CHAPTER 16 • Oceanic DOC Measurements
309
quires additional hydrogen, an additional reduction chamber maintained at high temperature, and a stable electrometer, all of which increase the costs. Although its sensitivity is very high and it is almost interference-free, frequent calibrations are needed
and unattended operation is not allowed.
The quantitation step implies the calibration of the system and the transformation
of the obtained signal in the needed chemical information. The influence of the analytical system itself on the overall signal is strongly related to the quality of the analytical results. If there is a consistent, but unrecognised blank signal, this leads to a
bias in the results. Much more dangerously, an inconsistent, unrecognised signal in
the analysis leads to data corruption with, at best, decreased precision and at worst,
seriously misleading results. The sources of blank signals are:
i. The system/instrument blank, which is a composite signal resulting from the sum of
the contributions of the components of the analytical system itself, e.g. carrier gas,
catalyst, reagents, pipework etc. It is due to anything that adds or subtracts CO2 to
that produced, or which adds non-C02 compounds, which interfere with the detection.
ii. The water blank (a signal that derives from the presence of residual organic carbon in
the low organic carbon water commonly used for the preparation of standard solutions).
iii. The sample blank, due to production of non-C02 components during oxidation or
combustion which can give spurious signals at the detector.
iv. The sample memory blank, due to memory effects observed when the signal generated is to some extent influenced by the previous samples that have been analysed.
Another term for this phenomenon might be a carry-over blank.
v. The procedural blank, which is the signal that derives from the three components
identified above plus any additional factors, which may arise because of organic carbon contamination or removal during sample collection and storage procedures.
Considerable efforts have been made to quantify all of the possible influences of
the various blank contributions identified above (Benner and Strom 1993), but the
variety of systems in use in different laboratories has made a consistent approach difficult. These problems are particularly evident in the analysis of sea water, for which
the low Toe concentration is associated with high salt content. A number of authors
have published uncorrected data together with their determined blank values to try
and circumvent this problem (e.g. Ogawa and Ogura 1992; Kaplan 1992; de Baar et al.
1993 and Zweifel et al. 1993, 1995).
The contribution of the different blanks in the WCO and HTCO methods was
analysed by Cauwet (1994) using MilliQ water. In WCO method the blank includes the
reagent blank and the instrument blank, giving, after conditioning of the automated
system, a stable base line. When samples are processed the water blank is absent, giving rise to a lower base line by 12 ±1 IlM C with respect to carrier gas. In the HTCO
method the blank was estimated by injecting volumes of water ranging from 50 to
200 jll. From the linear correlation obtained, the blank was estimated (2.5 jlM C with
100 III injection). Some instruments (e.g. Shimadzu-TOC 5000) have the possibility
of collecting the pyrolized water and using it for blank checking. This blank check made
from carbon-free water gives a good estimation of the instrument blank. Cauwet (1994)
reported that this blank depended strongly on the catalyst used. It was high for 3%
PtlAl20 3 (42 jlM C) and lower when using Pt impregnated quartz wool (51lM C).
309
quires additional hydrogen, an additional reduction chamber maintained at high temperature, and a stable electrometer, all of which increase the costs. Although its sensitivity is very high and it is almost interference-free, frequent calibrations are needed
and unattended operation is not allowed.
The quantitation step implies the calibration of the system and the transformation
of the obtained signal in the needed chemical information. The influence of the analytical system itself on the overall signal is strongly related to the quality of the analytical results. If there is a consistent, but unrecognised blank signal, this leads to a
bias in the results. Much more dangerously, an inconsistent, unrecognised signal in
the analysis leads to data corruption with, at best, decreased precision and at worst,
seriously misleading results. The sources of blank signals are:
i. The system/instrument blank, which is a composite signal resulting from the sum of
the contributions of the components of the analytical system itself, e.g. carrier gas,
catalyst, reagents, pipework etc. It is due to anything that adds or subtracts CO2 to
that produced, or which adds non-C02 compounds, which interfere with the detection.
ii. The water blank (a signal that derives from the presence of residual organic carbon in
the low organic carbon water commonly used for the preparation of standard solutions).
iii. The sample blank, due to production of non-C02 components during oxidation or
combustion which can give spurious signals at the detector.
iv. The sample memory blank, due to memory effects observed when the signal generated is to some extent influenced by the previous samples that have been analysed.
Another term for this phenomenon might be a carry-over blank.
v. The procedural blank, which is the signal that derives from the three components
identified above plus any additional factors, which may arise because of organic carbon contamination or removal during sample collection and storage procedures.
Considerable efforts have been made to quantify all of the possible influences of
the various blank contributions identified above (Benner and Strom 1993), but the
variety of systems in use in different laboratories has made a consistent approach difficult. These problems are particularly evident in the analysis of sea water, for which
the low Toe concentration is associated with high salt content. A number of authors
have published uncorrected data together with their determined blank values to try
and circumvent this problem (e.g. Ogawa and Ogura 1992; Kaplan 1992; de Baar et al.
1993 and Zweifel et al. 1993, 1995).
The contribution of the different blanks in the WCO and HTCO methods was
analysed by Cauwet (1994) using MilliQ water. In WCO method the blank includes the
reagent blank and the instrument blank, giving, after conditioning of the automated
system, a stable base line. When samples are processed the water blank is absent, giving rise to a lower base line by 12 ±1 IlM C with respect to carrier gas. In the HTCO
method the blank was estimated by injecting volumes of water ranging from 50 to
200 jll. From the linear correlation obtained, the blank was estimated (2.5 jlM C with
100 III injection). Some instruments (e.g. Shimadzu-TOC 5000) have the possibility
of collecting the pyrolized water and using it for blank checking. This blank check made
from carbon-free water gives a good estimation of the instrument blank. Cauwet (1994)
reported that this blank depended strongly on the catalyst used. It was high for 3%
PtlAl20 3 (42 jlM C) and lower when using Pt impregnated quartz wool (51lM C).