Congratulations, now can you actually link to the paper instead of summarizing it in half a sentence as is "caused by high-strain gradient built up inside the device"? This is useless reporting.
From my reading it sounds like we've known for sometime that it was due to strain on the board, but the work provides some insight by mapping stresses, and SEM imaging.
From the conclusion:
"In summary, a methodology has been developed to incorporate DIC technique with SEM images for mapping the local strain fields during whisker growth. By processing SEM images taken in consecutive time intervals over the area of interest on Sn-Cu finishes, the local strain evolution due to whisker growth can be obtained. This methodology provides an alternative way for exploring the complicated process of whisker growth through the page 9 of 13 morphology changes. From the DIC results, it is proposed that strain or stress gradient, instead of an overall compressive stress field is the key for whisker growth. Results from SEM and DIC analysis also indicate that the whisker growth is a continuously dynamic process, during which the subsequent whisker is triggered by the redistribution of strain or stress field after local strain relaxation. The findings have advanced the understanding of whisker growth mechanisms and may provide insight for developing whisker mitigation technology for lead-free solder alloys."