Table 1.  Changes Made to the Original Version of the CACM During This Worka  			
Change	Reaction Numbers	Adaptations Made	Reasons/Notes
1	(79)	increase yield of RO221 to 0.1 and decrease that of  RAD3 to 0.74	Increasing the yield of RO221 enhances the formation of ring-retaining products.  This causes simulations to be in line with observations of Forstner et al. [1997] and AROL to be consistent with other aromatic parent species.
2	(97)-(109)	increase RAD reaction rate constants with NO2 by a factor of 1.7 and adjust those with O2 by a factor of 0.9	These changes also enhance the formation of ring-retaining products.
3	(128), (136), (156), (269), (284), (298), (306), (315), (329), (345)	decrease rate constants of acyl peroxy radical reactions with HO2 by a factor of 3.5 	At low VOC and NOx levels in the original versions of the modules, acyl radical consumption is dominated by reaction with hydroperoxy radicals, leading to acid products.  This generates high SOA yields in cases with low initial VOC concentrations, behavior not observed in laboratory chamber experiments [Odum et al., 1996, 1997a, 1997b; Hoffmann et al., 1997; Griffin et al., 1999].
4	(197)-(202)	decrease yield of species UR7 to 0.3 and add formation of species UR10 at a yield of 0.7b 	In the initial development of the model it was assumed that UR7 would not contribute significantly to SOA due to its relatively high vapor pressure.  However, its formation in high yield and its lumping into a less volatile surrogate lead to overpredictions of SOA.  UR10 is not a partitioning species in the model.
5	(221), (228), (242), (249), (256)	increase reaction rate by 20%	Faster isomerization of benzyl peroxy radicals leads to increased formation of anhydride and furan type products to be more in line with the results of Forstner et al. [1997].
6	(222)-(224), (229)-(231), (243)-(245), (250)-(252)	decrease yield of species RPR9 (or its corresponding product, RP11, RP12, or RP13, in the latter sets of reactions) to 0.3 and add formation of a new species (UR35) at a yield of 0.7	In the initial development of the model it was assumed that RPR9 (or its corresponding product in the latter sets of reactions) would not contribute significantly to SOA due to its relatively high vapor pressure.  However, its formation in high yield and its lumping into a less volatile surrogate lead to overpredictions of SOA.  UR35 is not a partitioning species in the model.
7	(235)	increase reaction rate by 10%	Faster isomerization of benzyl peroxy radicals leads to increased formation of anhydride and furan type products to be more in line with the results of Forstner et al. [1997].  A slower isomerization rate in this case relative to those listed above is one pathway by which AROH leads to higher yields of SOA than does AROL.
8	(236)-(238)	decrease yield of species RP11  to 0.5 and add formation of UR35 at a yield of 0.5	In the initial development of the model it was assumed that RP11 would not contribute significantly to SOA due to its relatively high vapor pressure.  However, its formation in high yield and its subsequent formation of UR26 lead to overpredictions of SOA.  In this case, 0.5 is used (as opposed to the 0.3 used above for RPR9) to provide another pathway by which AROH leads to higher yields of SOA than does AROL.
9	(334)	decrease the yield of UR26 to 0.67 and add the formation of UR24 at a yield of 0.33	UR26 is a major constituent of SOA from aromatic ring degradation products, which was originally simulated to be too high.  UR24 is not a partitioning species in the model.
10	(350)-(361)	treat highly functionalized alkyl nitrate (AP) products as unreactive species (set k350 through k361 = 0)	Structure-activity-relationship-based rate coefficients lead to rapid consumption of the AP products, leading to a peak followed by a decline in SOA concentrations in model simulations.  No such peaks have been observed in chamber experiments [Odum et al., 1996, 1997a, 1997b; Hoffmann et al., 1997; Griffin et al., 1999].
aReaction numbers and product designations refer to those presented by Griffin et al. [2002a].			
bIn reaction (197) these values need to be multiplied by CF(28).