Borophosphosilicate glass (BPSG) thin films have been widely used in semiconductor device fabrication as interlayer dielectric films due to their excellent planarization, flow properties, and passivation barriers against moisture penetration. FTIR spectroscopy combined with chemometric techniques has been shown to accurately measure boron and phosphorus concentrations in BPSG thin films. A neural network is a set of simple, highly interconnected processing elements imitating the architecture of the human brain, which are capable of learning information presented to them. In this work Reflectometry measured at the same location as the FTIR is used to accurately measure film thickness. The thickness along with the FTIR data are then presented to a trained neural network which predicts the boron and phosphorus concentration in the film. Reflectance, FTIR and an independent measurement of concentration, XRF in this case, from a set of calibration wafers is required to train the neural network. Predicted concentrations from both calibration and test wafers were an average of 1 percent off XRF measurements. This measurement program was implemented on a Nanometrics NanoSpec 8000XSE, a standalone metrology tool, and will simultaneously measure thickness, index and both boron and phosphorus concentrations. This measurement can be easily transferred to the NanoSpec 9000, which is an integrated metrology tool. This measurement system is ultra-compact and was designed specifically for integration into semiconductor processing equipment. Integrated measurements of thickness, optical constants and dopent concentration would improve overall equipment effectiveness through advanced process control. This would provide process engineers immediate feedback and should reduce scrap and monitor wafers.