Once the wavefunction is in hand, all observable properties can, at least in principle, be computed. This can include all varieties of spectral properties, which are particularly valuable in ascertaining molecular structure. The full theoretical and computational means for computing spectral properties are mathematically involved and beyond the scope of this chapter. Instead, we will focus on the use of quantum chemical computations to help identify chemical structure. The purpose of this chapter is to inspire the routine use of computed spectra to aid in structural identification.As discussed in Chapter 1, the full three-dimensional structure of a compound can be optimized with almost any of the quantum computational techniques. Since most quantum chemical computations are still performed on a single molecule in the gas phase, these computed structures can be most readily compared to gas-phase experimental structures. In the following chapters, we will present a number of case studies where computed and experimental geometries are compared. To get a sense of the quality of computed geometries, a few selected cases are discussed next.
COMPUTED BOND LENGTHS AND ANGLESSchaefer and Allinger 1 examined a series of small organic compounds with a number of different computational methods. Table 2.1 compares the C-C and C-H bond lengths (r e ) in some very small compounds, and finds the coupled-cluster singles and doubles (CCSD/TZ2+P) results in excellent agreement with the limited experimental values. Importantly, the bond lengths predicted by the much more computationally efficient B3LYP/6-31G* are very similar to the CCSD results. For a group of 10 small hydrocarbons, including some alkenes and alkynes, the B3LYP/6-31G* C-C and C-H distances are on an average too small (compared with r g values) by only 0.0031 and 0.016 Å, respectively. Similar excellent performance is also seen for a group of 19 organic compounds containing oxygen or nitrogen.