Alkali metal doping is essential to achieve highly efficient energy conversion in Cu(In,Ga)Se 2 (CIGSe) solar cells. Doping is normally achieved through solid state reactions, but recent observations of gasphase alkali transport in the kesterite sulfide (Cu 2 ZnSnS 4 ) system (re)open the way to a novel gas-phase doping strategy. However, the current understanding of gas-phase alkali transport is very limited. This work (i) shows that CIGSe device efficiency can be improved from 2% to 8% by gas-phase sodium incorporation alone, (ii) identifies the most likely routes for gas-phase alkali transport based on mass spectrometric studies, (iii) provides thermochemical computations to rationalize the observations and (iv) critically discusses the subject literature with the aim to better understand the chemical basis of the phenomenon. These results suggest that accidental alkali metal doping occurs all the time, that a controlled vapor pressure of alkali metal could be applied during growth to dope the semiconductor, and that it may have to be accounted for during the currently used solid state doping routes. It is concluded that alkali gas-phase transport occurs through a plurality of routes and cannot be attributed to one single source.Control of alkali doping is crucial for a range of technologically relevant chalcogenide materials, from photovoltaics (CdTe, Cu(In,Ga)Se 2 , Cu 2 ZnSn(S,Se) 4 ) 1-5 and thermoelectricity (Pb(S,Se,Te)) 6-9 potentially to superconductivity (KFeSe 2 ) 10,11 and quantum computing (Bi 2 Te 3 12 , MoS 2 and WSe 2 13 ). In the case of Cu(In,Ga)Se 2 (CIGSe) solar cell material, the current alkali metal doping procedures are overwhelmingly based on condensed state reactions. Two common approaches are taken. Either by indirect control of the diffusion from a sodium-containing substrate or back contact [14][15][16][17] , or by deliberate doping from the precursor surface through a post deposition treatment (PDT), e.g. by NaF or KF evaporation onto the surface of the absorber to form a tens of nanometer thick layer, followed by annealing [18][19][20] . Control of the sodium content in the former case is difficult as substrates are never identical 21 , and in the latter case at least one extra step is required to add the alkali metal. The subject has been extensively reviewed by Salomé et al.
22. CIGSe thin films are always grown in a controlled atmosphere containing a certain pressure of selenium. The semiconductor requires selenium for its formation and to prevent its decomposition, given that the reaction is ruled by a solid/gas-phase equilibrium 23,24 . The question arises, are any other gas-phase chemical species involved in the equilibrium? All the main binary compounds of CIGSe have low vapour pressures; however, usually CIGSe contains also a considerable amount of sodium incorporated in the film. Is the vapour pressure of sodium or its
