Swedish Startup Angry Camel Redefining the Chickpea Protein Production

ALTERNATIVE PROTEINS

Harleen Singh

6/24/20266 min read

This is an AI generated illustration and does not represent actual product or service.

When most people think about plant protein innovation, they picture novel extraction chemistry or genetically tuned crops. Angry Camel AB's newly published international patent application (WO 2026/111638), filed out of Malmö, Sweden, tells a different story. The headline technology—supercritical CO2 (SC-CO2) extraction—isn't new. What's new is getting chickpeas to behave well enough inside that process to make it worthwhile. And that turns out to be mostly a milling problem.

For an industry hunting for credible egg white and lecithin alternatives, that's a more interesting story than it sounds.

The egg replacement problem hasn't gone away

Egg white ovalbumin and egg yolk lecithin are workhorses of food formulation—emulsifying mayonnaise and dressings, stabilizing foams in baked goods, binding plant-based dairy alternatives. They're also exposed to avian flu-driven price volatility, allergen labeling requirements, and a growing segment of consumers actively avoiding animal-derived ingredients.

Chickpea protein has long been pitched as a candidate replacement, largely because of its natural emulsifying and foaming behavior. But as the patent's background section lays out in some detail, chickpeas have been stubbornly resistant to the processing methods that work fine for soy, pea, and other legumes. The application frames this as essentially three compounding problems: starch interference, extreme hardness, and oil-soluble bitter compounds that conventional debittering can't touch without wrecking protein functionality.

Why chickpeas broke the standard SC-CO2 playbook

SC-CO2 extraction is already established for de-oiling soybeans, rapeseed, and sunflower. It's attractive because it avoids hexane and ethanol, runs at relatively gentle temperatures, and can selectively pull out lipophilic compounds—including the oil-soluble saponins and phenolics responsible for much of chickpea's "beany" bitterness.

The catch is feedstock geometry. Conventional SFE wants flaked material—roughly 2 mm flakes, under 0.5 mm thick—to give CO2 enough surface area to penetrate without channeling or clogging the extraction bed.

Chickpeas don't flake well. According to the application, their seed hardness exceeds 200 Newtons—more than double soybean's—so conventional flaking either produces brittle flakes that collapse into flour, or requires aggressive milling that generates heat, denatures protein, and creates fines that clog the CO2 bed. Adding moisture to soften the seed before milling introduces its own risks: microbial spoilage and premature starch gelatinization, both of which create downstream headaches.

This is the gap the patent is trying to close—not by reinventing SC-CO2, but by reinventing the feedstock that goes into it.

The fix: milling for morphology, not just particle size

The application's core process claim specifies dehulling chickpeas, then milling them using an "abrasive" method—explicitly excluding high-friction techniques like hammer milling or pin milling, which the inventors say generate too many fines. Roller milling, multi-stage roller milling, and blade milling are given as examples.

The target isn't just a particle size range (250–2000 µm, with a narrower preferred window of 250–700 µm), but a specific morphology: "grit-like or semolina-like"—free-flowing, granular, resistant to compaction. The patent is fairly explicit about why this matters operationally. Particles over 2000 µm hinder CO2 penetration and leave oil and bitter compounds behind. Particles under 250 µm—essentially flour—cause clogging, pressure drops, and starch swelling that hurts protein recovery later in the process.

In other words, the innovation is tuning the mechanical preparation step specifically to chickpea's hardness and starch profile so that an existing extraction technology can actually work on it. That's a less glamorous story than "novel CO2 chemistry," but for anyone who has tried to run unconventional feedstocks through SFE pilot equipment, it's the kind of detail that decides whether a process is commercially viable or stuck in the lab.

What the SC-CO2 step actually buys you

The patent's data—drawn from its own examples, so worth treating as evidence from the applicant rather than independently verified—makes a fairly striking comparison between SC-CO2 de-oiling and ethanol extraction.

On solubility, the disclosure reports a Nitrogen Solubility Index (NSI) of 76.9% for chickpea protein de-oiled with SC-CO2, versus 51.0% using ethanol, with the raw material baseline at 84.6%. NSI is essentially a proxy for how well a protein will perform in applications that need it to dissolve or disperse—emulsions, beverages, batters. A jump from roughly 51% to 77% is the difference between a protein that's marginal for emulsification work and one that's genuinely usable.

On flavor, the contrast is even sharper. The application reports total identified volatiles of 9.58% (GC-FID area) for the SC-CO2 product versus 60.98% for ethanol-extracted material, with a hexanal-related marker compound (terpinen-4-ol) at 0.616% versus 19.19%. Hexanal and related aldehydes are classic lipid-oxidation off-flavor compounds, and they're a big part of why legume proteins taste "green" or "beany." If those numbers hold up in independent testing, it's a meaningful palatability story—and notably, the patent claims this neutral taste profile is achieved without any separate debittering step, which traditionally relies on acid-base treatments that damage protein structure.

That's the commercial pitch in a nutshell: skip the debittering step, get better solubility and a cleaner flavor, all from one process change.

Two products, two purification paths

After SC-CO2 de-oiling, the disclosure describes forming an aqueous slurry (pH adjusted to either an alkaline range of 8.0–9.5 or a near-neutral 6.0–7.5), clarifying it to remove starch and fiber, and then purifying the protein extract via one of two routes: ultrafiltration with optional diafiltration, or isoelectric precipitation.

These aren't interchangeable in outcome. The UF route, per the examples, produced a liquid concentrate with protein solubility around 90% and an NSI near 77%—useful as a liquid ingredient for beverages or dairy alternatives, or spray-dried into a powder. The isoelectric precipitation route achieved higher protein purity (around 89% dry matter) but markedly lower solubility (NSI of roughly 54%), which the patent attributes partly to the thermal pasteurization step applied in that example and partly to the IEP method itself.

What to watch for, and what's still unproven

A few things are worth flagging before treating this as a settled win.

First, the patent itself acknowledges residual oil isn't eliminated—just reduced to 2.0 wt% or below (with a preferred target of 1.0% or less). That's a big improvement over solvent-based de-oiling, but it's not zero, and residual lipid is exactly what drives oxidative off-flavors over shelf life. How this product performs after months on a shelf, rather than in a fresh pilot batch, is an open question the application doesn't address.

Second, several of the most flattering comparisons—particularly "superior functionality compared to commercial egg white powder in both emulsification and foaming properties"—come from the applicant's own examples, including some explicitly labeled "prophetic" (meaning predicted rather than experimentally demonstrated, a common feature of patent applications covering optional process steps like microbial pasteurization). That's standard patent drafting practice, not a red flag in itself, but it's a different evidentiary bar than peer-reviewed sensory panel data.

Third, there's a real economics question lurking behind the elegant chemistry. SC-CO2 extraction at 200–400 bar is capital-intensive, and the patent's bed-loading figures (0.25–0.60 kg of milled chickpea per liter of extractor volume) suggest a process that, while scalable, will need real throughput to compete on cost with conventional solvent extraction—especially since the broader market for SC-CO2 oilseed processing (used in things like premium edible oils and decaffeination) has historically served lower-volume, higher-margin applications than commodity protein ingredients.

The bigger picture

What makes this application interesting isn't a single breakthrough chemistry—it's a worked example of how feedstock-specific mechanical preparation can unlock an extraction technology that's been sitting on the shelf for decades. If the solubility and flavor data hold up at scale, chickpea protein gets a credible shot at applications where soy and pea proteins have struggled on taste. And for anyone evaluating ingredient suppliers in this space, the lesson generalizes: when assessing a "novel extraction" claim, ask what happened to the feedstock before it ever reached the extractor. That's often where the real engineering—and the real cost—lives.

For Angry Camel, the next test isn't the patent office. It's whether ChickBumin® can deliver these lab-scale numbers consistently in industrial runs, at a cost that food manufacturers will pay for an egg replacement that's allergen-free and clean-label by design.