Couple of very important justications in this paper.

http://www.nature.com/nmat/journal/v13/n9/full/nmat4065.html  

1. Pitfalls in measuring conversion efficiencies.

Independently of the cell embodiment used, it is good practice to carefully verify the presence of any hysteresis by collecting the photocurrent–voltage curves in both forward and reverse voltage scanning directions anda s a function of scanning speed.

Failure to do so leads to large overestimations ofthe device PCE.

Especially, PCE values previously reported in literature for planararchitectures should be carefully checked.

A further source of error lies in measuring cell efficiencies on devices smaller than 0.1 cm2. It has been shown that for such small cells opticalartefacts can inflate the values of the short-circuit photocurrent.

2. Doping of CH3NH3PbI3 with chlorideions. It has been suggested that replacingPbI2 with PbCl2 as a starting material to form CH3NH3PbI3 can lead to doping ofthe perovskite structure with chloride ions. However, no chloride ions could bedetected in the CH3NH3PbI3 crystals formed during the reaction despite extensiveanalytical efforts. Based on careful X-ray diffraction and elemental analysis, thegroup of Peter Erk has recently confirmed the absence of any chloride doping inCH3NH3PbI3 crystals formed from PbCl2 in this manner (keynote lecture, 6^thInternational Conference on Hybrid and Organic Photovoltaics, HOPV 14, 11–14May 2014, Lausanne Switzerland). Instead of substituting iodide in the CH3NH3PbI3lattice, the chloride ions become part of an amorphous lead-containing phase thatmelts at around 103 °C. In view of these unambiguous results, the use of formulas such as CH3NH3PbI3–xClx to indicate a chloride-doped state of the lead iodideperovskite is no longer justifiable and should be avoided.