Silicon is probably one of the most difficult elements to measure by ICP-MS, mainly because all the glassware you might use and the glass sample introduction system is usually made of glass.  I would suggest you use plastic volumetrics and beakers etc and replace the ICP-MS sample introduction system with one that isn't made from glass.  There are HF resistant spray chambers available and nebulisers made from plastics.  If this isn't possible the minimum would be to try to prepare things in non-glass containers.  Good luck

Silicon is probably one of the most difficult elements to measure by ICP-MS, mainly because all the glassware you might use and the glass sample introduction system is usually made of glass.  I would suggest you use plastic volumetrics and beakers etc and replace the ICP-MS sample introduction system with one that isn't made from glass.  There are HF resistant spray chambers available and nebulisers made from plastics.  If this isn't possible the minimum would be to try to prepare things in non-glass containers.  Good luck

My fellow Harrington provides some good suggestions above, borosilicate glass is definitely a major source of contamination for Si.

While this you're probably already doing this, it's also better to perform the analysis in collision cell mode, as there are a good number of interferences with ubiquitous elements on all of the isotopes of Si (N2 on 28, COH on 29, NO on 30, a number of divalents on both although CCT mode isn't as effective for addressing that and could be addressed by tuning for low divalents, etc) that would make the background appear high for Si when it's actually the interferents. Both using CCT mode and tuning for low divalents and low oxides may help if you're not already doing it.

Thanks a lot  to Chris and James Harington for their valuable good suggestions.

I would assume that the greatest fraction of your blank signal is due to spectral interferences rather than Si contamination. As James pointed out N2+, CO+ (especially from organic solvents) are major problems when trying to analyze Si with ICPMS. In your case the Naphta will certainly lead to a high abundance of CO, COH etc. which interfere with all Si-isotopes. If available I would suggest using higher mass resolving power with a sector field instrument. Eventually - but I am not sure how efficient this can be - collision reaction cells can help as well.

Wash your glassware with Nitric acid 2-4% atleast 5 times and then with MiliQ water 5 times. From my experience with ICP, I am quite sure that you will get rid of Si. 

I use Duo system from Thermo and had similar problem.