Science1 publisher2 min readPublished
A graphene oxide membrane takes half the caffeine out of coffee already brewed
Yihan Tian's group at UNSW tuned the channels between stacked graphene oxide sheets using alginate and calcium ions, then ran finished coffee through, removing about half the caffeine while four other compounds stayed detectable.
The Scientist · Science desk
What happened
- A graphene oxide membrane built at UNSW and reported in the Journal of Membrane Science removed around half the caffeine from coffee, in its best-performing version.
- Conventional decaffeination treats green beans before roasting, so Tian and Ren worked from the other end and filtered coffee that had already been made.
- Instead of picking whichever version stripped the most caffeine, the team scored several membranes together on caffeine removal, retention of selected compounds and water permeance.
- The group has also put the membrane material into hollow fibers, an early step toward testing the approach at a larger scale.
Compiled by The ScientistSomething wrong?How this is made
Why it matters
- capability Filtering the finished drink moves the caffeine decision downstream of roasting, so a single batch of beans could in principle serve both kinds of drinker.
- constraint At the reported performance the membrane is no use to anyone who needs caffeine near zero, and the authors themselves put a practical process further down the road.
- decision The next real go or no-go rests with a sensory panel, because a chromatogram showing four molecules present cannot settle whether the coffee still tastes like coffee.
- precedent Since the half-removal figure came at a filtration time chosen for fair comparison, a bigger number in the next paper should be read as tuning of the same chemistry.
Graphene oxide stacks in layers, and the gaps between those layers are where the separation happens. Water and other dissolved molecules travel through the channels, so anything that changes the channels changes what gets through. Tian, working with Tongxi Lin and Xiaojun Ren, added sodium alginate and crosslinked it with calcium ions to alter those channels and shift which molecules passed [4][5][2].
The retention side of the result is reported as presence. Trigonelline, N-methylpyridinium, choline and hydroxymethylfurfural were still detectable in the coffee after it went through the membrane [9]. The paper did not report a retained percentage for any of the four [18].
That distinction matters because the design problem here is a trade-off the group hit early: making a membrane better at one thing made it worse at another, and stripping more caffeine is of little use if too much of what you wanted to keep goes with it [10].
About half the caffeine came out [8]. At 50 percent removal, two cups of the treated coffee carry roughly the caffeine of one untreated cup [17]. For someone trimming a habit that is worth something. For anyone instructed to avoid caffeine it is not decaf, and the account says further work is needed before this becomes a practical decaffeination process [19].
The half figure also comes from one filtration condition. Time was held constant across the membranes so the versions could be compared fairly, and the researchers say the result does not necessarily represent the limit of what the membrane can do [12]. They expect optimization of the membrane and the filtration conditions to raise caffeine removal while keeping processing time practical [13].
Tian, a master's research student on UNSW's Team Graphene, said: "What attracted me most is that this research connects fundamental membrane science with a very practical, everyday product" [16][2].
The team plans to explore how filtration affects coffee's taste and nutritional properties, with sensory testing guiding the next stage of development [14]. Four named molecules surviving a pass through a membrane is a chemistry result, and coffee's character comes from a mixture of many compounds that contribute to flavor and aroma [6]. Whether the filtered cup still tastes like coffee to a drinker is a separate experiment, and it is the one that decides whether the membrane matters outside the ARC Centre of Excellence for Carbon Science and Innovation [1].
What to watch
- Sensory panel results: whether tasters can tell the filtered coffee from the untreated cup.
- A caffeine removal figure from an optimized membrane or a longer filtration time, and what it costs in water permeance.
- Whether the hollow-fiber version reproduces the flat membrane's separation when scaled up.