PDS_VERSION_ID         = PDS3
RECORD_TYPE            = STREAM
OBJECT                 = TEXT
  PUBLICATION_DATE     = 2016-09-19
  INTERCHANGE_FORMAT   = ASCII
  NOTE                 = "Remarks on APXS Oxide Abundance data.
          THESE NOTES CONTAIN IMPORTANT CAVEATS REGARDING 
          THE USE OF THIS DATA SET. READ THESE NOTES CAREFULLY
          TO AVOID MISINTERPRETATION OF THE DATA.
"
END_OBJECT             = TEXT
END

Notes for the September 19, 2016 release
-----------------------------------------

The derived APXS oxide abundances have been updated up to sol 4000. The 
data are extracted with the exact same method, program, calibration 
parameter set as previous releases. 

As a reminder the given errors are precision( i.e. repeatability) 2-sigma 
statistical errors. The expected accuracy found in the MER calibration paper
as relative deviations during the laboratory calibration is given as  
Na 11%, Mg 14%, Al 7%, Si 3%, P 15%, S 15%, Cl 30%, K 15%, Ca 7%, Ti 20%,
Cr 19%, Mn 8%, Fe 7%, Ni 16%, Zn 16%, Br 20%.

All samples are assumed homogeneous and normalized to 100%. The norm 
column indicates the standoff with a norm of 100 being in contact at the 
beginning of the mission. Note that the source decay has not been taken into 
account in the norm yet ( ~ 30% loss for sol 4000).

The data user is cautioned about the usage of data that have short integration 
times and/or high standoffs, which occurred more often later in the 
mission. Here especially for low abundant trace and minor elements the 
errors might be underestimated in the standard analysis for these unusual 
conditions.                       

Notes added for the December 4, 2012 release
--------------------------------------------

Now the nomenclature is FM1 on Spirit (thinner Be window),
FM2 is on Opportunity ( thicker window)

Small adjustments have been made for Na,Mg and Al on both instruments.

Heavy adjustments on Chlorine. Chlorine is now ~ 40 % higher!  S and Cl are
neighbors in z, but had big differences in APXS response.  This was most
determined by the sample SSK with 1500 PPM Cl.  Now we have more samples with
Chlorine.  Also the low Cl values of the calibration samples had very large
error bars in the terrestial analyses.  I took out these samples and added
some more Chlorides as samples.

The new Chlorine response value is fine in the systematics.
Response (resp = CPS/Wtpercent)
now :   S:1.30    Cl 1.15    before  S: 1.28   Cl 1.61  -> ~ 50 % difference
A simulation of GUPIX group showed that Cl is smaller than S and closer to 
Potassium.

Instrument background taken into account.  I added up all background spectra
taken into atmosphere.  Here I fitted in Cr, Ni, Au and Pb lines. These 
components are now constant in all spectra taken into account. This means that
the Ni values drop a bit.  Especially for low Ni values at rocks and also 
for soils with bigger standoff distances (constant background is then
relatively higher) it makes a bigger difference.


Notes from previous releases
----------------------------

Remarks for Version G7 of APXS tables

Sample type
OU = outcrop undisturbed
OB = outcrop brushed
SU = soil undisturbed
SD = soil disturbed
ST = soil trench
SC = soils crest
RU = rock undisturbed
RB = rock brushed
RR = rock ratted
C   = cobbles

Information:
The oxide data were derived with the fit program, method, and calibration
parameters described in Gellert, R., et al 2006.

Periodically the local undisturbed soil was used to check for constancy of
the instrument performance and calibration as described in Ming, D.W. et al 
2008.

No degradation or change in calibration was found.

The samples are assumed homogeneous, i.e. matrix effect corrections are 
applied assuming a homogeneous average matrix and the calibration stems from 
thick ( ~ mm ) powdered samples. This is known to be not true, especially 
in the high Silica soils.

In certain samples the results indicate that small soil grains are intermingled 
with, e.g. the Silica rich soil. As a result the low z elements of these thin 
soil grains contribute more to the signal than the high z elements, e.g. Fe.
For discussion of the effects, Squyres 2008 and Ming et al. 2008.

Caveats :
Touch and go (T&G) short spectra were taken in the morning and evening. 
In addition to the short measuring time, the temperature was also higher.
T&G are visual by the small measuring time column.
The errors for T&G are complicated! Especially for Na,Mg,Al,P they might be
too low influenced by the overlapping peaks by degraded FWHM at higher
temperatures.


Remarks G10:
No changes in calibration.
On MERB a change in the instrument's acceptance was taken into account after
sol 300.


All values are in wt % ( Ni,Zn,Br in wt percent ppm (ug/g) )
All values were assumed with Fe2+ ( even for known high Fe3+ )
All values were assumed with a homgeneous sample! Known not to be true for
spherules!
The given errors are statistical errors of the peak areas. They represent
precision, not accuracy!
For the accuracy levels achieved during calibration, see Gellert et al. 
2006


Columns:
spectrum  Name of the summed spectrum
Type  Type of target
Name  unofficial Target name
Norm  sum of the oxide weight percent before normalization to 100 %.
Time/h  measuring time of the spectrum


References:
Gellert, R., et al. (2006), Alpha Particle X-Ray Spectrometer (APXS): Results 
from Gusev crater and calibration report, J. Geophys. Res., 111, E02S05, 
doi:10.1029/2005JE002555.

Ming, D.W. et al. (2008), Results of the Alpha Particle X-Ray Spectrometer from 
Cumberland Ridge to Home Plate, J. Geophys. Res., 113, E12S39, 
doi:10.1029/2008JE003195

Squyres, S.W. (2008), Detection of Silica-Rich Deposits on Mars,  Science,
320, 1063 - 1067, doi: 10.1126/science.1155429