Abstract

In this work, a procedure was elaborated to quantify hydrogen peroxide (H₂O₂) after degradation tests of dyes present in a synthetic textile matrix. For this purpose, based on the intensity of radiation absorption of the peroxovanadium cation, which was formed by the reaction between the oxidant and vanadate ions, ultraviolet/visible spectrophotometry technique was used. The most suitable experimental condition was composed of concentrations of 0.05 mol L⁻¹ (NH₄VO₃) and 0.3 mol L⁻¹ (H₂SO₄). The system for dye treatment involved the photo-Fenton process under simulated sunlight. In this case, concentrations of 900 mg L⁻¹ (H₂O₂) and 4 mg L⁻¹ (iron) in pH 3 were the most efficient for degrading contaminants. An efficiency of 94.49% was obtained after 180 min of reaction, the time in which the presence of the oxidant was no longer verified. The kinetic monitoring showed a two-stage degradation, described with accuracy greater than 96% by the linear and non-linear kinetic models of pseudo-first order. Additionally, the degradation under natural solar radiation was also studied, which resulted in an efficiency of 92.45% after 360 min and in the presence of the residual oxidant. Finally, via mathematical modeling and employing a Multilayer Perceptron neural network, with a 3-10-2 topology and BFGS 387 training algorithm, it was possible to predict the degradation and H₂O₂ residual concentration with an accuracy greater than 98%. Therefore, the degradation study developed and the proposed methodology for determining residual H₂O₂ proved to be adequate and capable of contributing positively to related research.

Graphic abstract

中文翻译：

---

阳光下的光芬顿法降解纺织废水：分光光度法监测残留的过氧化氢并建立人工神经网络模型以预测处理

摘要

在这项工作中，精心设计了量化合成纺织品基质中存在的染料降解测试后的过氧化氢（H ₂ O ₂）的程序。为此，基于通过氧化剂和钒酸根离子之间的反应形成的过氧钒阳离子的辐射吸收强度，使用了紫外/可见分光光度法。最合适的实验条件是由0.05 mol L ^-1（NH ₄ VO ₃）和0.3 mol L ^-1（H ₂ SO ₄）组成。染料处理系统涉及在模拟阳光下的光芬顿工艺。在这种情况下，浓度为900 mg L^-1（H ₂ O ₂）和4 mg L ^-1pH为3的（铁）对降解污染物最有效。反应180分钟后效率达到了94.49％，此时不再确认氧化剂的存在。动力学监测显示出两阶段降解，通过伪一级反应的线性和非线性动力学模型描述，其准确度大于96％。此外，还研究了在自然太阳辐射下的降解，在360分钟后和存在残留氧化剂的情况下，效率为92.45％。最后，通过数学建模并采用具有3-10-2拓扑和BFGS 387训练算法的多层感知器神经网络，可以预测降解和H ₂ O ₂残留浓度的准确度大于98％。因此，开发了降解研究，并且所提出的用于确定残留H ₂ O _2的方法被证明是足够的，并且能够为相关研究做出积极的贡献。

图形摘要