Structural networks

Water boils at a much higher temperature than would be expected based solely on its molecular weight. The reason is that liquid water exhibits a highly structured network of hydrogen bonds. 

Diamond is a naturally occurring form of pure, crystalline carbon. Each carbon atom is surrounded by four others arranged tetrahe-drally. The result is a compact structural network bound by normal chemical bonds. This description offers a ready explanation for the extreme hardness and the great stability of carbon in this form. 

It should be noted that the ability of the filler to form the structural network is largely responsible for its reinforcing action [28, 33], These aspects of the problem have been discussed at length in a number of monographs (e.g. [7, 53, 54]) and we do not think it proper to dwell on them here. On the other hand the structurization in melt increases the viscosity of the material and hampers its processability. 

PECTIN FORMS A STRUCTURAL NETWORK WITHIN THE CELL WALL... 

We have created structured networks in whey proteins using mild heat and shear, to create reversible TWPs. By understanding on a molecular basis, the effects of shear, ways of creating new functionality can be developed. This will enable development of extrusion parameters that permit controlled denaturation of whey proteins. 

Walker, A.K., Cross, S.S., and Harrison, R.F., Visualisation of biomedical datasets by use of growing cell structure networks A novel diagnostic classification technique, Lancet, 354, 1518,1999. 

Wong, J.W.H. and Cartwright, H.M., Deterministic projection by growing cell structure networks for visualization of high-dimensionality datasets. /. Biomed. Inform., 38,322, 2005. 

Biomedical Datasets by Use of Growing Cell Structure Networks A Novel Diagnostic Classification Technique. 

Projection of Growing Cell Structure Networks for Visualization of High-Dimensionality... 

For a theoretical description of crosslinking and network structure, network formation theories can be applied. The results of simulation of the functionality and molecular weight distribution obtained by TBP, or by off-space or in-space simulations are taken as input information. Formulation of the basic pgf characteristic of TBP for crosslinking of a distribution of a hyperbranched polymer is shown as an illustration. The simplest case of a BAf monomer corresponding to equation (4) is considered ... 

Differences in Network Structure. Network formation depends on the kinetics of the various crosslinking reactions and on the number of functional groups on the polymer and crosslinker (32). Polymers and crosslinkers with low functionality are less efficient at building network structure than those with high functionality. Miller and Macosko (32) have derived a network structure theory which has been adapted to calculate "elastically effective" crosslink densities (4-6.8.9). This parameter has been found to correlate well with physical measures of cure < 6.8). There is a range of crosslink densities for which acceptable physical properties are obtained. The range of bake conditions which yield crosslink densities within this range define a cure window (8. 9). 

It is usually very difficult to transform a liquid into a glass since nearly all liquids or melts crystallize when undercooled. The question as to which liquids can be undercooled has recently been discussed by Turnbull (< ). From a more chemical standpoint Zachariasen s ideas are well accepted (20). He states that a glass can be formed if the liquid contains a structural network in the temperature range near the melting point. This network must be broken in order to form crystal nuclei and this is not possible if the liquid is undercooled too much, thus preventing nucleation... 

In addition to the above experimental point, one can raise a theoretical objection against the way in which Volkenstein et al. introduce the effect which the structure in a network has on its elastic behaviour. In their theory the Gaussian chain statistics are left unchanged in spite of the fact that the chain molecules run through bundles. Such a decoupling of chain statistics and bundles is unwarranted. In Fig. 29 c a schematic representation of the approach of Volkenstein et al. to a structured network is given. Only a two chain network is drawn, although it should, of course, be remembered that in reality a bundle structure will comprise parts of many molecules. 

Blokland (14) recently also considered the stress-strain behaviour of structured networks. His approach is schematically illustrated in Fig. 29. Consider a cubical lattice on which the chain configurations are laid out in a partially obstructed random walk. Of the N steps of each chain there will be on the average m steps which participate in a bundle structure... 

Fig. 29. Schematic representation of a part of a structured network and its theoretical treatment (a) ideal network structure without bundles (b) network with a simple"two chain bundle" (c) theoretical treatment by Volkenstein et al. (174). The separated "bundle illustrates the intactness of the original network chain statistics (d) theoretical treatment by Blokland (14). Each chain has a number of obstructed steps. No relation between the obstructed parts of different chains...

Some remarks in connection with Blokland s theory are appropriate. In the first place it must be pointed out that an additional parameter is introduced in the theory, which is supposed to reflect the existence of intermolecular obstructional effects ( %). It is questionable whether this parameter is an accurate reflection of a structured network,... 

The cooperative, infinite chains and cycles formed by O-H 0 hydrogen bonds in the a-cyclodextrin hydrates are a characteristic structural motif [109]. As with the simpler carbohydrate crystal structures described in Part II, Chapter 13, the hydrogen bonds can be traced from donor to acceptor in the cyclodextrin hydrate crystal structures. Networks of O-H 0-H 0-H interactions are observed in which the distribution of hydrogen bonds follows patterns with two characteristic motifs. One are the "infinite chains which run through the whole crystal lattice, and the others are the loops or cyclically closed patterns (a special case of the "infinite chains). As in the small molecule hydrates, such as a-maltose monohydrate, the chains and cycles are interconnected at the water molecules to form the complex three-dimensional networks illustrated schematically in Fig. 18.5, with some sections shown in more detail in Fig. 18.7 a, b, c. 

Keywords Crystal engineering Extended structures Networks Hydrogen bonding Self-assembly... 

At the first level, the sequence-to-structure network (NN1) was trained to classify mutually independent segments of residues (13-amino acid-long) in terms of the state of a single central residue. A new key aspect was the use of evolutionary information contained in multiple sequence alignments as input to a neural network in place of single sequences. Each sequence position was represented by the amino acid residue... 

A second level structure-to-structure network (NN2) was used to introduce a correlation between the secondary structure of adjacent residues in 17-amino acid windows. The structure context training provided better prediction of helix and strand lengths. The output (the three-state values of the central residue) from the first network was input to the second network. In addition, a spacer unit and a conservation weight unit were used for each residue, as well as the global information of the protein. The total number of input units for NN2 is 5 x 17 (for w =17) + 20 + 12. The secondary structure prediction program (PHDsec) is available via an automatic E-mail prediction service (Rost, 1996). 

Rost and Sander (1994b) developed another neural network system to predict the relative solvent accessibility (PHDacc). The one-level network system used the same input information as that in the PHDsec sequence-to-structure network, and mapped it to ten output units coded for ten relative levels of solvent accessibility. PHDacc was superior to other methods in predicting the residues in either of the two states, buried or exposed. Entirely buried residues (<4% accessible) were predicted best. 

---