How reliable are surface roughness measurements of micro-features? - Experiences of a Round Robin test within nine 4M laboratories
How reliable are surface roughness measurements of micro-features? - Experiences of a Round Robin test within nine 4M laboratories
复制标题
微观特征的表面粗糙度测量有多可靠？
DOI：
--
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
G. Tosello
中科院分区：
文献类型：
--
作者：
L. Mattsson;Pieter Bolt;S. Azcarate;Emmanuel Brousseaud;B. Fillon;Ronald Fowler;E. Gelink;C. Griffiths;C. Malek;S. Marson;A. Retolaza;A. Schneider;A. Schoth;Attila Temun;G. Tosello
摘要微小微机械加工特征的表面粗糙度不易验证。表面本身的统计变化可能是妨碍公差验证的限制因素。在本文中，我们研究了这种效果，我们还测试了10个不同的表面轮廓仪在一个非常好的指定表面积的性能。对于该区域，6个分析仪在± 6%的标准差窗口内产生相同的结果。对于其他领域，顶部的窄酒吧和在狭窄和深的渠道，一个更大的传播在循环结果被发现。关键词：表面粗糙度，微观计量，循环，表面轮廓仪，Ra，Pa 1。4 M多材料微型制造卓越网络已被证明是测试程序和揭示微型元件非标准化计量问题的绝佳论坛。在第一轮测试中，我们发现聚合物和计量部门的合作伙伴之间存在很大的表面粗糙度测量差异。此外，当对结果进行分析时，四个金属插入物的微脊的尺寸测量值远非完全重叠。第一次测试是相当开放的，没有向仪器操作员提供如何进行测量的详细说明，因为目的是了解不同实验室的测量实践，以及它是否足以提供可靠和准确的数据。结果为阴性。即使在一个简单的地面金属表面上，表面粗糙度值的分布也是惊人的，微结构的宽度和高度测量值也明显取决于测量的实验室。正如在关于微观计量挑战的综述论文中已经指出的那样[1]，在微小表面上评估表面粗糙度时会出现严重的问题，因为基于Ra，Rz，Rq，Rp，Rv参数的ISO标准化线轮廓的整个概念依赖于长迹线-默认情况下是采样长度的七倍，包括开始和停止长度。对于约1 µm的表面粗糙度，标准化采样长度为0.8 mm，这意味着需要访问5.6 mm的默认线轮廓。这远远不能接受，并且与许多微型元件的实际情况相矛盾。更新后的ISO标准4287 [2]和4288 [3]提供了指定基于主要轮廓的参数Pa、Pz、Pq、Pp、Pv值的机会，其中采样长度，即沿表面沿着测量的长度，可以设置为适合于被测量的表面特征的任意确定的长度，只要其与测量值一起声明即可。默认情况下，主轮廓仅通过2.5 µm的短波长截止进行过滤，即它不适用于亚微米特征粗糙度测定，如使用原子力显微镜进行测量。如果一个不熟练的操作员只是把样品放在测量台上，然后按下Ra测量按钮，那么很明显，仪器会得出完全错误的结果。因此，当要测量微米尺寸的特征时，绝对需要受过良好教育的仪器操作员。在对几个注塑模具嵌件进行了初步的计量循环赛之后，很明显，需要更严格的测量描述，并且本文中仅选择了一个在MEC卡迪夫铣削的嵌件进行计量循环赛。
Abstract Surface roughness of tiny micro machined features is not easy to verify. The statistical variation of the surface itself can be the limiting factor that hampers tolerance verification. In this paper we have studied this effect and we also test the performance of 10 different surface profilers over a very well specified surface area. For this area 6 profilers yielded the same result within a standard deviation window of ±6%. For other areas, on top of narrow bars and in narrow and deep channels, a much larger spread in the Round Robin results was found. Keywords: Surface roughness, Micro metrology, Round robin, Surface profiler, Ra, Pa 1. Introduction The 4 M Multi Material Micro Manufacture Network of Excellence has proven to be an excellent forum for testing procedures and throwing light on non-standardised metrology issues of micro components. In a first test round we found that large variations in surface roughness measurements existed among the partners in the Polymer and Metrology divisions. Also, the dimensional measurements of the micro ridges of four metal inserts were far from a complete overlap when the results were analysed. This first test was quite open and gave no detailed instructions to the instrument operator on how to perform the measurements, as the intention was to get an idea of the measurement praxis at the different laboratories and if it is good enough to provide reliable and accurate data. The result was negative. Even on a simple ground metal surface the spread in surface roughness values was astonishing, and the wi dth and height measures of the micro structures departed significantly depending on what lab did the measurement. As already pointed out in the review paper on micro metrology challenges[1] severe problems arise when surface roughness is to be assessed on tiny surfaces, as the entire concept of ISO standardised line profile based Ra, Rz, Rq, Rp, Rv parameters rely on long traces – by default seven times the sampling length including a start and a stop length. For a surface roughness of about 1 µm the standardised sampling length is 0.8 mm, which means a need to access a default line profile of 5.6 mm. This is far from acceptable and contradictory to the reality of many micro components. The updated ISO standards 4287 [2] and 4288 [3] give the opportunity to specify primary profile based parameters Pa, Pz, Pq, Pp, Pv – values instead, where sampling length, i.e. the measured length along the surface, can be set to an arbitrarily determined length adapted to t he surface feature being measured, as long as it is stated with the measured values. The primary profile is by default only filtered by the short wavelength cut off at 2.5 µm, i.e. it is not suitable for sub-micron feature roughness determination like measurements with atomic force microscopes. If a non-skilled operator just put the sample on the measurement table and press the Ra measurement button the risk is obvious that totally wrong results will come out of the instrument. It is therefore absolutely necessary to have well educ ated instrument operators when micrometer sized features are to be measured. After the initial metrology Round Robin on several injection mold inserts, it wa s obvious that a much more strict measurement description was needed and only one insert, milled at MEC Cardiff, was chosen for the Metrology Round Robin presented in this paper.