Passive grain boundaries (GBs) are essential for polycrystalline solar cells to reach high efficiency. However, the GBs in Cu2ZnSn(S,Se)4 have less favorable defect chemistry compared to CuInGaSe2. Here, using scanning probe microscopy we show that lithium doping of Cu2ZnSn(S,Se)4 changes the polarity of the electric field at the GB such that minority carrier electrons are repelled from the GB. Solar cells with lithium-doping show improved performance and yield a new efficiency record of 11.8% for hydrazine-free solution-processed Cu2ZnSn(S,Se)4. We propose that lithium competes for copper vacancies (forming benign isoelectronic LiCu defects) decreasing the concentration of ZnCu donors and competes for zinc vacancies (forming a LiZn acceptor that is likely shallower than CuZn). Both phenomena may explain the order of magnitude increase in conductivity. Further, the effects of lithium doping reported here establish that extrinsic species are able to alter the nanoscale electric fields near the GBs in Cu2ZnSn(S,Se)4. This will be essential for this low-cost Earth abundant element semiconductor to achieve efficiencies that compete with CuInGaSe2 and CdTe.

Thin-film photovoltaic (PV) devices can be fabricated using a solution-based synthesis procedure in which metal-chalcogenide nanocrystals with aliphatic coordinating ligands are suspended in a solvent to produce a printable ink (NC-ink). However, the aliphatic ligands that are used to solubilize and stabilize the nanocrystals operate as a significant source of carbon impurities that are incorporated into the final device absorber layer. Despite the ubiquity of this technique and the fact that carbon defects have been reported to be found across a spectrum of devices, the structure, properties, and influence of the carbon on PV device performance remain relatively unexplored. Our findings indicate that these organic ligands undergo a pyrolysis reaction during annealing, producing an electrically conductive, graphitic carbon that also reacts with chalcogens (S or Se) to produce heterocyclic moieties. In this work, we used oleylamine (OLA) and dodecylamine (DDA) to fabricate Cu 2 ZnSn-(S x Se 1−x ) 4 (CZTSSe) photovoltaic devices from a NC-ink. DDA, which has fewer carbon atoms and contains no double bond, produces CZTSSe devices with less carbon in the absorber layer; but this reduced carbon content does not translate to improved device performance. OLA, which is a larger molecule and contains one double bond, produces CZTSSe devices with more carbon in the absorber layer. However, OLA also produces more crystalline graphitic carbon and allows for the CZTS nanocrystals to grow to a larger size during annealing, which improves device performance significantly.

3464www.MaterialsViews.com wileyonlinelibrary.com result in a linear dispersion for low energy carriers, which can be described as zero rest-mass relativistic particles with exemplary transport properties. [1][2][3] Electron mobility as high as 200 000 cm 2 V −1 s −1 can be achieved once extrinsic factors such as scattering centers from underlying substrates, adsorbates, and defects within graphene itself are controlled. [4][5][6][7][8][9] In order to utilize graphene for next generation electronic devices it is necessary to have precise control of carrier concentration and polarity without disrupting its intrinsic properties. Carrier doping of graphene has so far been achieved via chemical doping, [10][11][12][13] substitutional doping, [14][15][16] electric fi eld modulation, [ 17,18 ] and metal contact doping. [ 19,20 ] Recent reports have also shown it is possible to modulate the properties of graphene by modifying the underlying dielectric surface with a self-assembled monolayer (SAM) resulting in doping control without compromising the intrinsic graphene performance. [21][22][23][24] However, these studies use exfoliated graphene rather than the more commercially viable CVD-based graphene. In addition, exfoliation severely limits the ability to do a systematic and statistical study based on multiple devices due to the taxing processing steps necessary to fabricate individual devices. In order to circumvent this issue, characterization of these devices is mainly attributed to multiple Raman scans on a few pieces of modifi ed graphene rather than multiple pieces of graphene. While such methodology is useful for obtaining an overall picture of the graphene doping environment it makes it diffi cult to understand the exact relationship between SAMs and graphene. SAMs are commonly used in organic fi eld-effect transistors to modify the work function of metal electrodes, [ 25,26 ] quench charge trap sites at the interface between semiconductor and metal or dielectric, [27][28][29] and modulate the position of the threshold voltage. [30][31][32] SAMs represent an ideal platform for control of graphene electronics as they can be designed and functionalized at the molecular scale to cater to specifi c device requirements. However, there is still a need for better understanding of how SAM-treated dielectrics modulate the doping of graphene devices. In particular, an understanding of how the SAM dipole, while taking into account metal electrode effects, infl uences graphene has yet to be studied. In this paper we Recent reports have shown that self-assembled monolayers (SAMs) can induce doping effects in graphene transistors. However, a lack of understanding persists surrounding the quantitative relationship between SAM molecular design and its effects on graphene. In order to facilitate the fabrication of next-generation graphene-based devices it is important to reliably and predictably control the properties of graphene without negatively impacting its intrinsic high performance. In this study, SAMs with varying dipo...