Error synthesis

Fig. 214. Sodium benzyl penicillin, (a) Approximate electron density map, b projection, with atomic coordinates appropriate to extended molecular configuration. (6) Error synthesis map. (c) The same electron density map as (a), with atomic coordinates appropriate to a curled configuration of the same molecule.

The work of Cox, Cruickshank, and Smith (1958) on the crystal structure of benzene at — 3° C (a little below the melting-point) illustrates well this sort of application of the error synthesis. Fig. 215 shows the error synthesis (or difference synthesis) map in the plane of the benzene ring after a series of refinements in which only carbon atoms were included in the structure amplitude calculations, and thermal vibrations were assumed to be isotropic with a temperature factor B = 6-0 A2. 

Fio. 215. Benzene. Error synthesis maps. Above—Structure Amplitude calculations based on carbon atoms only, and on isotropic thermal motions. Below—After hydrogen atoms and anisotropic thermal motions were included in calculations. (Cox, Cruickshank and... 

The traditional way to seek new zeolites was primarily based on a trial-and-error synthesis process. With the growing needs for applications of zeolite materials, the rational synthesis of zeolites with desired structures and functions has become the objective of synthetic chemists. However, the rational synthesis of zeolites with desired structures is still a great challenge due to the unclear formation mechanism of zeolite crystals. With the advancement of understanding the synthesis chemistry and strucmre chemistry of zeolites in recent... 

Historically, the discovery of one effective herbicide has led quickly to the preparation and screening of a family of imitative chemicals (3). Herbicide developers have traditionally used combinations of experience, art-based approaches, and intuitive appHcations of classical stmcture—activity relationships to imitate, increase, or make more selective the activity of the parent compound. This trial-and-error process depends on the costs and availabiUties of appropriate starting materials, ease of synthesis of usually inactive intermediates, and alterations of parent compound chemical properties by stepwise addition of substituents that have been effective in the development of other pesticides, eg, halogens or substituted amino groups. The reason a particular imitative compound works is seldom understood, and other pesticidal appHcations are not readily predictable. Novices in this traditional, quite random, process requite several years of training and experience in order to function productively. 

Rasmussen, J. 1979. Notes on human error analysis and prediction. In G. Apostalakis and G. Volta (Eds.), Synthesis and Analysis Methods for Safety and Reliability Studies, Plenum, New York. 

To facilitate the use of methanol synthesis in examples, the UCKRON and VEKRON test problems (Berty et al 1989, Arva and Szeifert 1989) will be applied. In the development of the test problem, methanol synthesis served as an example. The physical properties, thermodynamic conditions, technology and average rate of reaction were taken from the literature of methanol synthesis. For the kinetics, however, an artificial mechanism was created that had a known and rigorous mathematical solution. It was fundamentally important to create a fixed basis of comparison with various approximate mathematical models for kinetics. These were derived by simulated experiments from the test problems with added random error. See Appendix A and B, Berty et al, 1989. 

A more difficult criterion to meet with flow markers is that the polymer samples not contain interferents that coelute with or very near the flow marker and either affect its retention time or the ability of the analyst to reproducibly identify the retention time of the peak. Water is a ubiquitous problem in nonaqueous GPC and, when using a refractive index detector, it can cause a variable magnitude, negative area peak that may coelute with certain choices of totally permeated flow markers. This variable area negative peak may alter the apparent position of the flow marker when the flow rate has actually been invariant, thereby causing the user to falsely adjust data to compensate for the flow error. Similar problems can occur with the elution of positive peaks that are not exactly identical in elution to the totally permeated flow marker. Species that often contribute to these problems are residual monomer, reactants, surfactants, by-products, or buffers from the synthesis of the polymer. 

Conditions that are important to all chemical reactions such as stoichiometry and reactant purity become critical in polymer synthesis. In step growth polymerization, a 2% measuring and/or impurity error cuts the degree of polymerization or the molecular weight in half. In chain growth polymerization, the presence of a small amount of impurity that can react with the growing chain can kill the polymerization. 

Even as the computational prediction error rate is reduced to acceptable levels, many cases will be encountered in which the predictions are indistinguishable to within error. In a scenario in which several different in silico designs are given equivalent but favorable activity predictions, the end user s medicinal experience may help decide which to promote to synthesis. The quality of that decision at this point will be strongly influenced by how easy it is to understand the different contributions to the computational predictions. Interpretability is thus critical for synergistically utilizing the experience of the end user. 

---