Perovskite–silicon tandems have a deposition problem that has survived every efficiency record the field has set. The silicon bottom cell is textured — etched with microscopic pyramids that trap light — and the perovskite top cell has to coat those pyramids evenly, at a thickness under a micrometre, without pinholes and without wasting material. The routes that do this reliably, evaporation and spin-coating, are either vacuum-hungry or wasteful, and neither scales the way a printing press does. A paper published in Advanced Energy Materials in September, by a team at the Karlsruhe Institute of Technology (KIT) led by Raphael Pesch and Ulrich W. Paetzold, reports a different answer: print the absorber with an inkjet head. The champion tandem reaches 30.2% efficiency, and the team describes it as the first perovskite–silicon tandem above 30% with an inkjet-printed absorber. This article treats it as analysis — a September paper surfacing in trade coverage on November 2 — not as breaking news.

Why inkjet, and why it has been hard

Inkjet printing is the deposition method a factory manager would choose if the physics allowed it. It is additive: droplets land only where they are needed, so material waste is close to zero. It is maskless: the pattern is software, not a stencil or a shadow mask. And it is digital in the same sense a printer is digital, which is why it is the only perovskite deposition route with a genuine print-industry supply chain behind it. Ricoh, which builds those printheads, announced on October 27 that it will establish a production site for inkjet-printed perovskite solar cells at its Tohoku Site in Miyagi Prefecture, targeting several hundred megawatts of annual capacity by 2030 with expansion to gigawatt scale in view, backed by Japan’s GX Supply Chain Development Support Program. The industrial interest is real. The laboratory record for inkjet-printed tandems had not kept up with it.

The reason is the texture. An inkjet droplet is a liquid object, and a textured silicon surface is a landscape of peaks and valleys a few micrometres tall. A droplet that lands on a pyramid tip behaves differently from one that lands in a valley, and a film that dries from discrete droplets is prone to the coffee-ring patterns and drying fronts that are the failure mode of printed perovskites. The KIT team’s answer is not to print the whole absorber in one step. It is to split the absorber’s formation into two steps, and to print only the second one.

The process: evaporate the skeleton, print the rest

The fabrication route is a hybrid of three deposition methods, and each one does the job it is best at. The hole-transport layer — a self-assembled monolayer (SAM) — is dip-coated onto the textured silicon. For single-junction cells the SAM is 2PACz; for the tandem cells it is Ph4PACz. Dip-coating coats a textured surface conformally: the substrate is immersed and withdrawn, so the SAM assembles in the valleys as well as on the peaks. The same step is used again for the passivation layer above the absorber.

The absorber itself is formed in two steps. First, the inorganic precursors — the lead-halide skeleton of the perovskite — are deposited by thermal evaporation, a vacuum process that lays down a uniform, conformal scaffold over the texture. Then the organic precursors are inkjet-printed onto that scaffold, and the two react to form the perovskite. This is the hybrid two-step route the same KIT group introduced in 2024, when a Solar RRL paper showed it could reach 18.2% in single-junction cells, on par with spin-coated counterparts, with films free of macroscopic drying effects. The 2024 paper was the proof that the route worked; the new paper is the proof that it works on a textured tandem bottom cell at 30%.

Two details of the hybrid route deserve emphasis, because they are the reason it survives the texture where one-step printing struggles. First, the evaporated inorganic layer is already a continuous film before any liquid arrives, so the inkjet step never has to wet a bare textured surface — it only has to deliver the organic precursors into a scaffold that is already there. Second, the inkjet step deposits the organics as a precisely metered dose rather than a flood, which keeps the material yield near 100% and the solvent load low. A related KIT paper from the same group, published in Advanced Science in 2026, pushed the same idea further toward manufacturability by replacing toxic solvents with a green γ-valerolactone and dimethyl sulfoxide system for one-step inkjet-printed wide-bandgap cells above 17%, and by demonstrating a first tandem with a one-step inkjet-printed film above 28%. The September paper is the two-step branch of that programme, and it is the one that has now crossed 30%.

The device numbers, handled in order

The tandem is a monolithic two-terminal device: a wide-bandgap perovskite top cell on a textured silicon heterojunction (HJT) bottom cell. The champion reaches 30.2% power conversion efficiency, with an open-circuit voltage of 1.94 V, a short-circuit current density of 20.4 mA/cm², and a fill factor of 75.8%. It held above 30% through 10 minutes of maximum-power-point tracking.

Three qualifications travel with that headline, and they are all stated plainly in the coverage and the paper’s own framing. First, the fill factor of 75.8% is openly not yet optimized: the team attributes it to a tandem architecture that has not yet been tuned for the textured silicon surface, and it is the weakest of the three device parameters. For comparison, the Soochow–LONGi tandem this desk covered on November 10, which reached a certified stabilized 35.13%, paired its 2.007 V open-circuit voltage with a fill factor of 84.53%. The KIT voltage is close to the field’s best — 1.94 V against 2.007 V — but the fill factor is roughly nine points behind, and that gap is where most of the missing efficiency lives. Second, the 10 minutes of tracking above 30% is a stability check, not a durability test: it confirms the device does not collapse immediately under load, and nothing more. No longer-term stability data, damp-heat result, or outdoor exposure accompanies the coverage consulted. Third, no independent certification is claimed for the 30.2% figure. It is the team’s reported champion, and should be quoted as such, not as a certified record. The device area of the champion tandem cell is not stated in the trade coverage consulted, and is not claimed here — a gap worth naming, since a tandem figure without an area is a figure without a scale.

Read with those qualifications attached, 30.2% is still the number that matters, because of what it is first at rather than how high it is. The certified tandem record sits above 35%, and the Jülich rounded-pyramid tandem this desk covered two weeks ago reached 33.25% on a 1.014 cm² cell without any printing at all. The KIT result does not compete on the leaderboard. It competes on the process: it is the first time the perovskite absorber of a 30%-plus tandem has been deposited by inkjet printing, and the first time a printing route — the route with the print industry behind it — has produced a tandem efficiency in the range that makes a factory conversation serious.

The mini-module: 17.8% at 12.96 cm², and the scaling number

The paper’s second result is the one that tests whether the process survives the jump from a cell to a module. The team fabricated a single-junction perovskite mini-module with an active area of 12.96 cm², made of nine series-connected cell stripes of 1.44 cm² each, formed using laser scribing, with the absorber deposited the same way as the tandem’s. It reaches 17.8% efficiency.

The number to read alongside it is the scaling loss: compared with the median efficiency of the smaller cells, the module loses 0.53 percentage points of efficiency per decade of area scaling. That is a modest loss, and it is the more informative figure, because it is a rate rather than a point. But the honest reading cuts both ways. The mini-module is a single-junction device, not a tandem — the 30.2% tandem and the 17.8% module are different demonstrations, and no tandem module is reported. The module’s 17.8% is also well below the small-cell figures the same process produces, and the coverage attributes the remaining gap to laser-scribing precision and process control, which the team names as the next improvements needed before the laboratory process becomes an industrial one. A 12.96 cm² mini-module is a scaling signal, not a product.

Where it sits in the tandem arc

Place the KIT result beside the tandem stories this desk has covered in the last month, and the field’s division of labour sharpens. The Jülich paper of October 31 attacked the texture from the silicon side: shrink the pyramids, round their tips with a three-minute ozone etch, and the perovskite coats evenly — 33.25% on a 1.014 cm² cell, explicitly not a record. The Soochow–LONGi paper, covered November 10, attacked the buried interface from the chemistry side: a conjugation-controlled SAM that resolves the trade-off between anchoring and charge transport — a certified stabilized 35.13%. The KIT paper attacks the same stack from the deposition side: keep the texture and the SAM, and change how the absorber gets there. Three papers, three different layers of the same device. The Jülich result is higher and the LONGi result is certified; the KIT result is the one that can be printed.

The lineage inside KIT matters too, because it shows a programme rather than a single result. The 2024 Solar RRL paper established the hybrid two-step route at 18.2% in single junctions. The 2026 Advanced Science paper made the solvents green and the one-step variant tandem-capable above 28%. The September Advanced Energy Materials paper combines the two-step route with dip-coated SAM and passivation layers and takes it to 30.2% on a textured HJT bottom cell. Each step removed one obstacle to manufacturing — first the drying defects, then the toxic solvents, now the tandem integration.

The Ricoh announcement, six days before the trade coverage of the KIT paper, is the commercial echo of the same argument. Ricoh’s plan is to pattern-print every functional layer of its cells with its proprietary inkjet process, eliminating laser patterning and vacuum deposition, at a production site in Miyagi targeting several hundred megawatts a year by 2030. Ricoh’s process and KIT’s process are not the same — Ricoh prints all layers, KIT still evaporates the inorganic scaffold and dip-coats the SAM and passivation — and Ricoh published no efficiency, lifetime, or cost data in its announcement. But the two announcements land in the same month for the same reason: inkjet printing has moved from a laboratory deposition option to a manufacturing bet, and the laboratory evidence now includes a 30%-plus tandem.

What to watch

Three things, in order of how quickly they could change the assessment.

First, the fill factor. The team has named the unoptimized tandem architecture on textured silicon as the cause, and a fill factor in the low 80s — where the best tandems sit — would take the same voltage and current to roughly 33%. The next KIT tandem paper will be judged less on whether it prints than on whether the fill factor moved.

Second, the area. A stated device area for the champion tandem cell, and a tandem module of any size, would convert the 30.2% from a process demonstration into a scaling claim. The 12.96 cm² single-junction mini-module shows the printing route scales; a tandem module would show the tandem scales.

Third, durability. Ten minutes of maximum-power-point tracking is the shortest stability evidence this desk has covered for a tandem in months, and the paper’s own framing is about fabrication, not lifetime. A thousand hours under ISOS-L-1, let alone a damp-heat or outdoor result, would be the first evidence that a printed absorber ages like an evaporated one.

The bottom line

A KIT team has shown that the perovskite absorber of a silicon tandem can be inkjet-printed — evaporate the inorganic scaffold, print the organic precursors into it, dip-coat the SAM and the passivation — and the resulting device reaches 30.2%, with 1.94 V, 20.4 mA/cm², and a 75.8% fill factor, holding above 30% through 10 minutes of tracking. A 12.96 cm² single-junction mini-module made the same way reaches 17.8%, losing 0.53 percentage points per decade of area scaling against the smaller cells.

It is a September paper, not a breaking story, and its limits are its authors’ own: the fill factor is not yet optimized for the textured architecture, the tracking run is ten minutes, no certification is claimed, no tandem module is reported, and the champion cell’s area is not stated in the coverage consulted. Frame it as the first inkjet-printed tandem above 30% — a deposition result, not a record — and let the fill factor, the tandem module, and the durability data decide whether printing becomes the way tandems are made.

No investment recommendation is made or implied anywhere in this article.

Sources and further reading

  • Pesch, R., Rothbauer, L. P., Schenk, J., Fang, L., Kuechle, T., Petry, J., Dangudubiyyam, U. K., Thelen, R., Ritzer, D. B., Pan, T., Baumann, D. O., Gholipoor, M., Zhao, T., Rosemann, N. W., Li, J., Lemmer, U., & Paetzold, U. W. “Hybrid Two-Step Inkjet Printing and Dip-Coating Enable Scalable 30%-Efficient Perovskite/Silicon Tandem Solar Cells.” Advanced Energy Materials (September 2026), e71579. DOI: 10.1002/aenm.71579 — the primary source for the paper title, author list, and DOI, and, via the trade coverage below, the hybrid fabrication route (dip-coated SAM HTL, 2PACz for single-junction and Ph4PACz for tandems; evaporated inorganic precursors with inkjet-printed organic precursors; dip-coated passivation), the textured silicon HJT bottom cell, the champion tandem figures (30.2%, 1.94 V, 20.4 mA/cm², 75.8% fill factor), the 10-minute above-30% maximum-power-point tracking, the first-above-30%-with-inkjet-printed-absorber claim, the 12.96 cm² mini-module (nine 1.44 cm² laser-scribed stripes, 17.8%), the 0.53-percentage-point-per-decade scaling loss, and the unoptimized fill factor on textured silicon.
  • taiyangnews (November 2, 2026), “KIT Achieves 30.2% Efficiency In Inkjet-Printed Tandem Cells” — trade coverage of the same paper, the source for the November 2 announcement date, the process description, all device and module figures quoted above, and the authors’ stated next steps (laser-scribing precision and process control). No certification is claimed in the coverage, and no device area is stated for the champion tandem cell; neither is claimed here.
  • Pesch, R., Diercks, A., Petry, J., et al. “Hybrid Two-Step Inkjet-Printed Perovskite Solar Cells.” Solar RRL 8, 2400165 (2024). DOI: 10.1002/solr.202400165 — the 2024 origin of the hybrid two-step route: evaporated PbI₂ with inkjet-printed organic precursors, 18.2% champion single-junction cells on par with spin-coated counterparts, films free of macroscopic drying effects, and the group’s stated goal of processing full-size wafer silicon bottom cells.
  • Dangudubiyyam, U. K., Pesch, R., Karakaya, O., et al. “Eco-Friendly Solvent System for Inkjet Deposition of Wide Bandgap Perovskite Solar Cells Enabling Tandem Integration.” Advanced Science 13 (54), e76463 (2026). DOI: 10.1002/advs.76463 — the green-solvent branch of the same programme: γ-valerolactone and dimethyl sulfoxide solvent system, wide-bandgap inkjet-printed single-junction cells above 17%, and a first tandem with a one-step inkjet-printed film above 28%.
  • Ricoh Company, Ltd. (October 27, 2026), “Ricoh to establish manufacturing capabilities to advance mass production of inkjet-printed perovskite solar cells” — the company announcement, the source for the Tohoku Site production site in Shibata, Miyagi Prefecture, the several-hundred-megawatt annual capacity target by 2030 with gigawatt-scale expansion in view, the METI GX Supply Chain Development Support Program selection, and the September 2025 NEDO Green Innovation Fund project. No efficiency, lifetime, or cost data were published in the announcement, and none is claimed here.
  • Excitonic.com, “Rounded Pyramids Let Perovskite-Silicon Tandems Keep Their Light Trap” (November 2026) — the paired tandem story: Jülich submicron rounded pyramids, 33.25% on a 1.014 cm² cell, explicitly not a record; and “Twist the Molecule, Not the Process: Conjugation Control Solves the Tandem SAM Trade-Off” (November 10, 2026) — the Soochow–LONGi SAM, certified stabilized 35.13% on a 1 cm² cell, the fill-factor comparison used above.

A note on sourcing: the full body text of the Pesch paper was not directly readable through the browser fetch used for this article, so device, module, and process figures are reported from the taiyangnews coverage of the paper, with the paper’s title, author list, and DOI confirmed against the KIT Paetzold Lab publications listing. The champion tandem cell’s device area is not stated in the material consulted and is not claimed here. No certification, long-term stability, damp-heat, outdoor-exposure, cost, or yield data are stated, because none appears in the material consulted. Background on inkjet printing and tandem architectures is drawn from the standard perovskite literature and the cited KIT papers.