How do I change the structural layout when using semPaths?

Question

I am trying to plot an SEM model with the manifest variables and latent variable of the outcome variable on one side and the manifest and latent variables of the predictor, which is measured as a 2nd order factor model, on the other side of the plot.

I have tried different layout functions like tree, spring, and circle.

factor.modelSEM5 <- '
ConDCS =~ c1*o1_a+c2*o2_a+c3*o3_a
NegDCS =~ n1*o4_a+n2*o5_a+n3*o6_a
IntDCS =~ i1*o7_a+i2*o8_a+i3*o9_a

DCS =~ DC1*ConDCS+DC2*NegDCS+DC3*IntDCS
DCS ~~ 1*DCS

AUDITC =~ A1*b1_a_R+A2*b2_a_R+A3*b3_a_R
AUDITC ~~ 1*AUDITC

AUDITC ~ DCS
'
mySEM.fit5 <- sem(factor.modelSEM5, SEMProject)

semPaths(mySEM.fit5, whatLabels="std", layout="tree2",curvePivot = TRUE, 
rotation=2,sizeMan=6,posCol="black", sizeLat=6, edge.label.cex=.65)

With this code, all manifest variables (o1_a through b1_a_R) are lined up next to each other one on side of the plot.

Instead, I want the manifest variables,

• o1_a through o9_a, to line one side of the plot and the manifest variables, and
• b1_a_R through b3_a_R, to line the opposite side of the plot.

Thanks for your help.