基于实验大数据评价电子器件性能和制备的有效性

Method for Evaluating the Performance and Preparation Effectiveness of Electronic Devices Based on Experimental Big Data

在大量实验方案中快速找出成功率较高的优质方案不仅可以缩短实验时间,而且也能大幅降低实验成本。由于通过实验方式寻找最优方案的经济成本较高,而应用软件模拟的时间成本又很大;同时,在进行综合分析时,参数方法也存在指标权重确定困难、忽视指标个性信息等弱点。因此,以增强型AlGaN/GaN HEMT器件掺杂为例,给出一种快速寻找和筛选优质实验方案的数学模型。然后,给出一种综合评价器件优劣的非参数方法。通过实例分析表明:该方法在增强型AlGaN/GaN HEMT器件掺杂方案筛选和预估方面具有较好的效果和应用前景。对充分利用实验大数据来提高工作效率、预测技术前沿、有效降低制备成本等方面具有一定的指导意义。

It can not only shorten the experimental time,but also can greatly reduce the experimental cost if we can rapidly identify the scheme with high success rate in a large number of experimental schemes.Because the way by using experiments to find the optimal scheme will have a high economic cost,while the way by the simulation software will spend much time.At the same time,the parameter methods are difficult to determine the index weight and ignoring the index personality information.Therefore,taking the doping of enhanced AlGaN/GaN HEMT devices as an example,a mathematical model for quickly finding and screening high-quality experimental schemes is given.Then,a non-parametric method for comprehensive evaluation of devices is provided.At last,the example shows that above method has good effect and application prospect to select and predict the doping scheme of enhanced AlGaN/GaN HEMT devices.It has certain guiding significance for using the experimental big data to improve work efficiency,predict technology frontier and effectively reduce preparation cost.