the split depends of the moment when the transmitted or reflected beam reaches his initial value; maybe can you investigate on this way?

Best regards

The phenomenon you observed  is common when using a temperature protocol with a high peak inert mode temperature (e.g., the NIOSH protocol in which the highest temperature during the He mode is about 850), and is usually termed premature evolution of EC. To my experience, if you want to avoid the premature evolution of EC, you should  use a protocol similar to IMPROVE (i.e., you need to reduce the peak inert mode temperature to about 580). However, I don’t  think you really have to do so. Just mention the analytical method clearly when reporting the OC&EC results.  

Yes Mr Cheng, in my lab we use the Improve Protocol with a DRI Carbon anlayser, which takes in count the charring correction

Dear Yuan,

Thanks for your helpful comments. My other question is that which of the reported values by the instrument are reliable, when the "premature evolution of EC" exists?  In other words, are the values of thermal OC and its fractions (i.e. OC1, OC2, OC3, OC4) reliable when I observe a "premature evolution of EC" in my analysis?

Thermal OC and its fractions are reliable and could also be meaningful, regardless when the OC&EC split occurs (i.e., either before or after the introduction of O2). For example, some investigators used OC and EC fractions in combination with other components in PM for source apportionment studies. If you only want to get OC and EC concentrations, please use  the values given in the TXT file. As for the Sunset carbon analyzer, EC is defined as the carbon evolved off the filter after the transmittance signal returns to its initial value, regardless the temperature protocol used and when the OC&EC split occurs.

Regarding to the premature OC/EC split issue, one possibility is due to the presence of metal salts in the samples. Metal salts can reduce EC evolved temperature and enhance charring of OC (Wang et al., 2010). So EC loss is more pronounced in NIOSH type protocols(NIOSH 5040, STN, ACE-Asia etc) since it have higher inert mode temperature than IMPROVE (800~900 vs. 550). On the other hand, lower inert mode temperature can reduce early split problem, but may also cause OC not fully evolved. Since OC and EC are operational defined due to the lack of standard materials, it's quite challenging to say which protocol is more "correct" in determining OC and EC.

As mentioned by Dr. Cheng, the results of carbon fraction did not rely on OC/EC split. If your are using carbon fractions for source apportionment, IMPROVE is better than NIOSH. Because the retention time of IMPROVE is driven by FID signal while retention time of NIOSH is fixed. Carbon peaks in IMPROVE are well resolved thus have higher reproducibility than NIOSH. 

Reference:

Wang, Y., Chung, A., and Paulson, S. E.: The effect of metal salts on quantification of elemental and organic carbon in diesel exhaust particles using thermal-optical evolved gas analysis, Atmos. Chem. Phys., 10, 11447-11457, doi:10.5194/acp-10-11447-2010, 2010.

The catalytic effect of metals was already the reason to use manganese in  two step oxidation procedure decades ago

Thanks Harry for your comment.

Dear Wu and Yuan,

Your comments have been very helpful and informative. I am currently performing some analyses with lower inert mode temp to see if that fixes my issue.

Hi Mr Hasheminassab, OC/EC was "operational defined", which means different method results in different OC/EC. Pre-oxygen EC_OC is only a phenomenon, it is method charater. before explain, you need to check purity of inert environment, high purity Helium can have several ppm impurities. maybe oxygen, so oxygen absorber is needed for helium gas cylinder.

For usual research, we use common protocal(fixed temprature profile and laser method TOT/TOR), then we can compare results with other groups. Only if you hope to develop own method, change & develop temprature method.

For RT-4, there are several other aspects we need to be careful.

1 carbon denuder

VOC can be absorbed on quartz filter, and the carbon denuder can absorb VOC to decrease VOC's influence on OC1 results. You metioned high OC1, maybe carbon denuder should be changed, normally we should change it at least every year or less.

2 3-peak check.

Did you do this? this check is important for data quality. sunset lab use this protocal to check instrument, but little user know this. it tests back oven efficiency with CH4 standards gas. attached figure show an abnormal RT-4, which underestimate OC about 20%. if instrument is OK, the 3 peaks should equal with each other.

the attached figure is 3 peak check, peak in OC stage is 20% lower than EC or intercalib one,  showing mulfunction of back oven. Maybe the catalyst (MnO2) maybe should be changed.

Same thing, now I am experiencing a severe pre-oxygen EC_OC split. Attched is the result of a diesel sample. I am considering about changing the valve of oxygen.  Does anyone have any suggestions?  

I think same thing but we haven't enought experiences, the laboratory own the Sunset device for a few months. Guoliang Li suggests an oxygen leaek, why not, I asked myself a very similar question about the real level of oxygen during the helium period. To test the Sunset and determine the best protocole to analyse EC part in diesel particules, I try to generate known concentrations of graphite nanoparticules mixed with OC in order to obtain "samples reference". I hope that this work will yield results.

Here is follow up of my OCEC.

The problem is oxygen leakage, which result in the untable voltage of the flow sensor controlling oxygen. Thus, in the first Helium stage, there is indeed some oxygen going inside with helium (the flow sensor controls the flow by voltage, but when there is leakage, it's hard for him to identify how much oxygen you want or even whether you want it  through). I reached this conclusion from the check of the volts of every flow sensor as the attached pic. 

So I suggest that check the leakage first , and make sure the purity of Helium is UHP.