【作者】 吕品；

【导师】 唐祯安；

【作者基本信息】 大连理工大学 ， 微电子学与固体电子学， 2008， 博士

【摘要】 当前,室内空气质量问题已成为全球关注的热点。改善室内空气质量的首要前提是对室内空气污染物的检测,主要包括甲醛（HCHO）、苯系物、总挥发性有机物、氡气、石材放射、氨的检测等,其中对室内首要污染物——甲醛的检测更成为重中之重。现如今控制室内空气质量问题的主要途径是在空调上安装室内空气检测、净化装置,但空调散热器铝箔本身可能散发的难闻气味也会反过来污染室内空气质量。实现对室内甲醛和空调散热器铝箔散发气味的实时、准确的现场检测和识别是控制室内空气质量的两个关键检测技术。本文针对室内甲醛、空调散热器铝箔气味的检测,主要从甲醛敏感材料的制备、甲醛传感器性能、基于传感器阵列与BP神经网络的电子鼻对二元混合气体中0.06ppm甲醛的识别、空调散热器铝箔气味检测系统等几方面进行了系统的研究。用化学共沉淀法制备了SnO₂-NiO粉体,各种表征分析表明:该粉体属于纳米级,依据NiO的掺杂浓度和粉体煅烧温度的不同其晶粒尺寸在11nm～39nm范围内变化。低浓度NiO的掺杂使金属氧化物表面的吸附氧数量增加,表面活性位相应增加,还原性气体甲醛与吸附氧的反应加快,大大提高了对甲醛气体的气敏性能。低浓度的NiO掺杂也有效地抑制了SnO₂晶粒的生长,晶粒尺寸减小,比表面积增大,有利于对氧的吸附和气体表面反应。煅烧温度越高晶粒尺寸越大。NiO低浓度掺杂时XRD中只出现了SnO₂衍射峰,没有任何Ni物种的衍射峰出现,低浓度的NiO可能被隔离在SnO₂的表面。当纳米粉体中Sn与Ni的原子比达到4:5,经600℃煅烧后,XRD中出现了NiO衍射峰,体系发生了两相分离。调和SnO₂-NiO粉体涂覆在微热板敏感区经退火后制成甲醛传感器,用自行搭建的气体传感器自动测试系统对甲醛传感器的测试表明:低浓度NiO掺杂的SnO₂-NiO甲醛传感器的电特性行为表现为n型。该传感器对痕量甲醛响应灵敏、选择性好、稳定性好。最佳工作温度为300℃,在300℃下加热功率为180mW。其响应时间和恢复时间随着甲醛气体的浓度不同而变化。环境温度对传感器影响较大;相对湿度超过30%时对甲醛传感器的影响很小。空气中O₂的浓度对0.06ppmHCHO及其干扰气的响应几乎没有影响。基于气体传感器阵列与BP神经网络的电子鼻实现了对二元混合气中0.06ppmHCHO和低浓度干扰气体的定量识别。传感器阵列由所研制的甲醛和其他几种掺杂不同贵金属的SnO₂薄膜传感器构成。用主成分分析法对传感器阵列信号进行压缩降维后,构建了BP神经网络对0.06ppmHCHO进行了有效识别,结果表明:对单一成分的0.06ppmHCHO的识别率为88.8%,而在乙醇、甲苯、α-派烯、VOCsmixture等干扰气体存在时,对0.06ppmHCHO的识别率分别为92%、89.3%、90.0%和96.7%。该识别率可与文献报道的电子鼻对不同种类、不同浓度的气体的识别率相比拟。电子鼻能有效地识别二元混合气体中0.06ppmHCHO。针对空调散热器铝箔散发的气味研发了空调散热器铝箔气味检测系统。该系统具有重复性,可在现场实时检测。设定的阈值随环境温湿度的变化而变化。传感器阵列由所研制的甲醛传感器和其他4只商售传感器构成。采用sum of deltV数据处理方法和神经网络相结合的两级数据处理模式。实测数据表明:当sum of deltV的值与设定阈值的差值的绝对值大于0.1时,可只用sum of deltV数据处理方法对空调散热器铝箔散发的气味直接进行判定;当sum of deltV的值与设定的阈值很接近时,即两者的差值的绝对值小于0.1时,需用神经网络做进一步的判定。Sum of deltV数据处理方法和神经网络相结合的判定结果与气味专家的判定结果完全吻合。该检测系统已被国际某知名大公司用于生产线的产品质量检测。更多还原

【Abstract】 At present,Indoor air quality（IAQ） problems have become the focal point of worldwide attention.It is an important prerequisite to detect indoor air pollutants for improving IAQ, which include formaldehyde（HCHO）,benzene series,TVOC,radon gas,radiation,ammonia, et al.The detection of HCHO as the first pollutant is the most important.The main means of controlling IAQ is to install indoor air detection and purification unit on air-condition now. But bad odor from air-condition radiator aluminum plates will severely harm people health. Exact detecting and recognizing indoor HCHO and the odor from air-condition radiator aluminum plates in real time on the spot are two key detection technologies to control IAQ. The fabrication of material which is sensitive to HCHO,properties of HCHO sensor, electronic nose based on gas sensor array and BP neural network to recognize 0.06ppm HCHO in binary mixtures and system developed to detect the odor from air-condition radiator aluminum plates are systematically researched in the paper.SnO₂-NiO powders fabricated by chemical coprecipitation method belong to nanometer levels.The grain size of the powders changes in the range of 11 nm to 39nm according to the difference of NiO-doped concentration and the calcining temperature.Low concentration doped NiO greatly increases the quantity of adsorption oxygen and surface reactive sites.The reaction of HCHO gas and adsorbed oxygen speeds up to greatly improve the sensitive properties to HCHO gas.Low concentration doped NiO can effectively inhibit the growth of SnO₂ grains to decrease the grain size and increase specific surface area,which is beneficial to the oxygen adsorbing and the surface reaction between HCHO and adsorbed oxygen.The higher the calcining temperature of the powders,the larger the grain size.The SnO₂ peaks and no peaks of Ni species are observed in X-ray diffraction（XRD） pattern when NiO is doped at low concentration,because low concentration NiO is segregated on the surface of SnO₂.NiO peaks can be observed in XRD pattern of SnO₂-NiO powders calcined at 600℃at an atomic ratio of Sn:Ni=4:5 because the system takes place two-phase separation.The HCHO sensor is fabricated after annealing the MHP whose sensitive region the homogenized SnO₂-NiO powders are coated onto.The gas sensor characterization system is employed to test the HCHO sensor properties.The SnO₂-NiO HCHO sensor shows the electronic property behavior of n-type semiconductor.The HCHO sensor shows high response to low concentration HCHO,good stability and selectivity.The optimized working temperature is at 300℃and the power is about 180mW at 300℃.The response and recovery time change with the HCHO concentration.The environmental temperature has influence on the HCHO sensor and the relative humidity beyond 30%almost has no influence.The O₂ concentration in the atmosphere almost has no influence on the response to 0.06ppmHCHO and its interference gas.The quantitative recognizing of 0.06ppmHCHO and its interference gas in binary mixtures is realized by employing electronic nose technology based on the gas sensor array and back propagation（BP） neural network（NN）.The gas sensor array is composed of above-mentioned HCHO sensor and several noble metal-doped SnO₂ thin-film sensors.The BP NN is constructed to recognize trace HCHO after the sensor array signals are compressed and reduced the dimensionality by PCA.The results show that the recognition rate to the single component HCHO is 88.8%and that to the HCHO in the presense of an interference gas,such as alcohol,toluene,α-pinene and VOCsmixture is 92%,89.3%,90.0%and 96.7%, respectively.The recognition rate is commensurable with that in literatures reported to recognize different kinds,different concentrations of gases by employing electronic nose. Electronic nose can effectively recognize 0.06ppmHCHO in binary mixtures.The odor detecting system is developed to aim at the odor from air-condition radiator aluminum plates.The system owns the repeatability and can detect in real time on the spot. The threshold value should be adjusted with the change of environmental temperature and humidity.The sensor array is composed of above-mentioned HCHO sensor and four commercial sensors.The sum of deltV data processing method combineds with NN is employed to process the sensor array signals.If |sum of deltV-threshold value| is bigger than 0.1,the sum of deltV data processing method can directly give the judgement to tested aluminum plates.If |sum of deltV-threshold value| is smaller than 0.1,the NN is used to further judge.The judgement results of the sum of deltV data processing method combineded with NN completely coincide with those of the odor specialist.The detection system has been used to detect the product quality in the production line by a internationally well-known Co.更多还原