200 R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207
worldwide. Knowing only the total arsenic concentration in
the plant is insufﬁcient to understand the mechanisms of ar-
senic hyperaccumulation by this plant and effectively use
the plant to clean up arsenic contaminated sites.
The goal of this research was to develop a reliable and
robust analytical method for routine arsenic speciation in
a single run. Reverse-phase liquid chromatography (LC) is
the most popular LC separation mode due to its high sep-
aration efﬁciency, good sample loading tolerance, and abil-
ity to separate a broad range of different polarity samples.
Reverse-phase ion-pair chromatography has been developed
to routinely separate both non-ionic and ionic compounds
in a signal run using the same column . Ion pairing
reagents, including tetrabutylammonium hydroxide (TBAH)
[5–7], tetrabutylammonium phosphate (TBAP) , sodium
pentanesulfonate , methanesulfonic acid, and propanesul-
fonate acid , have been used by other research groups.
In the present research, a novel ion-pair reverse-phase
HPLC–ICP-MS was employed to perform arsenic speci-
ation, including arsenite [As(III)], dimethyl arsenic acid
(DMA), monomethyl arsenic (MMA), and arsenate [As(V)],
in environmental samples. The choices of HPLC separation
conditions were based on the selection and optimization
of ion-pairing reagent concentration, buffer concentration,
methanol concentration, pH, and mobile phase ﬂow rate.
The organic solvent and ﬂow rate effects as well as spectro-
scopic interferences were also considered for ICP-MS detec-
tion. This optimized ion-pair HPLC–ICP-MS method was
applied to determine arsenic species in Chinese brake fern.
The objectives of this research are to: (1) develop a
reliable analytical method for arsenic speciation in envi-
ronmental samples; and (2) apply this method to determine
arsenic speciation in the recently discovered arsenic hyper-
accumulating plant, Chinese brake fern. The arsenic specia-
tion information helps to better understand the mechanisms
of arsenic accumulation, transformation, and detoxiﬁcation
in Chinese brake fern. The HPLC–ICP-MS method can also
be widely applied to determine various elemental species
in environmental, biological, geological, and medical ﬁeld.
A VG plasma quadrupole II (VG Elemental, Winsford,
Cheshire, UK) ICP-MS was used. The ICP-MS was com-
puter controlled (Dell Dimension XPS 4100, Dell, TX, USA)
and VG instrument control software (Plasma Quad, Ver-
sion 4.30, VG elemental 1996) was operated under the OS/2
(IBM, USA) system. Tuning of the ICP-MS was performed
daily using a 100 ppb arsenic solution with a peristaltic pump
(Rainin, Woburn, MA, USA) and a Meinhard TR-30-A con-
centric nebulizer (Precision glassblowing, Englewood, CO,
USA) to maximize the signal response. After nebulization,
sample was transported to the ICP torch through a spray
chamber held at 5 ◦C. The quadrupole mass analyzer was
constantly scanned at m/z 75 for arsenic analysis. Data were
acquired in a time resolved acquisition (TRA) mode. There
was no ArCl interference on arsenic speciation in these stud-
The chromatography system consisted of a Spectra SYS-
TEM P2000 binary gradient pump (Thermo Separation
Production, Fremont, CA, USA), an Auzx 210 injector
valve with a 20 l loop and a Haisil 100 (Higgins An-
alytical, Mountain View, CA, USA) C18 column with
150 mm × 4.6 mm i.d. × 5 m particles. The optimized
mobile phase contained 10 mM hexadecyltrimethylammo-
nium bromide (CTAB) as the ion-pairing reagent, 20 mM
ammonium phosphate buffer, and 2% methanol at pH 6.0.
The stock solutions (1000 ppm of As) of arsenite [As(III)],
arsenate [As(V)], and DMA were prepared separately by
dissolving 0.1734 g NaAsO2 (MCIB, East Rutherford, NJ,
USA), 0.4164 g Na2HAsO4·7H2O (Sigma Chemical Co., St.
Louis, MO, USA), and 0.2133 g C2H6AsO2Na (Supelco,
Bellefonte, PA, USA) into 100 ml Milli-Q water. The stock
solution (100 ppm of As) of MMA was prepared by dis-
solving 0.0389 g CH3AsNa2O3·6H2O (Supelco) into 100 ml
Milli-Q water. All stock solutions were stored in refrigerator
at 4 ◦C.
The buffer solutions of HPLC–ICP-MS were prepared by
dissolving 2.1166 g NH4H2PO4 and 0.2133 g (NH4)2HPO4
into 1 l Milli-Q water for 20 mM ammonium phosphate
buffer. The pH of the buffer solutions was adjusted to 6.0 by
drop wise addition of diluted phosphoric acid or ammonium
hydroxide. The ion-pair reagent, CTAB, was prepared by
directly adding 3.6445 g of C19H42BrN to 1 l buffer solu-
tions to produce 10 mM CTAB solution. The mobile phase
was mixed using 98% (v/v) 20 mM ammonium phosphate
buffer and 10 mM CTAB solution with 2% (v/v) methanol
by HPLC pump.
Working solutions of arsenic were prepared daily by ap-
propriate dilution from the stock solutions with Milli-Q wa-
ter. All solutions and mobile phases were ﬁltered through
0.45 m Teﬂon ﬁlter (Gelman Instrument Company, Ann
Arbor, MI, USA). The mobile phases were degassed using
an ultrasonic bath for 20 min and also a helium sparge for
10 min before starting the chromatography.
2.3. Sample preparation and collection
2.3.1. Chinese brake fern cultivation and root exudate
The spores of Chinese brake fern were germinated in
an arsenic free soil mixture in a greenhouse for 3 months.
The fern plants with ﬁve to six fronds were transported
to a controlled hydroponic system with a relatively con-
stant temperature of 23–28 ◦C, humidity of 70%, and equal
amounts of artiﬁcial light. It took approximately 2 weeks
R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207 201
for the ferns to grow new roots. The fern roots were
washed free of soil by tap and deionized waters, respec-
tively. Then the ferns were transferred to a 20% strength
Hoagland nutrition solution and spiked with different lev-
els of arsenic species. The solution was buffered by 5 mM
2-2(N-morpholino)ethanesulfonic acid (MES) at pH 6.0.
After 2 days, the fern roots were washed again with tap wa-
ter, deionized water and phosphate buffer to assure arsenic
desorbed from the root free spaces. After washing with tap
and deionized waters again, each fern plant was placed in
150 ml deionized water to collect root exudate for 6 h. The
short collection time was used to minimize the impacts of
microbial activities on arsenic speciation. The root exudate
solution was ﬁltered immediately by a 0.45 M ﬁlter and
then analyzed immediately by the HPLC–ICP-MS method,
or stored below −80 ◦C for future analysis.
2.3.2. Chinese brake fern xylem sap collection
Chinese brake fern cultivation was the same as the proce-
dure described in Section 2.3.1. For this study, Chinese brake
ferns with similar size and age were selected. After the fern
roots were washed free of soil by tap and deionized waters,
respectively, the ferns were transported to a 20% strength
Hoagland nutrition solution for 1 week. Prior to harvesting,
the ferns were transferred to a new 20% strength Hoagland
nutrition solution, which was spiked with different levels ar-
senic species, and grown for 3 days. One or two fronds of
similar size were cut from the fern and placed immediately
into a pressure chamber for xylem sap collection. Xylem sap
was excreted from the cut of the frond by nitrogen gas in
the pressure chamber. Approximately 0.7–1.0 ml of xylem
sap was collected from each frond with a micropipette and
was stored at −80 ◦C for future analysis.
2.4. Method comparison
Arsenic speciation in the root exudates were also analyzed
by using an arsenic separation cartridge (Metal Soft Center,
Highland Park, NJ, USA) as a method comparison. This dis-
posable cartridge retained As(V) and allowed As(III) to pass
through to the ﬁltrate . Then As(III) concentration was
determined by a graphite furnace atomic absorption spec-
trometer (GFAAS) (SIMMA 6000, Perkin-Elmer, Norwalk,
CT, USA) to analyze the total arsenic in the ﬁltrate. Thus,
this method is only capable of separating As(V) from As(III)
in the sample.
3. Result and discussion
3.1. Ion-pair reverse-phase HPLC–ICP-MS arsenic
The chromatographic behaviors of arsenic species are
based on their acidic or basic properties (pKa value) as
shown in Table 1 . The elution order was predicted from
The formula and pKa value of arsenic species 
Compound Formula pKa
As(III) O=As–OH 9.3
Dimethylarsinic acid (DMA) 6.2
Monomethylarsinic acid (MMA) 3.6; 8.2
As(V) 2.3; 6.9; 11.4
their pKa values and veriﬁed experimentally by injecting the
four arsenic species individually. When the pH of the phos-
phate buffer solution was 6.0, arsenite was present as neu-
tral HAsO2 (pKa = 9.3), which is fully protonated and not
retained by the column. Hence, As(III) eluted ﬁrst with the
void volume. DMA (pKa = 6.2) was partially ionized at pH
6.0 and retained on the column a little longer than As(III),
therefore it eluted a little later than As(III); MMA (pKa1 =
3.6) and As(V) (pKa1 = 2.3 and present as H3AsO4) were
completely ionized and became anionic species, which re-
acted with hexadecyltrimethyl ammonium pairing cation and
were retained longer on the column. However, As(V) eluted
last due to its strong interaction with hexadecyltrimethyl am-
monium pairing cation. Under the chromatographic condi-
tions, the elution sequence was As(III), DMA, MMA, and
As(V), respectively, as shown in Fig. 1.
Hexadecyltrimethyl ammonium bromide was used as the
ion-pairing reagent, which is a compound with a polar head
(ammonium) and a non polar tail (hexadecyltrimethyl). The
reverse-phase mobile phase (98% ammonium phosphate
buffer and 2% methanol) was polar and the stationary
phase (silica-based bonded phase with C18 as ligand) was
non-polar. All of the four arsenic compounds formed anions.
Hence, they could be separated by the anion ion-pairing
reagent (CTAB), which could form hexadecyltrimethyl
ammonium cations. The ion-pairing reagent (CTAB) was
dissolved in the mobile phase, interacted with the stationary
phase, and was strongly retained by the column after a pe-
riod of time. Before CTAB dynamically coated the column,
the four arsenic species were not retained on the column
completely and rinsed out quickly. With longer equilibra-
tion times, the retention times of MMA and As(V) were
longer. Once the interaction between the ion-pairing reagent
and the column approached equilibrium, the retention times
for four arsenic species did not change during the whole
experiment. The concentration of the ion-pairing reagent,
which is CTAB in the present research, affected the time for
the column to achieve equilibrium. The higher the CTAB
concentration, the sooner the column equilibrated. The con-
centration of the ion-pairing reagent was typically in the
range of 1–5 mM . As shown in Fig. 2, it took 240 min
202 R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207
Fig. 1. Determination of arsenic species standard by HPLC–ICP-MS (all compounds present at 10 ppb).
to equilibrate the column with 1 mM CTAB; 130 min with
5 mM CTAB; but only 75 min with 10 mM CTAB holding
the other entire HPLC–ICP-MS conditions constant. After
the column equilibrated, As(V) retention times were con-
stant even at the different CTAB concentration conditions.
Therefore, 10 mM CTAB concentration was chosen as the
upper limit to equilibrate the column, and prevent clogging
the sampler cone at the same time.
The inﬂuence of other parameters such as buffer concen-
tration, methanol concentration, pH, mobile phase ﬂow rate,
and column degradation were also studied. The organic sol-
vent and ﬂow rate effects on ICP-MS were considered in
Fig. 2. Effect of CTAB concentration on column equilibrium time.
addition to achieve an effective separation and detection of
As(III), DMA, MMA, and As(V).
3.2. Arsenic speciation in the root exudates of Chinese
brake fern after treatment with different levels of As(V)
In these experiments, Chinese brake fern roots were
treated with 1.5, 15, or 150 ppm As(V) solution for 2 days
before root exudate collection (Fig. 3). When the fern roots
were treated with As(V), the predominant arsenic species in
the fern root exudate remained as As(V), ranging from 83
to 100% of the total arsenic concentration (Table 2). As(III)
R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207 203
Fig. 3. Concentrations of As(III) and As(V) in Chinese brake fern root
exudates. Roots were treated with 1.5, 15, and 150 ppm As(V) solutions
for 2 days.
was presented in some cases ranging from 0 to 17% of the
total arsenic concentration. This indicated that As(V) was
the main species in the fern root though some of the As(V)
was probably reduced to As(III) by the roots. This plant
survived in the solution containing up to 150 ppm As(V).
The total arsenic concentration in root exudates ranged
from 1405 to 2955 ppb for the 150 ppm arsenic treatment,
corresponding to 1–2% of the original treatment solution,
which indicated that the fern roots released some of the
arsenic taken up by the roots back into solution.
The same fern root exudate samples were also analyzed
using the arsenic cartridge-GFAAS method (Table 2). The
sum of As(III) and As(V) concentrations determined by
HPLC–ICP-MS and the total arsenic concentration deter-
Concentrations of arsenic species in Chinese brake fern root exudates after being exposed to arsenic for 2 days
HPLC–ICP-MS (ppb) Arsenic cartridge GFAAS (ppb)
Control As(III) As(V) Total As (As(III)/total As) × 100 As(III) As(V) Total As (As(III)/total As) × 100
R1-1 ND ND ND 1 2
R1-2 ND ND ND ND 1
R1-3 ND ND ND ND 2
R2-1 ND 38 38 ND 1 50 3
R2-2 ND 38 38 ND 3 48 6
T2-3 6 29 35 17 6 44 14
R3-1 ND 84 84 ND 2 97 2
RT3-2 2 118 120 2 10 170 6
R3-3 6 276 282 2 11 287 4
R4-1 ND 1483 1483 ND 12 1430 1
R4-2 ND 1405 1405 ND 30 1340 2
R4-3 234 2721 2955 8 389 2810 14
R1-1: root exudate blank sample no. 1; (As(III)/total As) × 100: (As(III) concentration/total arsenic concentration) × 100; ND: not detected; detection
limit for As(III), DMA, MMA, and As(V) is 0.5, 0.4, 0.3 and 1.8 ppb of arsenic.
mined by arsenic cartridge-GFAAS were in good agreement
for the majority of the samples. However, the As(III) con-
centrations were slightly different. In some cases, the ar-
senic cartridge-GFAAS method gave As(III) concentrations
ranging from 1 to 6% of the total arsenic concentration.
However, the HPLC–ICP-MS results showed that there was
no As(III) present. This was possible since the cartridge
retained only As(V), which meant that all species other
than As(V) passed through the column and were counted as
As(III). Therefore, it is expected that the cartridge-GFAAS
method may overestimate As(III) concentration if As(V) in
the fern sample is present in a complex form and hence
passes through the column. The average recovery of As(III)
was 98% using the cartridge, with arsenic concentrations less
than 500 g l−1 . The sensitivity of this GFAAS method
was adequate to analyze total arsenic in most fern samples.
However, DMA and MMA could not be determined by this
arsenic cartridge-GFAAS method due to the limitation of
the cartridge’s selectivity. Therefore, HPLC–ICP-MS is an
important analytical technique for arsenic speciation in Chi-
nese brake fern.
3.3. Arsenic speciation in the root exudates of Chinese
brake ferns after treatment with 15 ppm As(III), As(V),
DMA, or MMA
In this experiment, Chinese brake fern roots were treated
with 15 ppm As(III), As(V), DMA, or MMA for 2 days be-
fore root exudate collection (Fig. 4). For the control, when no
arsenic was applied, the only species present in the root ex-
udate was As(V) at a small level of 2–7 ppb (Table 3). When
the fern roots were treated with either As(V) or As(III), the
main arsenic species in the root exudates was As(V) com-
prising of 97–100% of the total arsenic concentration. This
204 R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207
Fig. 4. Concentrations of four arsenic species in Chinese brake fern
root exudates. Roots were treated separately with 15 ppm As(III), DMA,
MMA, or As(V) solutions for 2 days.
was consistent with our hypothesis that arsenic reduction
occurred mostly in the fronds . In a study to determine
the location of arsenic reduction in the Chinese brake fern,
Tu and Ma  reported that As(III) accounted for 24–34%
of the total As in the excised roots (detopped) that were
treated with 50 ppm As(V) for 1 day, whereas 30–39% of
As(V) was present when the roots were treated with 50 ppm
As(III). Their data strongly suggest that both As(III) oxida-
tion [when treated with As(III)] and As(V) reduction [when
treated with As(V)] occurred in the roots. Oxidation of
As(III) has rarely been reported in the plants, but it has been
reported for soil bacteria  and mineral leaching bacteria
. In the study of Tu and Ma , As(III) concentrations
in the solution spiked with As(V) in the presence of the
roots increase by 3–17% in comparison to the control with-
out plant [99.8–100% As(III)], suggesting an occurrence
of As(III) oxidation/reduction possibly by microbial activ-
ity in the solution or direct root exudation of As(V)/As(III)
into the solution. The fact that more As(V) was present in
the roots (24–34%) than the solution (3.1–17%), and more
As(III) was present in the solution (71–80%) than the roots
(61–70%) suggested that As oxidation occurred inside the
However, As(III) was also detected accounting for 0–3%
of the total arsenic concentration when treated with either
As(V) or As(III), which was consistent with the data dis-
cussed in Section 3.2. Such results indicated that signiﬁcant
oxidation of As(III) to As(V) occurred either in or outside
the roots, which could not be veriﬁed in this experiment. Al-
though the root exudates were collected with minimum con-
tact hours (i.e. 6 h) to minimize microbial-mediated arsenic
oxidation in the solution, such a process is still possible.
Approximately 76–87% of DMA was detected in the root
exudate when the plants were treated with DMA (Table 3).
Additionally, there was no MMA present with such treat-
Arsenic speciation in the root exudates of Chinese brake ferns after
treatment with 15 ppm As(III), As(V), DMA, or MMA solution for 2 days
Control As(III) DMA MMA As(V) Total
R1-1 ND ND ND ND ND ND 7 100 7
R1-2 ND ND ND ND ND ND 3 100 3
R1-3 ND ND ND ND ND ND 3 100 3
R1-3 ND ND ND ND ND ND 2 100 2
R2-1 5 2 ND ND ND ND 221 98 226
R2-2 5 3 ND ND ND ND 143 97 147
R2-3 ND ND ND ND ND ND 165 100 165
R2-4 ND ND ND ND ND ND 205 100 205
R3-1 ND ND ND ND ND ND 173 100 173
R3-2 ND ND ND ND ND ND 188 100 188
R3-3 9 3 ND ND ND ND 271 97 280
R4-1 4 6 14 22 32 50 15 23 64
R4-2 5 1 42 8 95 18 386 73 528
R4-3 4 5 13 14 42 46 32 36 91
R4-4 4 3 29 21 77 56 27 20 136
R5-1 ND ND 120 87 ND ND 18 13 138
R5-2 ND ND 84 87 ND ND 12 13 96
R5-3 ND ND 58 76 ND ND 19 24 77
R1-1: root exudate blank sample no. 1.
ments. A typical chromatogram is shown in Fig. 5C. This
suggests DMA was a stable arsenic form and the plant failed
to convert it to other less toxic forms. This was consistent
with our previous data . In a study to determine the ef-
fects of arsenic species on plant growth and arsenic uptake,
Chinese brake ferns were exposed to 50 ppm As(III), As(V),
DMA or MMA for 12 weeks in a greenhouse study .
All the ferns survived such arsenic exposure except those
treated with 50 ppm DMA, where all died after 12 weeks of
When the fern was treated with MMA, MMA was the
main species comprising of 18–56% of the total arsenic con-
centration. In addition, DMA was presented when the fern
was treated with 15 ppm MMA, as shown in Fig. 5B. The
concentration of DMA was far above the detection limit of
the HPLC–ICP-MS method accounting for 8–22% of the to-
tal arsenic concentration. The original 15 ppm MMA treat-
ment solution was also analyzed to verify there was no DMA
contamination as shown in Fig. 5A.
This experiment demonstrated that MMA converted to
DMA in the presence of Chinese brake fern roots. How-
ever, it was unclear whether such the methylation oc-
curred inside the roots or in the exudate solution. Odanaka
et al.  also found same phenomenon in rice plants
with GC–MID-MS–HG-HCT (gas chromatography with
a multiple ion detection mass spectrometry and hydride
generation-heptane cold trap).
R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207 205
Fig. 5. Typical chromatograms of arsenic species (A) original 15 ppm MMA treatment solution and Chinese brake fern root exudate with (B) 15 ppm
MMA treatment; and (C) 15 ppm DMA treatment.
3.4. Arsenic speciation in the xylem saps of Chinese brake
ferns after treatment with 10 ppm or 50 ppm DMA or
The role of xylem sap in the fern is to transport arsenic
from the roots to the fronds. In this experiment, the fern
roots were treated with 10 ppm or 50 ppm DMA or MMA
solutions for 3 days before xylem sap collection (Fig. 6).
For the control when no arsenic was applied, approximately
70–89% of the total arsenic concentration in the xylem sap
was present as As(III) and 10–25% was As(V) (Table 4).
It is important to note that As(V) was the only species in
the root exudates for the Chinese brake fern without arsenic
treatment (Table 3). This was in good agreement with previ-
ous research concerning the conversion of As(V) to As(III)
in the plant . Inorganic arsenic is the predominant form
of arsenic in the Chinese brake fern xylem sap in the control.
This also suggested that some arsenic reduction occurred ei-
ther in the roots after uptake or in the xylem during translo-
cation. However, a small amount of MMA, ranging from 1
to 7% of the total arsenic, was also detected, which may
indicate that methylation occurred inside the plant. Further-
more, no DMA was detected.
As shown in Table 4, the total arsenic concentration in
xylem sap can increase up to 344 ppm with 50 ppm DMA
treatment, which was approximately six times greater than
the solution concentration. This further veriﬁed the hyper-
accumulating property of Chinese brake fern.
206 R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207
Fig. 6. Concentrations of four arsenic species in Chinese brake fern xylem sap. Roots were treated separately with 10 ppm or 50 ppm DMA or MMA
solutions for 3 days.
When the Chinese brake fern roots were treated with ei-
ther 10 ppm or 50 ppm DMA, as expected, the predomi-
nant arsenic species in the fern xylem sap was DMA. DMA
ranged from 78 to 100% of the total arsenic concentration,
and there was no MMA present. This indicated that demethy-
Concentrations of arsenic species in Chinese brake fern xylem sap after the plants were treated with 10 ppm or 50 ppm DMA or DMA solutions for 3 days
Control As(III) DMA MMA As(V) Total (ppm)
C (ppm) % C (ppm) % C (ppm) % C (ppm) %
S1-1 1.0 89 ND ND 0.015 1 0.12 10 1.2
S1-2 0.29 70 ND ND 0.031 7 0.093 23 0.41
S1-3 0.055 84 ND ND 0.003 5 0.007 11 0.066
S1-4 0.14 72 ND ND 0.007 4 0.047 25 0.19
10 ppm DMA
S2-1 0.17 2 7.7 97 ND ND 0.093 1 8.0
S2-2 0.36 20 1.4 78 ND ND 0.042 2 1.8
S2-3 0.032 1 4.4 99 ND ND 0.024 1 4.4
S2-4 0.068 2 4.3 98 ND ND 0.011 0 4.4
50 ppm DMA
S3-1 0.091 0 65 100 ND ND 0.23 0 65
S3-2 0.083 0 38 100 ND ND 0.055 0 38
S3-3 0.11 0 30 100 ND ND 0.058 0 30
S3-4 0.11 0 344 100 ND ND 0.043 0 344
10 ppm MMA
S4-1 0.14 0 0.87 2 44 95 1.5 3 47
S4-2 0.52 7 0.40 5 5.9 80 0.54 7 7.3
S4-3 0.33 10 0.16 5 2.7 79 0.21 6 3.4
50 ppm MMA
S5-1 1.3 4 1.4 4 29 87 1.8 5 33
S5-2 0.43 5 0.15 2 8.3 87 0.71 7 9.6
S5-3 0.29 1 0.35 2 18 90 1.3 6 21
S5-4 1.2 13 0.31 3 7.2 80 0.32 4 9.0
S1-1–Xylem sap sample blank no. 1.
lation did not occur inside the plants. MMA remained the
primary species ranging from 79 to 95% of the total arsenic
concentration when treated with either 10 ppm or 50 ppm
MMA. Furthermore, DMA was also discovered in the xylem
sap treated with MMA. The concentration of DMA was far
R. Chen et al. / Analytica Chimica Acta 504 (2004) 199–207 207
above the detection limit of this HPLC–ICP-MS method,
which was 0.4 ppb, and accounted for 2–5% of the total ar-
senic concentration. This is in agreement with the conversion
of MMA to DMA in Chinese brake fern root exudate dis-
cussed in Section 3.3. Our results suggest that methylation
is possible in the Chinese brake fern. The relative standard
deviation of the total arsenic concentration ranged from 5
to 126% in either fern root exudate or xylem sap samples.
This is due to the variation among different fern plants.
Although demethylation from DMA to MMA was not
observed in the Chinese brake fern, conversion of DMA
or MAA to As(III) or As(V) was observed. However, it is
unclear how the conversion occurred.
HPLC–ICP-MS has been successfully used to perform
arsenic speciation in the root exudates and xylem saps of
Chinese brake fern after they were exposed to arsenic for
2–3 days in a hydroponic system. Our results conﬁrmed that
HPLC–ICP-MS was a reliable method that resulted in a low
detection limit and rapid analysis for arsenic speciation in
plant and aqueous samples.
When Chinese brake ferns were treated with either As(V)
or As(III), the primary arsenic species in the root exudates
was As(V), i.e. little arsenic reduction occurred in or outside
the roots, whereas signiﬁcant arsenic oxidation occurred ei-
ther in or outside the roots (Tables 2 and 3). This is consistent
with our hypothesis that arsenic reduction mostly occurred
in the fronds. When treated with DMA, DMA remained as
the dominant arsenic species with no MMA being detected
in the root exudates or in the xylem sap, suggesting DMA
is a stable arsenic form that the plant was unable to convert
to inorganic forms.
On the other hand, when treated with MMA, DMA
was detected in both the root exudates or the xylem sap
(Tables 3 and 4), suggesting methylation was possible in
this plant. The results from the root exudates suggested that
both arsenic reduction and oxidation occurred either inside
the roots or in the solution. If the latter is the case, then
microbial-mediated arsenic reduction and oxidation may
play a role. The results with root exudates and xylem sap
data suggested that DMA was more toxic to Chinese brake
since it failed to convert to other less toxic forms, and the
plant was capable of methylation (conversion from MMA
to DMA) in addition to converting MMA to As(III) and
As(V) in the roots or in the xylem.
This research was supported in part by the National Sci-
ence Foundation (Grant BES-0132114).
 A.H. Smith, E.O. Lings, M. Rahman, Bull. World Health Organ. 78
 M. Bissen, F.H. Frimmel, Fresenius J. Anal. Chem. 367 (2000) 51.
 L.Q. Ma, K.M. Komart, C. Tu, W. Zhang, Y. Cai, E. Kennelley,
Nature 409 (2001) 579.
 K.L. Sutton, J.A. Caruso, J. Chromatogr. A 856 (1999) 243.
 S. Wangkarn, S.A. Pergantis, J. Anal. At. Spectrom. 15 (2000)
 S.A. Pergantis, E.H. Heithmar, T.A. Hinners, Analyst 122 (1997)
 D. Beauchemin, K.W.M. Siu, J.W. Mclaren, S.S. Berman, J. Anal.
At. Spectrom. 4 (1989) 285.
 P. Thomas, K. Sniatecki, J. Anal. At. Spectrom. 10 (1995) 615.
 C. B’Hymer, K.L. Sutton, J.A. Caruso, J. Anal. At. Spectrom. 13
 X.C. Le, M. Ma, J. Chromatogr. A 764 (1997) 55.
 X. Meng, W. Wang, Presented at the 3rd International Conference
on Arsenic Exposure and Health Effects, San Diego, CA, 1998.
 R.L. David (Ed.), CRC Handbook of Chemistry and Physics, CRC
Press, Boca Raton, FL, 1994.
 M.S. Tu, L.Q. Ma, Environ. Exp. Bot., 2003, in press.
 S.E. Phillips, M.L. Taylor, Appl. Environ. Microbiol. 32 (1976) 392.
 H.M. Sehlin, E.B. Lindström, FEMS Microbiol. Lett. 93 (1992) 87.
 C. Tu, L.Q. Ma, J. Environ. Qual. 31 (2002) 641.
 Y. Odanaka, N. Tsuchiya, O. Matano, S. Goto, J. Agric. Food Chem.
33 (1985) 757.
 W. Zhang, Y. Cai, C. Yu, L.Q. Ma, Sci. Total Environ. 300 (2002)