Advanced Coatings in HIPIMS hybrid MF magnetron sputtering with the aid of EMICON closed loop feedback control

Recenly lots of requests on advanced coating configuration in HIPIMS magnetron sputtering: higher deposition rate and high coating layer quality are key demands. With HIPIMS magnetron sputtering, to have the highest thin film quality is for sure, however, the slow deposition rate is still a big concern comparing to MF magnetron sputtering. How to combine both advantages (high quality and high deposition rate) and how to perform the optimized coating layer in a reasonable and cost-effective configuration become a hot topic for R&D and industrials.

Magpuls Power supply can switch the output mode either MF or HIPIMS. And a special function to combine both in hybrid output waveform pattern via. a built-in PulsTrain waveform editor in the pulse unit. Recipes can be saved and recalled easily. For remote control, a full control over all parameters (time-settings, arc-settings, operating mode changing) to proceed the real-time online dynamic adjustments and controlling can be achieved via a LAN communication to a PC software which end user can design a special remote-control program running on the remote control PC. An EMICON system is the key technology to govern the film properties and quality. Through EMICON system’s setpoint for a closed loop PID control over the flow rate of MFC, the output voltage of a power supply, or even the pulse frequency of the pulse power supply, it’s very flexible to handle precisely the quality control of deposited layer in mechanical, optical and electrical properties. In the combination with EMICON (plasma emission monitor) online quality control and Magpuls stable output pulse patterned waveform, the plasma impedance can be kept very stable and the film growth has high quality in the best performances. Adding some end point conditions can enhance the power of online quality control in real time mode. Such an advanced coating configuration brings a new era for HIPIMS and MF hybrid coating to create new coating materials and applications.

1. Magpuls

2. Plasus

3. Applied Optivac Technology, Inc.

Homogeneity & Uniformity


Usually the definitions of two important quality factors – homogeneity and uniformity – are easily mixed up in a deposition process no matter which coating technologies are involved.  Note that many articles, papers or technical notes made the same confusion.  So, the definitions must be clear.

Homogeneity generally considers a very local spot area and measure/check the layer properties along the direction of the layer grown.

Uniformity considers the differences or tolerances among several measured data along one axis or in a big XY-plane area.

PEM system normally takes the responsibility of layer’s homogeneity control.

Uniformity can be achieved if the process chamber has a good layout or configuration for reactive plasma process.



To achieve a good homogeneity layer, PEM system was introduced to reactive plasma processês to ensure the composition fraction (or alloy fraction) in each compound or molecules formed by the reactive plasma process is nearly constant.

In general, a PEM system is able to handle magnetron reactive sputtering in a distance range 500mm-600mm along the long axis of magnetron for a good PID closed loop control under the operating pressure range: approx. 0.8mTorr~10mTorr. In this example, a SiOx layer is grown by a dual magnetron sputtering sources with a process control by a PEM system.  PEM system varies the oxygen flow quickly to fit the setpoint by a fast response PID calculation to give a feedback voltage to control the oxygen flow rate. A good PID closed loop control can bring the alloy fraction factor x in each grown SiOx layer has an almost constant x value in different growth time, for example:  T1, T2, T3 and T4.
This is what PEM system contributes to get a high homogeneity layer.


Uniformity is a very important factor to be concerned and it’s very complicated because there are many factors that can influence it.  Generally, process chamber’s configuration is most critical to get a good uniformity of the grown layer.

This example has 3 measuring points over the width 1300mm of the PET film to compare the thickness, refractive index and other data corresponding to the uniformity in 3 locations in the same XY-plane on the surface of PET film.

The factors often influencing the uniformity include:

-Pumping: if the pumping ports can not provide good pumping speed and the gas distribution inside the process chamber is not good, the uniformity would be influenced.
-Gas piping: if the gas piping delivering the reactive gas can not deliver the gas into the reactive plasma zone in the shortest time evenly, the uniformity would be influenced.
-Magnetrons: if magnetic constraint for electrons is not uniform, the uniformity would be influenced.



What’s the difference between the intensity listed in the database and the actual measurement?

This is an interesting question asked often while one touches and senses the plasma by optical emission spectrometry.


The radiation we are measuring with the EmiCon OES system is due to a two-step process in the plasma:

1. Ground state atoms are excited by electron collisions in higher electronic levels. The density of the excited atoms in a specific electronic level depends on the energy gap between the ground state (0 eV) and the higher electronic level and the electron temperature (i.e. electron energy distribution function EEDF, often Maxwelliam). The larger the gap the less is the density of the atom in this specific energy level.

2. From the higher electronic level the atom is decaying by spontaneous emission to some lower electronic level. This is the radiation we are measuring. The intensity given in the SpecLine database is the probability that the atom in the specific higher electron level will decay into the given lower electronic level.

This means for the Hg spectrum:
A.The lines at 404 nm, 435 nm and 546 nm are all decaying from the electronic level 7s2S with energy 7.73 eV. These there lines show an intensity distribution as given by the intensity values of the SpecLine database.
B. The line at 365 nm is decaying form the electronic level 6p2P° with energy 8.85 eV and the line 579 nm is decaying from the electronic level 6p1P° with energy 8.84 eV. Allthough these lines have a higher probability to decay than the lines form the 7s2S level, the denstiy of atoms in the 6p levels is much smaller due to the higher energy of the excited level (8.85 eV and 7.73 eV for the 7s level).

In general, the intensity of a line is ruled first by the excitation process (i.e. the energy of the higher electronic level) and second by the probabilty of decaying form the higher level to a lower level.

The remaining differences in the intensities of the lines is due to the sensitivity distribution of the EmiCon system, i.e. the system is most sensitive between 450 and 550 nm and less sensitive below 400 nm.

~~~ supported by Dr. Thomas Schütte of PLASUS ~~~

新一代的OES (EMICON)在等离子工艺应用上的重要性

回顾等离子工艺使用OES (Optical Emission Spectroscopy)的历史,早在1980年代,已有不少大型的玻璃工业尝试使用OES来协助等离子工艺在量化生产的稳定性,同时,也利用OES的特殊性能开发等离子工艺能够使用的新材料。1990年初期,OES著名的代表产品称为PEM (Plasma Emission Monitor)正式在生产大楼帷幕玻璃的等离子溅射流水线上使用,从single Low-e的能源玻璃进化到更具节能效果的double Low-e规范,到了1990年末期,更应用在平面显示器的镀膜工艺上。在2000年以前发展的PEM,多半采用 窄通滤光器 NBPF (Narrow Band-Passed Filter)搭配一个极为灵敏的 光电倍增管 PMT (Photo Multiplier Tube),透过一套单晶片的处理器计算侦测讯号强度与内部设定强度的差异,将差异值输出到控制反应气体的流量单元: 早期使用的多半是一个 流量计 (flow meter)搭载一颗 压电陶瓷阀 PZT  (Lead zirconate titanate: Pb[ZrxTi1-x]O3 0≤x≤1) valve,反应气体的流量经过变化,再次侦测等离子的强度,不断用回路的方式进行修正,让实际侦测的强度趋近内部设定值。为了稳定PEM的控制,PID计算的控制器也成为PEM Closed Loop Control (PEM闭锁回路控制)中很重要的一个角色。由于等离子工艺进步迅速,加上应用市场快速开发,对于镀膜的质量要求越来越严苛。不仅镀膜的质量在同质性(homogeniety: 镀膜期间随著厚度增加时的质量均匀性)的要求增高,同时对镀膜速率的提升也有很大的期许。

SpecLine分析等离子光谱的结果 (感谢德国PLASUS公司提供图片)

因此,2000年开始,新一代的PEM (EMICON: EMISSION CONTROLLER) 不再使用频宽不够精密的窄通滤光器,改用线性的电荷耦合器件CCD,大幅提升光谱的分辨率(resolution),不但谱线的位置准确,也因为光谱的分辨率提升,把主要用来监控的原子谱线与旁边夹杂的其他杂讯可以轻易地分离,让控制的范围加大且大幅改善控制的精度。这十年在分光仪的工艺上有长足的进步,不仅光谱的分辨率越来越好,也改善CCD的感光度,这些进步的优点,也替新一代的PEM EMICON打开更多应用的市场。虽然新一代的PEM EMICON在价格上较旧型的PEM贵了许多,但不可抹灭的,新一代的PEM EMICON提供更多更精准的性能让生产与研发有更好的信赖度与发展空间。长期使用的成本估算却是远较旧型要节省不少,最重要的是能够让生产线稳定,增加产能也增加营收。

现今,热门的应用如:大气等离子、大楼帷幕玻璃、平面显示器的生产、触摸屏生产、微机电、太阳能电池、半导体、装饰镀膜、超硬膜与光学镀膜都必须搭載新一代的PEM EMICON,来保障生产的稳定与信赖。

以下是新一代PEM EMICON在等离子工艺中不可或缺的几项重要特徵:

  1. 在过渡区域(hysteresis region)能快速有效地稳定在设定点
  2. 可提高镀膜速率(deposition rate)
  3. 可做线上的质量管理(online QC)

对于等离子工艺系统(或设备)商而言,新一代的PEM EMICON不但能够提升等离子工艺研发的能力,更可以建立自我的等离子工艺资料库,大幅缩短在客户端装机验收的时间,也能够提供客户快速的检修服务。特别是溅射的流水线(inline sputter coater)与批量型的生产机台(batch type),十分适用。除了等离子工艺的研发,同时也兼具等离子工艺系统(或设备)的除错功能,让设备商有能力自我改善设备的设计,让性能与稳定性提升。



  1. 等离子工艺监控器 (plasma monitor and process controller)
  2. 脉冲式直流电源 (pulsed DC power supply) 或是 中频交流电源 (MF AC power supply)
  3. 磁控溅射靶 (magnetron sputtering source) 或称 阴极靶 (cathode source)


对反应气体(reactive gas)采用高速的PID闭锁回路控制(PID closed loop control),将检测到的等离子信号强度比对内部设定的预设值,根据比对的结果,转成电压信号将差异值输出到负责供应气体的质流量计(MFC: mass flow controller),改变供应气体的流量来修正实际等离子体实际反应的状态。再撷取改变后的等离子体的信号强度,又经比对产生修正的电压信号,重复修正气体流量来变化等离子体的行为表现。如此周而复始,一直修正到实测的等离子体信号与内订的预设值接近,甚至相同。在反应式的溅射镀膜期间,需要能够快速反应改变气体的流量,主要目的在于稳定溅射出来的金属靶材原子或分子与反应气体之间能够维持在一个固定的比例(合金比例)。如果,变化反应气体的速度太慢,合金比例只能在饱和区维持,且镀膜速率将会很慢。要能够提高镀膜速率又能改变合金比例且维持稳定的镀膜状态,这个等离子工艺监控器扮演极为重要且关键的角色。

PID Closed Loop Contorl

Continue reading “反应式等离子溅射镀膜工艺的三宝”

另外一個迷思: PEM的閉鎖迴路控制要用MFC或是PZT valve


到底要用MFC (Mass Flow Controller)就能夠擔負製程的重責大任? 還是需要使用PZT valve?
1. 感測頭: 將電漿光譜的光線收集傳遞給偵測器。這通常是由光學鏡頭與光纖組成的硬體,將光訊號傳遞到偵測器做光電訊號的轉換。因為是光速在傳遞訊號,所以需要的時間可以忽略。
2. 偵測器: 有感度十分靈敏的光電倍增管PMT (Photo-Multiplier Tube),通常轉換時間小於1 ns; 另外一種是CCD偵測器,轉換時間大約需要1~2 ms。
3. CPU / MPU軟體處理: 光訊號轉成電的訊號,經由電腦軟體的計算與處理產生可做控制的訊號輸出,需要的時間大約5-10 ms。
4. MFC或是PZT Valve: 控制反應性氣體流量的閥門從接收到電腦傳來的控制訊號到調整閥門開啟的大小到達指定的位置所需的時間,好的MFC最快可以在幾個ms達成,PZT Valve可以在小於1 ms內達成。
5. 氣體管路: 氣體經過氣閥後需要流經氣體管路到達真空噴出的位置,這段時間視真空鍍膜系統的真空抽氣設計來決定長短,通常需要20 ms(系統有很好的真空抽氣設計)到200 ms不等。
6. 擴散: 氣體抵達真空內部的噴嘴位置,會依照當時真空度的製程條件以擴散的方式離開噴嘴到達電漿製程的區域,這段時間通常小於1 ms.



到底要用MFC (Mass Flow Controller)就能够担负工艺的重责大任? 还是需要使用PZT valve?

1. 传感器: 将等离子体光谱的光线收集传递给传感器。这通常是由光学镜头与光纤组成的硬件,将光信号传递到传感器做光电信号的转换。因为是光速在传递信号,所以需要的时间可以忽略。
2. 传感器: 有感度十分灵敏的光电倍增管PMT (Photo-Multiplier Tube),通常转换时间小于1 ns; 另外一种是CCD传感器,转换时间大约需要1~2 ms。
3. CPU / MPU软件处理: 光信号转成电的信号,经由计算机软件的计算与处理产生可做控制的信号输出,需要的时间大约5-10 ms。
4. MFC或是PZT Valve: 控制反应性气体流量的阀门从接收到计算机传来的控制信号到调整阀门开启的大小到达指定的位置所需的时间,好的MFC最快可以在几个ms达成,PZT Valve可以在小于1 ms内达成。
5. 气体管路: 气体经过气阀后需要流经气体管路到达真空喷出的位置,这段时间视真空镀膜系统的真空抽气设计来决定长短,通常需要20 ms(系统有很好的真空抽气设计)到200 ms不等。
6. 扩散: 气体抵达真空内部的喷嘴位置,会依照当时真空度的工艺条件以扩散的方式离开喷嘴到达等离子体工艺的区域,这段时间通常小于1 ms.




經常有人認為在真空電漿濺鍍系統上安裝了PEM (Plasma Emission Monitor)或是Emission Controller就可以改善鍍膜的均勻性。把PEM當作神看待,卻不知道這是個大錯特錯的觀念。PEM系統能夠控制鍍膜製程的穩定性,也就是鍍在待鍍物上的膜層的品質是穩定的,在成膜的方向上穩定地把相同品質的薄膜一直鍍到待鍍物上。這是PEM最大的效用,卻無法因為安裝使用了PEM而改善鍍膜厚度在待鍍物的面積分布上的均勻性。



经常有人认为在真空电浆溅镀系统上安装了PEM (Plasma Emission Monitor)或是Emission Controller就可以改善镀膜的均匀性。把PEM当作神看待,却不知道这是个大错特错的观念。PEM系统能够控制镀膜制程的稳定性,也就是镀在待镀物上的膜层的质量是稳定的,在成膜的方向上稳定地把相同质量的薄膜一直镀到待镀物上。这是PEM最大的效用,却无法因为安装使用了PEM而改善镀膜厚度在待镀物的面积分布上的均匀性。


EMICON System的工作原理介紹




為了研究存在電漿中的各種不同帶電離子與分子的成分與分布,在不影響電漿反應機制的條件下,還能採用回綬控制的方式來穩定製程,光學偵測法是不接觸電漿也不影響電漿作用的唯一選擇,其中能夠完全表現電漿特性的就是採用分光光譜學的量測技術加上特殊設計的回綬控制機制來達到分析電漿物種與穩定製程的雙重目的。其中以OES (Optical Emission Spectroscopy)光放射光譜學的技術最為成熟。為了達到回綬控制的目的,電漿放射出來的光線經由一個準直鏡光學鏡頭焦聚到石英光纖,傳送到分光儀把電漿的光線用光柵分光到一個陣列式的光偶合感測器(Arrayed CCD),每一個感測器上的獨立感測單元代表一個分光後的特定波長,整個陣列分佈把光的資訊轉化成有用的電壓資訊,電腦透過USB電纜取得這一瞬間電漿光線轉化成陣列對應電壓的光譜資訊,再由軟體針對選定的特殊譜線資料做進一步的PID回綬控制。





为了研究存在电浆中的各种不同带电离子与分子的成分与分布,在不影响电浆反应机制的条件下,还能采用回绶控制的方式来稳定制程,光学侦测法是不接触电浆也不影响电浆作用的唯一选择,其中能够完全表现电浆特性的就是采用分光光谱学的量测技术加上特殊设计的回绶控制机制来达到分析电浆物种与稳定制程的双重目的。其中以OES (Optical Emission Spectroscopy)光放射光谱学的技术最为成熟。为了达到回绶控制的目的,电浆放射出来的光线经由一个准直镜光学镜头焦聚到石英光纤,传送到分光仪把电浆的光线用光栅分光到一个数组式的光偶合传感器(Arrayed CCD),每一个传感器上的独立感测单元代表一个分光后的特定波长,整个数组分布把光的信息转化成有用的电压信息,计算机透过USB电缆取得这一瞬间电浆光线转化成数组对应电压的光谱信息,再由软件针对选定的特殊谱线数据做进一步的PID回绶控制。


Improvement of the response time for a fast PID closed loop control

From time to time, there exists a critical issue to handle multi-channel OES system for a reactive sputtering process which demands high response time of the reactive gas supply. For example, a 3-channel PEM system shown in the figure below has three spectrometers in the PEM controller. Each spectrometer’s configuration has its own recording interval (=exposure time multiplies average number), which might be different from others due to slight differences in the alignments of collimators and the plasma density are different in each gas supply section. For the individual control of each channel, the recording intervals are different, but the spectra data acquisition from spectrometers to the computer’s host software needs to collect all three channels sequentially at a time. That means, if the recording interval settings for 3 channels are 100ms, 125ms, 150ms individually(for example), then the complete system recording interval for a closed loop control is the sum of three individual settings, i.e. 375ms (=100ms + 125ms + 150ms). This result is too bad for a fast PID closed loop control and such kind of control can not provide the highest quality of the film deposited on the substrate. Therefore, the new developed EMICON xMC series is able to acquire spectra data for each channel individually without waiting the others. In this way, each gas supply section can work properly in their optimized configuration to maintain the best quality of the film deposited. The high speed PID closed loop controlled process is achievable finally.

Of course, this technology can be used for any kind of applications with the demands on different settings for the recording intervals, but the response time is demanded to be very high. Now, only Plasus EMICON MC/HR series can do it.

如何應用OES tool與RGA tool


常有人問道: 對需要精準控制的電漿(等離子)製程(工藝)中,到底要使用OES還是RGA呢?

  • RGA能看到環境的氣體成分比例與變化,但是無法看到實際參與電漿製程(等離子工藝)的物種與變化,OES卻可以。這是最大的差異。
  • 其次,RGA的反應遲緩,OES可以快速反應。 因此,RGA對於製程(工藝)上出了狀況是很難追蹤與了解實際電漿製程(等離子工藝)的變化。只有使用OES才有機會知道製程(工藝)中真實發生的情形,進而可以找到解決之道。
  • RGA可以當做OES的輔助,兩者相輔相成。

結論: OES掌握製程(工藝)的穩定性,RGA去處理整體環境的穩定性。


常有人问道: 对需要精准控制的电浆(等离子)制程(工艺)中,到底要使用OES还是RGA呢?

  • RGA能看到环境的气体成分比例与变化,但是无法看到实际参与电浆制程(等离子工艺)的物种与变化,OES却可以。这是最大的差异。
  • 其次,RGA的反应迟缓,OES可以快速反应。 因此,RGA对于制程(工艺)上出了状况是很难追踪与了解实际电浆制程(等离子工艺)的变化。只有使用OES才有机会知道制程(工艺)中真实发生的情形,进而可以找到解决之道。
  • RGA可以当做OES的辅助,两者相辅相成。

结论: OES掌握制程(工艺)的稳定性,RGA去处理整体环境的稳定性。