Control production of polyester resins by NIR spectroscopy M. Blanco a, ⁎, J. Cruz a , M. Armengol b a Departament de Química, Unitat de Química Analítica, Universitat Autònoma de Barcelona, E-08193 Bellaterra, Barcelona, Spain b Powder Coatings Department, Cray Valley Ibérica, S.A. E-08470 Sant Celoni, Barcelona, Spain ABSTRACT ARTICLE INFO Article history: Received 25 March 2008 Accepted 17 April 2008 Available online 25 April 2008 Keywords: NIR spectroscopy Multivariate calibration PLSR Acid value Hydroxyl value Model transfer The control of the esterification reaction for production of polyester saturated resins is followed usually by determination of the acid value (AV) and hydroxyl value (OHV).These parameters are determined by titrimetry, but these methods are slow, intensity working and produce waste. In this paper an alternative methodology is proposed, based in the construction of multivariate models on NIR spectroscopic data and different models are constructed in order to apply to different steps of the production process. The ensuing methodology provides models of good predictive ability and constitute an advantageous alternative to existing titrimetric reference methods as regards expeditiousness and environmentally compatible. The multivariate calibration models established were also used with a different instrument; to this end, the spectra recorded with the original equipment were subjected to Piecewise Direct Standardization (PDS) in order to make them equivalent to those provided by the new equipment. Also, PLS calibration was reproduced by using the same samples, spectral treatment, wavenumber range and number of factors as in the original model, and the AV and OHV results thus obtained were similarly good. © 2008 Elsevier B.V. All rights reserved. 1. Introduction Polyester resins are polymers resulting from the condensation of one or more dicarboxylic acids with one or more dialcohols and the characteristics and physical properties of which can be modified by altering the reactant proportions; this allows a variety of resins ranging from solid to semi-fluid in texture that are insoluble in water but soluble in organic solvents to be obtained. The earliest synthetic resins were obtained in the 1940s and used as reinforcements for fibreglass in the production of the first composite plastics. Although their market and uses have grown substantially since then, synthetic resins continue to be used largely in this manufacturing in order to obtain layered composites known as “fibreglass-reinforced plastics” (FRPs).The most widely used polyester resin is polyethylene terephthalate (PET), which is obtained by condensing terephthalic acid with ethyleneglycol; PET is a thermo- plastic the mechanical and chemical properties of which make it especially suitable for food and drink packaging. In addition, polyester resins are widely used to manufacture plastic paint [1] and these resins are used to prepare reticulated powdered paint for electrostatic coating. The development of this kind of resins has always been very influenced by the environmental factors, because of the absence of solvents and it is an agreed alternative with the ecological principles. This type of paintings has had a fast growth from their introduction for 35 years due to their fast cured and the optimal qualities of the formed covering, with very good mechanical properties, resistance and durability. The plastic paint polyester resins studied are obtained by condensation of neopentyl glycol and terephthalic acid, but addition- ally contain minor amounts of other acids and diols intended to facilitate specific uses. Table 1 lists some of the more usual components of these resins. The American Society for Testing and Materials (ASTM) [2] recommends using the acid value (AV) and hydroxyl value (OHV) to monitor esterification reactions. Determining these two parameters (particularly OHV) involves long analysis times that lengthen the production cycle and result in increased costs and decreased productivity as a result. The production process and the quality of the end product can be improved by using rapid, effective analytical techniques to ensure appropriate, efficient analytical control of the different steps of the manufacturing process. A variety of techniques have so far been used for the analysis and characterization of resins [3–5] all of which, however, involve labour- intensive, time-consuming analyses. Near infrared spectroscopy (NIR) is a novel, effective alternative in his respect as it allows the spectra for untreated solid and liquids samples to be expeditiously recorded and affords the simultaneous determination of physical and chemical parameters from a single spectrum. In addition, it lends itself readily to at-line and in-line implementation (with fibre-optic probes in the latter case), which makes it suitable for process control purposes [6]. The nature of NIR spectra, however, requires the use of multivariate regression techniques to construct effective calibration models the most widely employed among which is Partial Least-Squares Microchemical Journal 90 (2008) 118–123 ⁎ Corresponding author. Tel./fax: +34 93 5811367. E-mail address: marcel.blanco@uab.es (M. Blanco). 0026-265X/$ – see front matter © 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.microc.2008.04.004 Contents lists available at ScienceDirect Microchemical Journal journal homepage: www.elsevier.com/locate/microc