472
M. Pettine . L. Patrolecco . S. Capri
experimental system. Samples were measured in triplicate with a fixed c.v. of 2%; otherwise, further replicates were automatically carried by the instrument.
Particulate Organic Carbon. For pac analysis, the filters were dried at 60°C overnight,
re-weighed to determine particle loading and homogenized in an agate mortar mill.
Blank filters were processed the same way. Powered, homogenized samples (10-15 mg)
were accurately weighed (±0.01 mg) into tin or silver cups (9 x 5 mm) to determine
particulate nitrogen (PN), total particulate carbon (PC) or particulate organic carbon
(PaC). Samples in silver cups were acidified with 20 fJl5 M ultrapure HCl and kept at
50-60°C for 30 min to remove inorganic carbon. Care was taken to ensure complete
saturation of the sample with HCl and to avoid sample loss. Acid treatment was repeated
until effervescence was no longer observed (generally three times). PC, pac and PN
were determined by high temperature oxidation using a Carlo Erba NA 1500 series 2
C/H/N/O/S analyser. Samples were run in duplicate. The sulphanylamide standard was
used to construct the calibration curve. Carbon and nitrogen contents were expressed
as percentage of total solid and as concentration (mg tl) of filtered water sample. Average blank levels were 2.5 ±0.8 flg C and <0.5 flg N; detection limits (calculated as three
times the standard deviation of the blank for carbon, and according to sensitivity of
the instrument for nitrogen) were 2 and 0.5 flg for carbon and nitrogen, respectively,
corresponding to 0.02% C and 0.005% N for a 10 mg sample. Analytical precision was
±2.3 and ±1.8% of the measured values for total and organic carbon and ±1.6% for nitrogen. Particulate inorganic carbon (PIC) was determined by the difference between
PC and poe. Only pac data will be discussed in this paper.
Colloidal Organic Carbon. A limited number of sea water samples were processed with
a tangential flow ultrafiltration (UF) system (Amicon DC10LA) to investigate the molecular weight distribution of the DaM pool. For DOC fractionation, about 30 1 samples were filtered aboard, immediately after collection, through 0.4 flm polycarbonate
filters (Nucleopore, 0 142 mm) in a closed pressurized system (N2), using a Sartorius
Teflon -covered apparatus. To avoid contamination, the first 1-2 I of the filtrate were discharged. Filtered samples were then passed in cascade through two polysulphone membranes, which have nominal molecular weight cut -offs of 10 kDa (hollow-fibre cartridge,
Amicon, HlOPlO) and 1 kDa (spiral-wound cartridge, Amicon, SlONl), respectively.
Operating pressures were 20-25 psi at the inlet and 10-15 psi at the outlet of the system. UF cartridges were thoroughly cleaned, recirculating sequentially 0.1 N NaOH and
0.01 N HCl solutions for at least 30 min before and between samples and rinsing several times with about 5 1 Milli-Q water. Finally, after conditioning the UF cartridges with
about 11 filtered sample, about 20 1 of sample were processed and the concentrated colloidal fraction was reduced to about 21, thus achieving a lO-fold concentration factor.
Following this procedure, we fractionated the DOCpool into a low-molecular-weight
or "truly dissolved" fraction (DOC < 1 kDa) and two colloidal size classes, operationally defined as low-molecular-weight COC (1 kDa < LCOC < 10 kDa) and high-molecular weight COC (10 kDa < HCOC < 0.4 flm). Each sample was stored frozen in duplicate into 25 ml high density polyethylene (HDPE) bottles (as previously described for
DOC). In the laboratory, DOC analyses were performed on 0.4 flm polycarbonate filtered samples and their four different molecular weight fractions (>10 kDa; <10 kDa;
1-10 kDa; <1 kDa), in order to calculate partial mass balances on each cartridge. DOC
M. Pettine . L. Patrolecco . S. Capri
experimental system. Samples were measured in triplicate with a fixed c.v. of 2%; otherwise, further replicates were automatically carried by the instrument.
Particulate Organic Carbon. For pac analysis, the filters were dried at 60°C overnight,
re-weighed to determine particle loading and homogenized in an agate mortar mill.
Blank filters were processed the same way. Powered, homogenized samples (10-15 mg)
were accurately weighed (±0.01 mg) into tin or silver cups (9 x 5 mm) to determine
particulate nitrogen (PN), total particulate carbon (PC) or particulate organic carbon
(PaC). Samples in silver cups were acidified with 20 fJl5 M ultrapure HCl and kept at
50-60°C for 30 min to remove inorganic carbon. Care was taken to ensure complete
saturation of the sample with HCl and to avoid sample loss. Acid treatment was repeated
until effervescence was no longer observed (generally three times). PC, pac and PN
were determined by high temperature oxidation using a Carlo Erba NA 1500 series 2
C/H/N/O/S analyser. Samples were run in duplicate. The sulphanylamide standard was
used to construct the calibration curve. Carbon and nitrogen contents were expressed
as percentage of total solid and as concentration (mg tl) of filtered water sample. Average blank levels were 2.5 ±0.8 flg C and <0.5 flg N; detection limits (calculated as three
times the standard deviation of the blank for carbon, and according to sensitivity of
the instrument for nitrogen) were 2 and 0.5 flg for carbon and nitrogen, respectively,
corresponding to 0.02% C and 0.005% N for a 10 mg sample. Analytical precision was
±2.3 and ±1.8% of the measured values for total and organic carbon and ±1.6% for nitrogen. Particulate inorganic carbon (PIC) was determined by the difference between
PC and poe. Only pac data will be discussed in this paper.
Colloidal Organic Carbon. A limited number of sea water samples were processed with
a tangential flow ultrafiltration (UF) system (Amicon DC10LA) to investigate the molecular weight distribution of the DaM pool. For DOC fractionation, about 30 1 samples were filtered aboard, immediately after collection, through 0.4 flm polycarbonate
filters (Nucleopore, 0 142 mm) in a closed pressurized system (N2), using a Sartorius
Teflon -covered apparatus. To avoid contamination, the first 1-2 I of the filtrate were discharged. Filtered samples were then passed in cascade through two polysulphone membranes, which have nominal molecular weight cut -offs of 10 kDa (hollow-fibre cartridge,
Amicon, HlOPlO) and 1 kDa (spiral-wound cartridge, Amicon, SlONl), respectively.
Operating pressures were 20-25 psi at the inlet and 10-15 psi at the outlet of the system. UF cartridges were thoroughly cleaned, recirculating sequentially 0.1 N NaOH and
0.01 N HCl solutions for at least 30 min before and between samples and rinsing several times with about 5 1 Milli-Q water. Finally, after conditioning the UF cartridges with
about 11 filtered sample, about 20 1 of sample were processed and the concentrated colloidal fraction was reduced to about 21, thus achieving a lO-fold concentration factor.
Following this procedure, we fractionated the DOCpool into a low-molecular-weight
or "truly dissolved" fraction (DOC < 1 kDa) and two colloidal size classes, operationally defined as low-molecular-weight COC (1 kDa < LCOC < 10 kDa) and high-molecular weight COC (10 kDa < HCOC < 0.4 flm). Each sample was stored frozen in duplicate into 25 ml high density polyethylene (HDPE) bottles (as previously described for
DOC). In the laboratory, DOC analyses were performed on 0.4 flm polycarbonate filtered samples and their four different molecular weight fractions (>10 kDa; <10 kDa;
1-10 kDa; <1 kDa), in order to calculate partial mass balances on each cartridge. DOC