too thick a layer, or taken up again by rain, it molds. The defect that follows is called moldy, or musty. It cannot be corrected in roasting, because it is already set in the green bean.

On the Indonesian route, the FAO trial is explicit. Fungal infection increases with the storage time of the moist parchment, and the cuppings confirm it with a higher frequency of moldy notes. Aspergilli of the ochraceus group, which include Aspergillus ochraceus, a producer of ochratoxin A, were found in 5 samples of moist parchment out of 6 after two weeks of storage. At shorter durations, they were not detected. Ochratoxin A was measured in 7 samples out of 42, from 0.07 to 2.24 ppb. Four of these seven came from parchment stored moist for 14 days. Prolonged storage also produces sour notes, the sour defect.

Second risk specific to wet hulling: the bean is hulled soft. The machine must be cushioned for this, its steel rib being replaced by rubber. The bean comes out bare and of a deep green, a color often sold as a signature of origin. The cuppers of the FAO trial, for their part, judged it too green.

Third risk, common to the whole family: fermentation that overflows. The fermented defect and the phenolic taste come from a mucilage left active too long, or from a wet lot that starts up again. In the FAO trial, all four cuppers noted the fermented character on the same sample. The word stinker, in the vocabulary of the trade, designates the single bean that is enough to spoil a cup.

Fourth risk, contractual and not sensory: the word itself. Buying semi-washed without specifying the actual handling amounts to specifying nothing. A Brazilian semi-washed and a Sumatran semi-washed do not share a single parameter.

Benchmarks

• Mucilage in the Colombia variety14.85% of the fresh weight of the fruit
• Pulp in the Colombia variety43.58% of the fresh weight of the fruit
• Moisture of the parchment at wet hulling35 to 45%
• Reference Colombian fermentation before washing14 to 20 h
• Water, natural fermentation4.17 to 20 L per kg of dry parchment coffee
• Water, BECOLSUB mechanical demucilaging0.7 to 1.0 L per kg
• Water, ECOMILL mechanical washing after fermentation0.3 to 0.5 L per kg
• Mucilage removed by the DESLIM demucilagermore than 98%
• Final moisture measured on pulped natural lots from southern Minas Gerais9.32 to 11.69%
• Ochratoxin A, FAO trial in Sumatra7 positive samples out of 42, from 0.07 to 2.24 ppb
• Quotation groups of the International Coffee Organization4, none for semi-washed