Thanks for posting the log file and your follow-up comment.
Yes, mixing 3D and 2D files to create a 3D output file is the cause of your problem - if you want combine to generate a 3D output file (by changing the line #layer 1 in your SpecDef file to !#layer 1), all your input files need to be 3D files with the same number of layers. Otherwise, when mixing 3D and 2D files, some arrays will remain unfilled for the upper layers when combine reads 2D variables specified in the SpecDef file. If such variables are then used in any equations (like the expression containing the LOG function), these functions may cause a program error triggered by the large positive or negative uninitialized values. If they are just ‘passed through’ to the output file, the output file would contain bad values for these variables for the upper layers since they couldn’t be read from the (2D) input file.
If you want a 3D output file, the solution is to modify your SpecDef file to only reference variables for which you have 3D information available. If calculating sharp cut-off PM2.5 species is important for your work, you will have to make sure that you change your CCTM run script settings to create APMDIAG files for the same layers as your ACONC files. You would also have to remove any species from your SpecDef file that reference variables from the 2D METCRO2D file.
Please see posts 10 - 15 in this thread for an earlier discussion of the same problem.