EESemi.com Forum Archives

EDX Dead Time and Input Rate

               

EDX analysis is one failure analysis technique that takes time and experience to master.  Used for identifying the elemental composition of a sample, EDX analysis is performed on a SEM that is equipped with an EDX system. An EDX system operates by counting X-ray emissions from a sample according to their energy content.  The X-ray counts are plotted in histogram format (with the energy levels on the x-axis), forming an EDX spectrum. The EDX specturm formed contains peaks that correspond to the elements present in the sample.

   

One challenge in operating an EDX system is ensuring that enough X-ray emissions are counted and plotted to come up with a valid EDX spectrum.  "Dead time", or the period wherein no emission counts are made, must not be excessive.  Excessive dead time means that not enough counts are being registered, causing an EDX spectrum with no peaks. The "standard" for the correct 'dead time' versus count rate varies from one EDX system to another, so it is important for a failure analyst to be familiar with his or her system and its calibration standards. The archived forum thread below reflects the challenges in using an EDX system, as well as the differences among various EDX systems used in the industry.

  

Posted by Longier: Sat Aug 19, 2006 12:31 pm    Post subject: EDX question about deadtime and input rate?

 

Using the Oxford/Hitachi and Horriba/Hitachi EDX system, I have the following questions:

1. The relation between the input rate and deadtime? The Horriba service engineer tells me that we should ensure the input rate no less than 5kcps and the deadtime to be about 30% when performing the microanaylsis , however, Oxford engineer give some advice that should promise the the input rate to be approximately 3 kcps and the deadtime maybe less than 10% in this case. in fact, the Horriba system and Oxford system have the same hardware and software.

2. Generally, I use 20KV acceleration voltage to do EDS, to get the most reliable EDS results, how to balance the input ratemeter and deadtime?

Posted by FARel Engr: Mon Aug 21, 2006 6:17 pm    Post subject:

 

It's really a balancing act to avoid getting too many xray counts (which may indicate that some 'invalid' xrays are being counted and can cause spectrum distortion) or getting too little of them (which can result in an erroneous spectrum if the 'right' xrays are not being counted). To get the dead time that we want, we try to optimize the probe current and the spot size, assuming that everything in the hardware is perfectly aligned and working.

Posted by Longier: Mon Aug 21, 2006 10:59 pm    Post subject:

 

Very thanks for your reply. Would you tell me how to optimize the probe current and the spot size to get the satisfied deadtime? do you have some experience or standards sharing with me?

Posted by FARel Engr: Tue Aug 22, 2006 11:17 am    Post subject:

 

Sorry, but no standards for these inputs can be defined since xray emission and counting vary from one specimen to another. That's the reason why the standardization was defined in terms of the output, i.e., dead time and counts per sec. Good luck!

Posted by Cvill: Thu Feb 01, 2007 2:15 pm    Post subject:

 

How different is the Horriba, Oxford and Kevex?

For Kevex, deadtime and cps can be adjusted using the fast discriminators in the pulse processor setup. However, it doesn't help me much because if I go for 1 to 2kcps, the deadtime shoots up to 100% and sometimes give an "No x-ray detected" message.

What am I doing wrong? Does it have to do that I'm trying a 5kV analysis? We're doing some sulfur contamination evaluation and this was required by our customer. Is there a better way to determine sulfur?

Posted by FARel Engr: Fri Feb 02, 2007 7:33 pm    Post subject:

 

Try to get a set-up that achieves about 30% dead time only. If you've done everything and still can't get any x-ray counts, then 5 kV may indeed be too low for what you are trying to detect.

What is the lowest K-alpha peak of sulfur - do you remember? If I recall it right, it's somewhere around 2.3 kV right? If so, then 5 kV should be enough to detect it if it's present.

What is the specimen you're trying to analyze? If it's a bond pad, then 5 kV should have no problems detecting Al, so Al should at least register in the spectrum. If Al can't be detected at 5 kV, then there's something wrong.

I suggest that you check your EDX with a cal standard, just to make sure that the hardware is really OK.

Posted by Longier: Thu Mar 27, 2008 6:26 pm    Post subject:

 

Now I use the Germany Bruker-AXS company's SDD XFlash Detector 4010 (No Liquid Nitrogen EDS system), this EDS can reach energy resolution 125eV (less than 133 eV), however, the conventional LN2 EDS can just reach 133 or 137eV and the 125eV can be specified at 100,000 cps.

The deadtime is less than 10% when the input rate arrived 10,000cps, the performing time is also 10 times faster than LN2 EDS system. No need for anybody to add LN2 or maintenance is the most improtant and convenient.

      

Back to the 'Best of Forums' Archives

      

HOME

 

      

Copyright © 2008 EESemi.com. All Rights Reserved.