From: JOHN PHILO <JOHN.PHILO@amgen.com>
  To  : rasmb <rasmb@bbri.eri.harvard.edu>
  Date: 19 Jul 1995 13:38:08 U

Re: Quartz windows

        Reply to:   RE>Quartz windows

Cindy Turgeon wrote:
>I have also seen negative spikes during scanning occurring randomly
>and at different radial distances for each cell. 

I believe these are more likely to be lamp mis-fires than cracks or
scratches.  These seem to occur randomly, and in my instrument I would
estimate that the frequency is about 1 every 20 scans at .001" spacing and no
averaging (more frequent if you average).

With regard to how to deal with these problems:  My own practice is to
discard those data points which are clearly mis-fires.  My software and the
UConn packages adopt a convention where data points are ignored if the
standard deviation entry in the second column of the data file is set to a
negative value (I do this manually using Notepad).  This convention allows
you to eliminate bad points while maintaing integrity of the original raw
data (since you can bring the point back again).  I do not know if the
Beckman software follows this convention.

To deal with scratches or dirt, I have had good success in subtracting them
out using a baseline file.  One approach to obtaining a baseline is to use a
scan at a wavelength where the sample does not absorb (although the zero
offset may not be exactly the same at other wavelengths, and the effect of
scratches does seem to be somewhat wavelength-dependent).  For velocity data
you also have the option to refill the sample side of the cell with buffer at
the end of a run. For equilibrium data you can also take the sample to high
speed to force all the sample to the bottom, leaving a clear region where you
can use the data to try to subtract the dirt/scratch distortions.  The
problem with the latter approach is that radial expansion will shift the
apparent positions of the dirt/scratches to higher radii, so you must
radially shift the baseline file to compensate.  If you do this, it will
work, though not perfectly.

>I would be very interested to know if anyone else has detected
>hairline cracks in their windows, and if so, did they see any negative
>spikes in their scans?

I have seen some small cracks develop, but in my experience these usually
give absorbance spikes, not dips, and the window in question will break
within a short time.

>Also, I am very curious about what others have experienced regarding >the
'life expectancy' of their windows.

I have some windows that have been in use for 2-3 years for equilibrium runs.
 The lifetime of those I use for velocity work at 60K is shorter.  Overall,
however, I would attribute most of my window attrition to my own
mis-handling.

'Regards,

John Philo, Protein Chemistry, Amgen