用于ISE硝态氮检测的土壤原位前处理装置设计与试验

前处理自动化程度是影响土壤养分检测效率和可行性的关键环节之一。该文设计开发了1种适用于电极法硝态氮原位测定的前处理装置,融合土壤水分校准算法,装置具备水分测定、称重、注液浸提、高速旋转离心等功能,性能测定结果表明:装置两自由度机械平台纵向控制精度达到0.1 cm,旋转定位控制精度为1?,注液精度可达0.2 m L,位移量及注液量具有较高重复性;较常规手动操作,利用该装置进行土样前处理耗时明显降低,自动化及时效性明显提高,人工劳动强度下降;利用装置前处理土样,电极法测定结果与标准光学检测结果接近,两者回归方程调整决定系数为0.87,平均绝对偏差为20.97 mg/kg,均方根误差为22.34 mg/kg,较悬液直测精度明显提高。装置浸提环节仍需进一步优化,研究结果可为集成式土壤养分原位检测装置研发提供参考。

Design and experiment of prototype soil pretreatment device for ISE-based soil nitrate-nitrogen detection

Abstract: Rapid lower-cost detection methods for soil macronutrient, including soil samping, sample pretreatment and nutrient determination, are urgently needed. In this paper, a prototype soil rapid pretreatment device was introduced for the ion-selective electrode (ISE) based in-situ soil nitrate-nitrogen detection. Necessary pretreatment steps of soil weighting, extracting and filtration were achieved in the integrated device. The mechanical structure of the device was composed of 4 major components, including a two-dimensional work bench, a high-speed centrifuge, a liquid pump and a self-designed electric control unit. The two-dimensional work bench was built with a vertical sliding table and a rotary table. On the purpose of eliminating the "suspension disturbance" of ISE, the centrifuge was used to obtain the clear soil extract solution. The peristaltic liquid pump, which conducted the quantitative injection of the soil extractant solution into centrifugal tube, was involved for the soil extractant injection and stirring. The electric control unit consisted of a microprocessor of STM32F103 and its interface circuit, which was used for the precise control of the two-dimensional work bench. The soil moisture sensor was applied to reduce the interference of soil water content. Net weight of the testing sample could be calculated. According to the testing results, the displacement accuracy of the vertical sliding table was less than 0.1 cm, while the root mean square error (RMSE) and mean relative error (MRE) were less than 0.04 cm and 0.4% respectively. The one-way displacement of 10 cm consumed less than 5 s. The accuracy of the angular movement for the rotary table was 1°. The rotating speed was calculated through the ratio between rotation angle and time, which could reach 10°/s. The two-dimensional work bench possessed good repeatability with the coefficient of variation (CV) of less than 0.40%. The accuracy of the liquid pump was also evaluated. Two injection volumes of 12.5 and 25 mL were selected. A volume accuracy of 0.2 mL was obtained. The RMSE, the MRE and the injection time were examined to be 0.07 mL, 0.24% and 13 s, respectively. The continuous injection operation possessed a high repeatability with the CV of less than 0.1%. Therefore, the extractant injection performance was verified to be reliable and accurate. The performance of the self-designed device was estimated in the nitrate-nitrogen detection of wet soil. Comparisons were conducted among 3 groups of ISE-based detections and the standard spectrometric detection. The 3 groups of ISE-based detections included the fast pretreatment group, the soil suspension group and the conventional pretreatment group. The fast pretreatment group was processed by the self-developed pretreatment device discussed above. Samples of the soil suspension group were prepared without filtration operation. The conventional pretreatment group was conducted under the national site-specific soil testing recommendation. Compared with the standard spectrometric detection, the poor detection accuracy with the biggest absolute error (AE) of 76.86 mg/kg was determined in the soil suspension group. The slope of regression equation was 2.27, which indicated a non-negligible deviation from the real condition. As expected, the ISE detection, conducted by the self-designed device, was much closer to the standard spectrometric. The adjusted determination coefficient (R2) of the regression equation was 0.87 and the equation slope was 1.31, which was less than 2.27 obviously. The average AE was decreased to be 20.97 mg/kg. The RMSE of 22.34 mg/kg was found. Meanwhile, the labor intensity was greatly reduced. However, the fast pretreatment group still exhibited the obvious difference with the conventional pretreatment group. In all, the device discussed in this study demonstrates a promising result, which might provide references for the ISE applications of in-situ soil macronutrient rapid detections. Further modification on the mechanical structure for extracting is required to enhance its performance.