Graphene oxide beneath the monolayer lifts HZB's all-perovskite triple junction to 27.3%

The 12.0% monolayer-only cell functions as the anomaly under investigation rather than a baseline for a working contact ^[12], because the same chemistry behaves well on pure lead perovskites and misbehaves only where it meets tin-lead ^[7]. The bilayer's margin over it, 10.1 percentage points or a factor of 1.84 ^[19], is therefore a measure of how completely a bare monolayer fails on this material. It confirms the diagnosis rather than ranking the bilayer against the contact the field has actually been using ^[6].

Two changes arrive together when the graphene oxide goes down first. Chemically, the phosphonic acid head groups of MeO-2PACz bond to the oxygen groups on the sheet, and the resulting molecular reorientation deepens the work function and raises conductivity ^[9][10]. Physically, the hydrophilic surface acts as a nucleation template, so the perovskite crystallises evenly and the films come out without the nanovoids seen in the monolayer-only devices ^[11]. HZB's optoelectronic characterisation attributes the improvement almost entirely to reduced electronic and ionic extraction losses ^[13], which is consistent with either route, since a void-riddled film also extracts badly. The account reports no graphene-oxide-only device that would separate the two ^[17].

In the stack, the gain is optical as much as electronic. Gold and PEDOT:PSS came out of the interconnecting layers and indium tin oxide plus the bilayer went in, so fewer near-infrared photons are absorbed before they reach the 1.25 eV bottom sub-cell ^[14][2]. The finished triple junction sits 5.2 points above the bilayer's own single-junction tin-lead device ^[20]. That subtraction is not a clean attribution (the two are different devices with different band-gap budgets), but it is the scale of what the upper junctions bought in this build, and passing the single-junction ceiling is the only reason to stack at all ^[21].

The durability half of the claim is thinner. Physics World reports the strategy improving operational stability without a retention figure or a test protocol ^[15], and gives no device count, cell area or outside calibration behind the 27.3% ^[18]. That gap matters more here than in most efficiency reports, because the case against PEDOT:PSS was never only its parasitic absorption; it was acid and water working on the perovskite underneath ^[6]. A contact that improves extraction while the layer below it still degrades solves half the problem. The portable part today is the measurement recipe: fast hysteresis plus bias-assisted charge extraction will tell another group whether ions at the buried interface are what is costing it charge collection ^[8].