EVALUATION OF VOLUME-TEMPERATURE RELATIONSHIPS IN THE NEMATIC PHASE OF MBBA NEAR ITS NEMATIC-ISOTROPIC TRANSITION BY MEANS OF COMPUTER MINIMIZATION METHODS

Ray-I Chang, J. C. Gysbers1975 · DOI

TL;DR

This paper presents a robust method for analyzing the complex volume-temperature behavior of liquid crystals near their phase transitions. By employing two independent computer minimization techniques, the authors successfully fit experimental data to a five-parameter nonlinear equation, providing a more accurate and globally consistent description of these critical phenomena.

What Problem Does This Paper Address?

Understanding the behavior of materials near phase transitions is crucial in many scientific fields. Liquid crystals, in particular, exhibit fascinating pretransition phenomena where their properties gradually change as they approach a distinct phase change, like the transition from a nematic to an isotropic liquid state. Describing these subtle changes accurately requires sophisticated mathematical models. The authors aim to precisely characterize the volume-temperature relationship of MBBA (a common liquid crystal) in its nematic phase, just before it transitions to the isotropic phase. Previous attempts to fit this data often involved arbitrary assumptions, like fixing certain parameters, which could lead to non-unique or less accurate solutions. This paper seeks to overcome these limitations by using advanced computational methods to fit a five-parameter nonlinear equation to experimental data without making such restrictive assumptions.

Methods

The core of this paper lies in the application of two distinct computer minimization methods to fit experimental volume-temperature data to a five-parameter nonlinear equation: $V = a + bT + c(T_c - T)^n$.

  1. PROSE (JOVE) Constrained Optimization Method: This method utilizes penalty functions to transform a constrained optimization problem into a sequence of unconstrained ones. The objective function is augmented with penalty terms for constraint violations, and these penalties are gradually reduced. This iterative process aims to find the optimal parameters that minimize the sum of squared differences between experimental and calculated volumes. The authors specifically mention that this method, when applied to their fitting function, is likely to yield a unique solution due to the "mildly nonlinear function with no oscillating terms."
  2. APL Nonlinear Minimization Method: This approach first employs a random minimization technique to locate an approximate global minimum. The original five parameters ($a, b, T_c, n$) are transformed into new variables ($a, d, c, p, m$) to remove explicit constraints and simplify the minimization process. The objective function is then differentiated with respect to these new parameters, and the resulting equations are used to iteratively refine the parameter values. The process continues until a preassigned number of trials is reached, aiming to ensure the final set of parameters is close to the true global minimum.

Both methods were used to analyze the same experimental dataset, and their results were compared.

Datasets

The authors used a dataset of thirty volume-temperature data points for distilled MBBA, collected in their laboratory. This data was obtained from a single experiment and covers the pretransition region of the nematic phase, extending up to the beginning of the first-order nematic-isotropic transition temperature ($T_B$). The data was presented in Table I, with temperature readings in millivolts from a thermocouple and corresponding volume readings in arbitrary units from a capillary.

Key Insights and Ideas

  • Global Solution Robustness: The paper strongly suggests that both independent minimization methods converge to a global solution, meaning they are not trapped in local minima of the error function. This is a critical finding for ensuring the reliability of the fitted parameters.
  • No Arbitrary Assumptions Needed: By employing these advanced minimization techniques, the authors were able to fit the five-parameter equation without making arbitrary assumptions about parameters like the coefficient of volume thermal expansion ($b$), which was a limitation in previous studies.
  • Universality of the Approach: The authors highlight that the described minimization techniques are broadly applicable to analyzing pretransition and second-order phase transitions in various liquid crystalline systems, not just for volume-temperature data but also for other physical properties like density, specific heat, and ultrasonic attenuation.
  • Potential Impact of Computer Word Length: The slight differences in the optimized parameters obtained by the two methods, and the fact that feeding PROSE parameters into the APL program reproduced the APL results, led the authors to suggest that the computer's word length might be a factor in achieving the best results for nonlinear least squares problems.

Key Results

Both the PROSE (JOVE) and APL methods successfully fitted the experimental volume-temperature data to the five-parameter nonlinear equation. The resulting SUMSQ (sum of squares of experimental minus calculated values) values were very low, indicating an excellent fit:

  • PROSE (JOVE) SUMSQ: 0.10658
  • APL SUMSQ: 0.11550

While both methods yielded global solutions, the optimized parameters differed slightly between the two methods. For instance, the parameter $a$ was 5.7046 for PROSE and 1.9921 for APL, and $b$ was 34.5649 for PROSE and 32.4444 for APL. The authors note that these slight differences might indicate a broad global minimum in the parameter space.

Limitations and Future Work

  • Parameter Differences: The authors acknowledge the slight differences in the optimized parameters obtained by the two methods. They suggest this could be due to the global minimum being "rather broad" and the methods approaching it from different directions.
  • Computer Word Length: The paper hints that the computer's word length might influence the precision of the results, suggesting this as a factor to consider in future work.
  • Future Applications: The authors explicitly state that the described computer minimization methods are expected to have "wide application in data analyses pertaining to pretransition and second order transition in liquid crystalline systems." This implies a call for researchers to adopt these techniques for similar studies.

Summary updated

Abstract

The volume (V)-temperature (T) data for MBBA in the pretransition region slightly below its nematic-isotropic transition temperature are fitted to the five-parameter nonlinear equation V = a + bT + c(Tc - T)n be means of two independent methods : the Control Data Corporation PROSE constrained optimization method and an APL nonlinear minimization method. Both methods yield global solutions. The computer minimization methods described here are expected to have wide application in data analyses pertaining to pretransition and second order transition in liquid crystalline systems.
BibTeX
@inproceedings{chang1975evaluation,
  author = {Ray-I Chang and J. C. GYSBERS},
  title = {{EVALUATION OF VOLUME-TEMPERATURE RELATIONSHIPS IN THE NEMATIC PHASE OF MBBA NEAR ITS NEMATIC-ISOTROPIC TRANSITION BY MEANS OF COMPUTER MINIMIZATION METHODS}},
  booktitle = {Le},
  year = {1975},
  doi = {10.1051/jphyscol:1975128},
}
Figure