# Comprehensive dataset on macro-porous PVDF flat sheet membranes for membrane distillation: Materials characteristics, morphology and performance data

## TL;DR

This paper provides a detailed dataset on creating macro-porous PVDF flat sheet membranes for membrane distillation using a vapor-assisted process, including experimental designs and performance metrics.

## Contribution

The paper introduces a comprehensive dataset from a structured experimental design to optimize PVDF flat sheet membranes for membrane distillation.

## Key findings

- A 33-run face-centered composite design was used to study six factors affecting membrane properties.
- Linear regression models showed strong relationships between process factors and membrane responses (R²: 67–94%).
- SEM micrographs captured morphological variations across the design space.

## Abstract

This data article presents a structured dataset from the transfer of a proven PVDF hollow-fiber formulation to macro-porous PVDF flat sheet membranes via a vapor-assisted non-solvent induced phase separation (VNIPS) process designed for membrane distillation (MD). The study uses N-methyl-2-pyrrolidone (NMP) as a solvent and water as a non-solvent. A face-centered composite design (33 membranes) was implemented to efficiently sample six controllable factors at three levels each: polymer content (16–20 wt %), solvation temperature (30–60 °C), wet casting thickness (300–500 µm), vapor-induced phase separation (VIPS) conditioning time (60–360 s at fixed 90 % RH), coagulation bath temperature (25–50 °C), and bath solvent content (0–50 wt %). For each treatment, one 100 × 200 mm membrane was cast on a lab-scale line (dope dispense + doctor blade → VIPS tunnel with controlled humidity/airflow → NIPS bath), then transferred to polyester fleece and dried at 40 °C for 24 h. Materials comprised PVDF (Mw ≈ 534 kDa), N-methyl-2-pyrrolidone (NMP, 99.5 %), and RO water.

The dataset includes: (i) full experimental design with factor settings for all 33 runs; (ii) raw and processed characterization results—mean thickness, water contact angle, liquid entry pressure, porosity, and permeate flux—for each membrane; (iii) linear regression models linking process factors to responses, constructed by retaining only coefficients with p < 0.05 and summarized by goodness-of-fit (R²: 67–94 % across endpoints); and (iv) scanning electron microscopy (SEM) micrographs of the top surface, bottom surface, and cross-section of each membrane, capturing morphological changes across the design space.

The accompanying tables also report the materials list and an optimization scenario (target-value approach) that returns factor settings maximizing wetting resistance and flux within the explored design bounds. These data enable reuse for multiple purposes: reproducing and extending VNIPS DoE studies; meta-analysis of factor–response relationships in phase inversion casting; benchmarking inverse design, response-surface, or machine-learning models; informing scale-up constraints and uncertainty analyses; and guiding MD membrane pre-screening under alternative objective functions or constraints.