Analytica Chimica Acta 539 (2005) 317325

Kinetic method for the determination of trace amounts of copper(II)

in water matrices by its catalytic effect on the oxidation of
1,5-diphenylcarbazide
Gaston Adrian Crespo a , Francisco Javier Andrade b,1 ,
Fernando Alberto Inon a, , Mabel B. Tudino a
a Laboratorio de Analisis de Trazas. Departamento de Qumica Inorganica, Analtica y Qumica Fsica, Universidad de Buenos Aires,
Pabellon 2, Ciudad Universitaria, (C1428EHA) Buenos Aires, Argentina
b Department of Chemistry, Indiana University, Bloomington, IN 47405, USA

Received 25 November 2004; received in revised form 22 February 2005; accepted 7 March 2005
Available online 25 March 2005

Abstract

The chemical reaction between Cu(II) and 1,5-diphenylcarbazide (DPC) is revised from an analytical perspective. It is shown that instead
of a CuDPC complex, an oxygen mediated DPC oxidation to diphenylcarbazone (DPCO) is the main reaction involved where Cu(II) acts
as catalyst. Therefore, the signal observed corresponds to the oxidation product DPCO. Under certain specific conditions a chelate complex
between Cu(II) and DPCO can also be observed. Kinetic parameters of these reactions are calculated. These results set the basis for the
development of analytical methodologies for the determination of Cu(II) at ultra-trace levels. Since the study of interferences revealed Mg(II)
as the most serious, an optimised flow injection system which allows the in-lineelimination of Mg(II) is presented. The limit of detection
(k = 3) calculated for 10 blank replicates is 0.05 g L1 (injection volume 20 L), the analytical sensitivity equals 48 g1 L and the sample
throughput is 120 h1 . Obtained results for Cu(II) determination in natural waters as well as the validation of the analytical methodology will
be given and discussed.
2005 Elsevier B.V. All rights reserved.

Keywords: Cu; DPC; Catalytic oxidation; Spectrophotometric determination

1. Introduction diphenylcarbazide (DPC) deserves a special attention as it

has been used for more than a century with this purpose [1].
Despite of the instrumental improvements of the last The applications of DPC for the analytical determination
decades, colorimetric techniques are still attractive from an of metal species (Cr(VI), Cu(II), Hg(II), etc.) were firstly
analytical point of view since they are cheap, sensitive and reported by Cazeneuve [2]. The reaction with Cr(VI) received
easy to implement in routine laboratories. In the field of by far more attention due to its high sensitivity and selectivity.
trace metal determinations, a plethora of organic colorimet- However, many efforts were devoted to the use of DPC for
ric reagents are still widely employed. From this list, 1,5- quantitation of other metals [36]. The different analytical
techniques have been summarized by Feigl [7].
Despite this wide range of applications of DPC, little was
Corresponding author. Tel.: +54 11 45763360; fax: +54 11 4763341. known about the mechanism of its reactions with metals:
E-mail address: finon@qi.fcen.uba.ar (F.A. Inon). either DPC or diphenylcarbazone (DPCO, the oxidation
1 On leave from Laboratorio de Analisis de Trazas. Departamento de

Qumica Inorganica, Analtica y Qumica Fsica, Universidad de Buenos

product of diphenylcarbazide) were suggested as probable
Aires, Pabellon 2, Ciudad Universitaria, (C1428EHA) Buenos Aires, ligands. Once it was shown than DPCO was the chelating
Argentina. agent, the efforts were oriented to study the conversion of

0003-2670/$ see front matter 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.aca.2005.03.013
318 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325

DPC into DPCO. Bose proposed [8] that the reaction of was prepared by dissolving 0.2430 g of sodium hydroxide
Cr(VI) and DPC gives a neutral complex Cr(III)-DPCO. (Merck, Darmstadt, Germany) in 750 mL of DDW with fur-
Pflaum and Howick [1] confirmed these results and con- ther addition of NaOH 0.1 mol L1 to adjust pH.
cluded that DPC is oxidised by Cr(VI) to DPCO, which The solution of DPC was prepared by dissolving 0.0250 g
in turns is reduced to Cr(III) and then, complexed by the of DPC (Merck) in the minimum volume of absolute ethanol
oxidised ligand. Thus metal species were divided into those (Merck) and then made up to 200 mL with DDW.
able to oxidize DPC, such as Cr(VI), Cu(II) or Hg(II), and DPCO was purified according to the Gerlach and Frazier
those requiring a previous oxidation step of DPC, which procedure [16] and solutions were prepared as stated above
was thought to be produced by air from the environment. for DPC.
Additionally, some works regarding the purity of the reagent A 1.000 g L1 Cu(II) (nitrate salt, Merck) was used as
showed that DPCO is present in commercial DPC [9,10]. stock solution. Copper working solutions were prepared by
Nonetheless, several decades later many questions regard- suitable dilutions of the stock solution. A pyrophosphate so-
ing the mechanism of these reactions remain unanswered. lution was prepared by dissolving 1.00 g of tetra-sodium py-
The mechanism of oxidation of DPC into DPCO was recently rophosphate (SigmaAldrich, St Louis, MO, USA) in 100 mL
studied by means of electrochemical and spectrochemical with DDW. A diluted acetic acid solution (1 105 M) was
techniques [11,12] but, no confirmation of the possible inter- prepared from glacial acetic acid (Merck).
mediates was given probably due to the similarity between A 0.01 M disodium ethylenediaminetetraacetic acid
their properties. Moreover, little information on decomposi- (EDTA, Na2 H2 Y) solution, prepared by dissolving 0.3720 g
tion products of DPCO, such us 1,5-diphenylcarbadiazone Na2 H2 Y2H2 O (Merck) in 100 mL of DDW, was employed
(DPDO), were shown [13]. for the study of physicochemical properties of DPC and
In a work reported by Turkinton and Tracey on cop- DPCO.
per determination [14], the analytical signal was ascribed Metal complexing reagents other than EDTA and sodium
to the formation of a complex of large molar absorptivity pyrophosphate were used to eliminate interferences through-
between Cu(II) and DPC, whereas Stoner and Dasler [15] out this work: (a) 3.7 104 M sodium citrate prepared by
evaluated its extraction in benzene. A change on the com- dissolving 0.100 g of C6 H5 Na3 O7 2H2 O (Merck) in 100 mL;
plex colour above a certain Cu(II) concentration was ob- (b) 3.1 104 M sodium fluoride obtained by dissolving
served, but no additional explanations were given. Balt and 4.200 g in 100 mL DDW and further dilution 1:4000 in
Van Dalen [5] found different Cu species forming the com- DDW.
plex and some works based on the application of this reac- For some experiments, solutions of H2 SO4 (Merck),
tion were reported furthermore [6]. However, the new ex- NaOH and KOH (Merck) were prepared by dissolving suit-
perimental evidence was not escorted by a revision of the able amounts of each reagent in DDW.
analytical strategy. In a recent work [12], it was shown that
the oxidation of DPC into DPCO is favoured by basic media 2.2. Samples
but, no reaction mechanisms or analytical applications were
presented. Different types of water samples were analysed in order
This work is devoted to revise and to analytically ex- to evaluate matrix effects on copper recovery. Spikes
ploit the reaction between Cu(II) and DPC. It will be shown of 10 g L1 of Cu(II) were carried out in: tap wa-
that depending on the conditions of the media, either a Cu- ter (250 S cm1 ),post-single stage osmosis water (6.8 S
DPCO complex or a DPC to DPCO Cu-catalysed conversion cm1 ), mineral water (120 S cm1 ), MilliQ water, and
can be observed. Kinetics information about this reaction is groundwater (60 m deep, 550 S cm1 ). Conductivity values
given, and some further experimental evidence regarding the (corrected at 25 C) are given for comparative purposes.
decomposition of these products is provided. These results
show that a methodology for the determination of Cu(II) 2.3. Apparatus and software
at ultra trace levels is readily available. Different analyti-
cal systems were evaluated in order to generate an interfer- Absorbance measurements were performed with a multi-
ence free method for the determination of Cu(II) in natural wavelength diode array HP 8453 spectrophotometer fur-
waters. nished with a cell holder and a cell-stirring module
(Hewlett Packard, Palo Alto, CA, USA) controlled through
the ChemStation software package. The spectrophotome-
2. Experimental ter cell, a standard QS 1 cm path length cell (Hellma,
Mulheim/Baden, Germany) fitted with a stirring bar, was
2.1. Reagents and solutions placed into the cell holder and filled through a flow injec-
tion (FI) assembly (dispensing mode). The FI system shown
Doubly deionised water (DDW, 18 M cm1 , Milli-Q in Fig. 1A was built using a peristaltic pump (IPC, Ismatec,
Water system, Millipore, Bedford, MA, USA) was used in all Glattbrugg-Zurich, Switzerland), a 10-port VICI rotary valve
the experiments, unless stated. Buffer solution of pH 12.00 (Valco Instruments, Houston, TX, USA), and 0.5 mm id
 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325 319

Fig. 1. Flow systems used for (A) kinetics experiments and (B) determination of Cu(II) in water samples: (A): A = Copper solutions, B = buffer carrier, C = DPC
solutions, L1 and L2 loops (injection volumes = 115 L and 117 L, respectively), W1 and W2 = waste; main pump was used in dispense mode (3.25 mL).
Auxiliary pump was manually switched on for loading the DPC and Cu solutions. (B): A = DPC solution (4.0 104 M, q = 0.7 mL min1 ), B = NaOH (0.17 M,
pH = 12.4, q = 0.7 mL min1 ), C = HAc (1.0 105 M, q = 1.5 mL min1 ), L = sample loop (20 L), D and E confluence points, R1 and R2 reactors (38 cm and
103 cm, respectively), G = samples, W1 and W2 = waste. Auxiliary pump was manually switched on for loading the sample loop. For more details see text.

PTFE tubes (Cole Parmer, Chicago, IL, USA). In order to che Werkstatten LF 521) furnished with a 2-poles platinum
save time and reagents, a second peristaltic pump (MS Re- cell (cell constant = 1.004 cm1 ).
glo, Ismatec) was used for loading solutions (samples) into
the valve loop. This assembly was employed only for ki- 2.4. Procedure
netic measurements as it will be discussed below. The flow
injection methodology for the determination of Cu(II) was A series of experiments was conducted in order to evaluate
developed using the experimental set up shown in Fig. 1B some physical and chemical properties of DPC and DPCO.
which is basically the same described above except for the Aqueous solutions of both reagents at different pH were pre-
employment of a 6-port VICI rotary valve (Valco Instru- pared as follows: an aliquot of the ethanolic solution of the
ments) for injection and a Shimadzu 10 VP spectrophotome- reagent was transferred to a flask, and a given amount of acid
ter (Shimadzu, Kyoto, Japan) variable wavelength UVvis (1 M H2 SO4 ) or base (1 M NaOH or KOH) was added before
detector connected to a Shimadzu Chromopac RC 6 for making up to volume with DDW. In some cases, the basic
detection. solutions (NaOH or KOH) were also 0.001 M in EDTA (or
Samples were also analysed by graphite furnace a similar chelating agent) as it will be explained below. The
atomic absorption spectrophotometry (GFAAS). A Shimadzu pH of these solutions was kept constant in 0.02 units. For
AA6800 graphite furnace-atomic absorption spectropho- some kinetics studies, solutions were mixed directly into the
tometer (Shimadzu, Kyoto, Japan) equipped with a Shimadzu spectrophotometric cell.
ASC-6100 autosampler, pyrolitic coated graphite furnaces, a UVvis spectra of these solutions were measured. Addi-
single element hollow cathode lamp of Cu (324.8 nm) and tionally, solutions of DPC and DPCO in acid, neutral and
a deuterium lamp for background correction was employed. basic media were extracted with CH2 Cl2 and the UVvis
The GFAAS program for Cu determinations in water samples spectra of the organic and aqueous layers were recorded.
is shown in Table 1. In order to study the reaction with Cu, solutions of
A Jenway ion meter Model 3345 (Jenway LTD, Essex, 0.01 mg L1 Cu(II) at different pH were prepared. An aliquot
UK) equipped with a glass electrode was used for pH mea- of these solutions was transferred to a spectrophotometric
surements. Conductimetric measurements were performed cell. Then, a given amount of either DPC or DPCO was added
by using a digital conductimeter (Wissenschaftlich Technis- under continuous stirring. UVvis spectra as function of time
were taken. Proper blanks were run in each experiment. This
Table 1 methodology was suitable for obtaining qualitative data but
GFAAS program for the determinations of Cu in water samples
not for quantification purposes since the manual operation
Stage Temperature ( C) Time (s) Heat Argon flow dramatically increases the blank level and thus, the limits of
(L min1 )
detection for Cu. The FI system shown in Fig. 1A was em-
1 120 10 R 0.1 ployed for mixing solutions directly in the flow cell and for
2 250 10 R 0.1
calculating kinetic parameters. Buffer solutions of different
3 700 10 R 1.0
4 700 7 S 1.0 pH values were used as carriers and dispensed with the peri-
5 700 3 S 0.0H staltic pump (total dispensed volume = 3.25 mL). After 1 s of
6 2500 3 S 0.0H starting the pump, the valve was commuted to the inject po-
Note: R means ramp, S means Step, H means high sensitivity. Sample volume sition. The loop holding the DPC solution was of 117 L and
20 L. that containing the Cu(II) samples was equal to 115 L. Vol-
320 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325

umes were properly calibrated, but no details will be given 3. Results and discussion
here. Loops were filled with each solution before commuting
the valve. A magnetic stir bar was used for homogenisation. 3.1. On the properties of DPC, DPCO and DPDO
The commutation of the 10-port valve triggered the spectra
acquisition. In order to fill the flow cell before the acquisition The chemical structures commonly proposed for DPC,
of the first spectra, a time delay of 10 s was allowed. DPCO and DPDO are shown in Fig. 2 and the UVvis spectra
In the next stage, the same flow dispenser system was used of the pure compounds are shown in Fig. 3. The spectrum of
for optimising the determination of Cu(II). The pH of the fi- DPC does not undergo significant changes with pH (data not
nal solution and the concentration of DPC were evaluated shown), which confirms that DPC is neutral and that it can
for attaining the best sensitivity. The optimum values were be quantitatively extracted in organic phase.
4 104 M for DPC and 0.17 M (pH = 12.4) for NaOH. Un- However, some changes are observed in DPC spectrum
der these conditions, the working curves for Cu(II) at different when time elapses. Whilst no variations are observed in acid
reaction times were obtained. conditions, a broad peak centred at 495 nm is observed in
Fig. 1B shows the flow system employed for Cu(II) deter- basic media (from pH 8 onwards). The solution becomes
mination in water. The basic set up includes a triple chan- notoriously red but this colour disappears as soon as it is
nel flow system with two mixing confluences. A diluted acidified. Nonetheless, a new spectrum in acid media differs
acid acetic solution (105 M) was used as carrier as some from the original one (Fig. 3), thus revealing a decomposition
memory effects were observed with water when samples in of DPC at high pH values.
basic media were analysed. The two confluences allow us, Since traces of metals acting as catalysts should be aid-
firstly, to adjust the pH of the solution to its optimum value ing to DPC decomposition in basic media, experiments were
(pH = 12.4) by mixing the carrier stream with a NaOH solu- repeated using basic solutions containing EDTA 1 mM as
tion and, secondly, to incorporate the DPC solution into the metal complexing agent. The addition of EDTA drastically
flow system. The flow rate of each solution, the length of the reduces the colour development and hence, the absorbance
mixing reactors and the sample volume were optimised for at 495 nm as time elapses. In fact, when no EDTA is added
maximizing sensitivity. In order to ensure a compromise be- the absorbance at this wavelength after 500 s is around 0.13,
tween sensitivity and sample throughput the latter was fixed at whereas when EDTA is added the absorbance is 0.001 at the
120 h1 . same reaction time. Similar results are obtained with different

Fig. 2. Chemical structures commonly proposed for DPC, DPCO and DPDO and keto-enolic equilibrium for DPCO: A = 1,5-diphenylcarbazide (DPC),
B = 1,5-diphenylcarbazone (DPCO), C = 1,5-diphenylcarbadiazone (DPDO). B-B1: keto-enolic equilibrium, B2 : possible quinoidean structure.
 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325 321

mechanisms could help in the improvement of the analytical

capabilities.
Regarding stability with time, DPCO is stable in acidic
media, but decomposes slowly in basic conditions yielding a
colourless solution. As the colour faints, a pale yellow solu-
tion with a weak band in the region of 320 nm appears whilst
the band at 495 nm vanishes. If the solution is left in an open
glass, the yellowing begins in the interface with air which
allows one to suspect that an oxidation by oxygen is being
promoted.
Literature [13] reports DPDO (C in Fig. 2) as the final
product resulting from the decomposition of DPC. In com-
parison to DPCO, DPDO presents a new double bond in its
molecule, which generates even more possibilities for iso-
meric structures. This fact reinforces the importance of know-
ing the structure of the precursor, DPCO, by studying its
Fig. 3. Relevant UVvis spectra of DPC, DPCO, DPDO in basic and/or acid spectrochemical behaviour. The spectra recorded for the de-
media: A = 1,5-diphenylcarbazide (4.37 105 M, pH = 12.4), B = 1,5-dip- composition of DPCO remains the same under pH changes
henylcarbazone (2.91 105 M, pH = 2.85), C = 1,5-diphenylcarbadiazone but basic solutions show a slight blue fluorescence which in-
(6.97 105 M, pH = 12.4) and D = 1,5-diphenylcarbazone (2.91 creases by irradiation with UV light (250 nm). This fluores-
105 M, pH = 12.4). Spectra were taken immediately after the addition of
the acid or basic buffer solution.
cence enhancement should be attributed to a photochemical
induced isomerisation but no studies on the actual chemical
form of the products of this photochemical reaction have been
complexing agents. Thus, although this reaction may spon- carried out.
taneously happen in basic conditions, it seems evident that
it is catalysed by metals. This fact becomes important as the 3.2. On the reaction between Cu(II) and DPC
reported analytical technique for copper determination [14]
proceeds in a basic medium. As it has been mentioned above, a solution of DPC in
Fig. 3 also shows the spectra of solutions of DPCO at dif- basic medium develops a red colour as function of time.
ferent pH. Strong differences are observed between the acid If Cu(II) at the g L1 level is added to the basic solution
and basic forms, confirming DPCO as a weak acid. Exper- in absence of metal complexing agents, a faster but simi-
iments with DPCO in acid media give way to a yellowish lar process is observed. In fact, the product formed in the
solution with a maximum at 300 nm, whilst those in basic presence of traces of Cu(II) presents similar spectroscopic
conditions show a strong red colour with an intense band at characteristics than DPCO: a red product with a maximum
495 nm. A new addition of acid turns the colour into yellow of absorbance at 495 nm that remains in the aqueous phase
again, revealing a reversible acidbase reaction. Extractions after extraction. If the reaction proceeds, the colour forma-
in organic solvents show the acid form as a non-charged com- tion reaches a maximum and then, a staining of the solution is
pound and the basic form as a charged one. A graph of ab- observed.
sorbance at 495 nm (A495) versus pH (data not shown) gives Some observations regarding this reaction must be men-
a pKa of about 8.5, which is close to the value already reported tioned. Firstly, the rate of these processes is influenced
in the literature [4] (pKa = 8.55,). by the concentration of Cu(II). Thus, if a quantity above
Even though DPCO structure is normally represented by of 10 mg L1 of Cu(II) is added, no colour formation is
the keto-form shown in Fig. 2B, the enolic form is more observed and the spectrum obtained is similar to that of
accurate at the moment of explaining the acidity of the the decomposition of DPCO. Secondly, the amount of
compound. The release of H+ (B1 in Fig. 2) should give product is independent of the amount of added Cu(II)
way to the charged basic form, but, the bathochromic shift since all the reagent decomposes if enough time is given.
of 200 nm observed in the spectrum of the basic form is From these experimental observations, it can be concluded
hard to explain just through the negative charge achieved that Cu(II) acts as a catalyst in the conversion of DPC
by the molecule. Probably, the generation of a more con- into DPDO, being DPCO an intermediate product of this
jugated compound such as the quinoidean structure (B2 in reaction.
Fig. 2) should explain the shift observed after the release of Actually, the effects of Cu(II) ions on the decomposition
the H+ ion Additionally, the double bond between the ni- of DPC are roughly mentioned in the literature and there is
trogen atoms can generate isomeric structures which may no clear explanation about the way this decomposition oc-
show different chemical and spectroscopic properties. Al- curs. Krumholz and Watzek [17] and Ayres and Johnson [18]
though, it is not the aim of this work to elucidate the struc- propose the autoxidation of DPC mediated by Cu, but the
ture of these compounds, a better knowledge of the involved authors attribute the discoloration to the formation of a com-
322 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325

plex of copper. Turkington and Tracey refer to a CuDPC For a given reaction (Eq. (1)), the rate law can be written
complex as the reaction product but, it is clear that the ana- as shown in Eq. (2).
lytical signal is due to the DPCO and not to the complex. As aA + bB cC + dD (1)
a matter of fact, the reported optimum pH for the complex
1 [A] 1 [B] 1 [C] 1 [D]
formation is about 5 [13,19] when DPCO is still fully pro- = = = k1 [A]a [B]b
tonated. However, the reaction reported by Turkington and na t nb t nc t nd t
Tracey takes place at pH above 9 when the anion of DPCO = k1 [C]c [D]d (2)
is prevalent.
In summary, the reaction between Cu(II) and DPC in- For the copperDPC reaction, the following rate law can be
volves numerous intermediate species and reactions and it derived assuming that the main reactants are Cu(II), DPC and
is dependent on the concentration levels of copper and the OH :
pH of the medium. High pH seems to favour the oxidation 1 [DPC] 1 [Cu(II)]
of DPC to DPCO, probably (at least in part) because of the =
nDPC t nCu(II) t
decrease of the redox potential of oxygen. At the same time,
1 [OH ] c
it allows the formation of the anion of DPCO that shows = = k1 [DPC]a [Cu(II)]b [OH ]
a high molar absorptivity. Lower pHs favour the formation nOH t
of the neutral Cu(II) complex that can be extracted into the where a, b and c are the orders of the reaction with respect
organic layer yielding a pink (max = 565 nm) compound. At to DPC, Cu(II) and OH , respectively. In order to evaluate
very high Cu(II) concentrations DPC is rapidly converted into the rate constant (k1 ), it is necessary to know the exact stoi-
the colourless DPDO, and DPCO is not observed. This fact chiometric coefficients of the CuDPC reaction (nDPC, nCu
might be related with the results reported by Ramirez et al. and nOH, respectively). Nevertheless, this requirement is not
[11], who attempted to perform the electrochemical oxida- necessary for calculating the order of the reaction with re-
tion of DPC with a carbon paste electrode. The authors found spect to every reactant. In this work, only the orders of the
very difficult to detect DPCO as intermediate product since reactants were calculated by means of the calibration curve
the redox potentials for the oxidation of DPC to either DPCO described below.
or DPDO were very close. Thus, Cu(II) should be acting as
a very selective catalyst in this reaction. 3.4. Calibration curve for DPCO
In order to prove the nature of the species involved in
the oxidation process, nitrogen was bubbled into solutions The purity of commercial DPCO was determined by titra-
of DPC and Cu(II) in basic medium. These solutions were tion, either with potentiometric and conductimetric detection
then mixed under nitrogen atmosphere and the absorbance using NaOH as titrating agent. An average purity of 44.37%
was followed as function of time. It was observed that the (w/w) was obtained with no statistical significant differences
absorbance at 495 nm grows until a plateau is reached. This (95% confidence level) between methods. Therefore, it was
plateau was considerable lower than that obtained under oxy- purified as described elsewhere [16].
genated conditions. If some air is allowed to enter (or bub- The calibration curve for DPCO was constructed in the
bled into the solution), the absorbance grows again until a 2.26 107 to 6.83 105 M concentration range. The sys-
new plateau is reached. Therefore, the oxidation occurs at tem shown in Fig. 1A was employed for all the mea-
some extent due to the oxygen present in the solution. As surements. In order to correct baseline variations, a lin-
a matter of fact, those solutions where the time of nitro- ear regression was performed between the absorbance at
gen bubbling was varied, reached different plateau heights 495 nm the absorbance at 540 nm and the concentration
suggesting that the system reacts depending on the oxy- of DPCO ([DPCO]). The parameters of this regression
gen content in the solution. Moreover, as it was mentioned were R2 = 0.9994, slope = (104.3 1.2) 102 M1 and inter-
above, the discolouration of the DPCO solution obtained in cept = (3.6 0.8) 103 .
the spectrophotometer cell begins in the region close to the It was not possible to build a simple linear calibration
liquidair interface, thus indicating that the second oxida- model for DPC, as DPCO and DPDO also absorb in the spec-
tion step (from DPCO to DPDO) is also produced by oxy- tral region in which DPC presents a significant absorptivity
gen. (see Fig. 3). Moreover, as DPCO decomposes to DPDO, it
is not possible to assess DPDO concentration. Therefore, the
3.3. Evaluation of the kinetic parameters of the reaction initial reaction rates of the reaction between Cu(II) and DPC
were determined by the amount of DPCO generated.
It could be expected that the new insight on the involved Fig. 4 shows the time evolution of the full UVvis spectra
mechanisms provide a way of optimising the performance of of DPC in basic media. From these spectra, the DPDO con-
the analytical technique. For doing so, some kinetic parame- centration was determined and the initial reaction rate was
ters of the reaction were evaluated. For practical reasons, O2 obtained by fitting the concentration versus time profile us-
concentration was considered to be constant and equal to the ing an exponential function. The slope at t = 0 was calculated
saturated concentration in contact with air. through this function.
 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325 323

Fig. 4. Time evolution of the spectrum of a CuDPC solution in basic media:

Fig. 5. Calibration curves using the system of Fig. 1A. Reaction time:
the system of Fig. 1A was used with DPC = 0.02 M, Cu(II) = 50 g L1 and
100 s (A), 30 s (B) and 20 s (C). DPC concentration = 0.021 M (pH = 12.4),
NaOH carrier (pH = 12.4). Arrows indicate time evolution.
Cu(II) concentration range = 0.1 to 1.2 107 M. Error bars represent S.D.
(n = 3).
3.5. Order for DPC

Solutions of DPC (n = 7) in a final concentration range is proportional to the acquisition time, which can be adjusted
between 1.4 105 to 7.8 104 M with a Cu(II) con- in order to change the dynamic linear range of the method (see
centration of 5.2 108 M (0.0032 mg L1 ) at pH = 12.4 Fig. 5). For reaction times of 20, 30 and 100 s sensitivities
were measured using the system of Fig. 1A. The regres- of 5.5 0.3, 7.7 0.8 and 19 1 106 L g1 and limits
sion of log(slope) versus log([DPC]) presents a slope of of detection of 0.5, 0.35 and 0.30 g L1 were obtained, re-
1.005 0.004 (R2 = 0.9999) which allows us to confirm that spectively. As it can be seen, the sensitivity nearly doubles
the order for DPC is 1. when the reaction time is increased from 20 to 100 s. Never-
theless, this increase in sensitivity is not directly reflected on
3.6. Order for Cu(II) a decrease in the limit of detection, as an increase in blank
variability is also observed. This can be explained due to the
Using the same procedure, solutions of Cu(II) (n = 6) fact that the reaction is very sensitive and the reaction cell is
in a final concentration range between 2.7 108 and open. Therefore, further experiments were focused to develop
1.1 107 M at pH = 12.4 and with a DPC concentration of different flow injection.
7.8 104 M were prepared. The regression of log(slope)
versus log([Cu(II))] allows us to confirm that the order 4.2. A ow injection system for the determination of
for Cu is 1, as the slope for this regression is 1.06 0.01 Cu(II) in different water samples
(R2 = 0.992).
Different configurations of flow systems for the determi-
3.7. Order for OH nation of Cu(II) have been evaluated, but only that with the
best performance is presented here. This system is schemati-
Using the same procedure, solutions of NaOH (n = 6) in cally shown in Fig. 1B, and basically consists in the pH con-
the pH range of 1112.5 were prepared with a Cu(II) concen- ditioning of the injected plug of sample followed by its mix-
tration of 5.2 108 M (0.0032 mg L1 ) and a DPC concen- ing with the DPC solution. A dilute acetic solution (HAcO)
tration of 7.8 104 M. The regression of log(slope) versus has been chosen as carrier since memory effects have been
log([OH ]) allows us to confirm that the order for OH is 1 noticed in neutral and basic media (probably due to the ad-
as the slope for this regression is 1.020 0.005 (R2 = 0.988). sorption of copper hydroxide on the tube wall). Flow rates,
injection volumes and reactor lengths have been optimised.
These values are shown in the legends of Fig. 1B.
4. Analytical applications of the CuDPC reaction From the kinetics experiments it can be concluded that
sensitivity will be improved when both pH and DPC con-
4.1. A batch methodology for the determination of Cu(II) centration are increased. These parameters were studied in
the concentration ranges of 1113 for pH and 5.0 105 to
The determination of Cu(II) can be carried out using the 4.0 104 M for DPC. Nevertheless, some practical prob-
system of Fig. 1A. In this case, the sensitivity of the reaction lems impede us to work above the upper limits of concentra-
324 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325

tion already mentioned. In the case of DPC, solubility prob- Therefore, the interference of Mg(II) was faced in a sec-
lems were found for 4 104 M onwards and this value was ond stage. In this case, pyrophosphate has been considered
chosen as the more suitable. For pH, extreme alkalinity raised due to their anti-tartar properties. The results were very sat-
blank values as they are directly proportional to NaOH con- isfactory, showing the total elimination of the interference at
centration. Tubing deterioration was noticed in this case. In a concentration of Mg(II) as high as 200 mg L1 and with-
order to keep low black values together with good sensitiv- out affecting the copper recovery. In fact the sensitivity for
ity for Cu concentrations ranging between 0.1 and 50 g L1 Cu(II) in DDW was not statistically different (significance
Cu2+ , pH was fixed at 12.4. Moreover, precipitation of cupper level 95%) to that obtained for a standard addition procedure
hydroxide in stronger basic conditions was prevented. on mineral water (Mg2+ = 45 mg L1 ).
Regarding flow injection variables (coil length and flow Taking into account these results, the in-lineconditioning
rate), the values were chosen in order to reach a reaction time of the sample was evaluated replacing the dilute acetic acid
of 30 s, which allows one to achieve good sensitivity for the solution by a solution containing 600 mg L1 of sodium py-
selected range of concentrations, and a sample throughput of rophosphate.
120 h1 . The working conditions were selected in order to Calibration curve has been carried out using Cu(II)
achieve these figures of merit compatible with proper opera- standard solutions in the 125 g L1 concentration
tion of the system as probably some of them were impover- range. The sensitivity and the intercept obtained are
ished at expenses of the improvement of others. 0.025 0.002 g1 L and 0.005 0.031, respectively (er-
Once the flow system was selected, a study of possible ror values were calculated for a 95% confidence level). For
interferences in waters was carried out. The concentrations injections (n = 3) of solutions containing 2 and 25 g L1 of
of some known interfering cations of the reaction [14] were Cu(II), reproducibility values (as coefficient of variation) of
selected according to those occurring in natural waters [20]. 0.5 and 0.1% were found, respectively. No noticeable changes
Table 2 shows the recovery of Cu(II) for synthetic samples in baseline record were found when MilliQ water was in-
spiked with the interfering ions. In all the cases negative er- jected (n = 3). Therefore, the standard deviation of the blank
rors have been obtained which is in agreement with that re- was estimated as the instrumental standard deviation of the
ported by Turkington and Tracey [14]. The behaviour of the baseline, which is 0.0004. Injection of an unknown water
interferent ions should be attributed either to the oxidation of sample (n = 3) yielded a signal of 0.0056 with a standard de-
DPC directly into DPDO (no DPCO generation is observed) viation of 0.0005. A limit of detection of 0.05 g L1 was
or to the inhibition of the oxidation of DPC due to changes then calculated.
of the redox potential of the environment Our study revealed With this system, different water samples un-spiked and
Fe(III), Mn(II) and Mg(II) as the most important interferences spiked with 10 g L1 of Cu(II) were analysed. As it can
in copper determination under the tested conditions. be seen in Table 3, the recoveries of Cu(II) of the spiked
In order to overcome this drawback two different strate- samples are excellent. In order to validate de methodology,
gies were tested. The first one was the use of a chelating the same solutions were measured by GFAAS. The slope and
agent for Fe(III) and Mn(II) (such as citrate ions) [21]. Differ- the intercept of the regression between the concentrations
ent solutions of 10 g L1 of Cu(II) containing 15 mg L1 found by the flow system versus those found by the reference
of Fe(III) and increasing amounts of sodium citrate were methodology were 0.999 0.030 and 0.1 0.2, respectively.
prepared and injected into the system. These experiments Therefore, the results obtained by the proposed methodology
showed that 0.17 mM of citrate was required to eliminate the are traceable to those obtained using GFAAS.
interference of iron, losing only 10% of Cu(II). This citrate A comparison of the figures of merit of both techniques
level was also enough for eliminating the Mn(II) interference, is shown in Table 4. Sample throughput and reproducibility
but no that produced by Mg(II).
Table 3
Studies on Cu(II) recovery in water samples using the developed FI system
Table 2
Sample Spiked (g L1 ) Found (g L1 )
Interference from some metals at concentration levels commonly found in
natural waters DDW water 0 <LD
DDW water 10 10.1 0.1
Analyte Concentration Recovery (%)
Osmosis water 0 0.2 0.1
Fe(III) 1 mg L1 58 Osmosis water 10 10.3 0.1
Cr(III) 50 g L1 77 Tap water 0 1.9 0.1
Cd(II) 50 g L1 98 Tap water 10 12.0 0.1
Co(II) 50 g L1 78 Groundwater sample 0 0.8 0.1
Mn(II) 1 mg L1 100 Groundwater sample 10 11.1 0.1
Mn(II) 5 mg L1 50 Mineral water 0 0.5 0.1
Mg(II) 25 mg L1 56 Mineral water 10 10.7 0.1
Notes: Ions has been added as nitrate or chloride salts to a 10 g L1 solu- Notes: error values () are the confidence interval at 95% of significance
tion of Cu(II). Recovery% indicates the ratio between the amount of Cu(II) obtained for the prediction of a new concentration from a new average signal
quantified and the actual amount of the analyte. (number of replicates = 3, number of standards = 4). [22]
 G.A. Crespo et al. / Analytica Chimica Acta 539 (2005) 317325 325

Table 4 Acknowledgements
Comparison of figures of merit between the proposed method and GFAAS
Figure of merit Proposed methodology GFAASa The financial support of UBACyT, INQUIMAE and
LOD (g L1 ) 0.05 0.1 CONICET is acknowledged.
Sensitivity (g1 L) 401 451
Characteristic mass (pg) 4
Sample throughput (per hour) 120 15
Injection volume (L1 ) 20 20
References
Cost + +++
Reproducibility <1% 35% [1] R.T. Pflaum, L.C. Howick, J. Am. Chem. Soc. 78 (1956) 4862.
[2] P. Cazeneuve, Bull. Soc. Chim. Fr. 23 (1900) 701.
a Concentration values are informed for 20 L injection volume. [3] S. Balt, E. Van Dalen, Anal. Chim. Acta 27 (1962) 422.
[4] S. Balt, E. Van Dalen, Anal. Chim. Acta 27 (1962) 188.
[5] S. Balt, E. Van Dalen, Anal. Chim. Acta 29 (1963) 466.
are better in the proposed methodology, whereas LOD and [6] S. Balt, E. Van Dalen, Anal. Chim. Acta 30 (1964) 434.
sensitivity are practically the same [7] F. Feigl, Spot Test, Elsevier, Amsterdam, 1954.
[8] M. Bose, Anal. Chim. Acta 10 (1954) 201.
[9] G.J. Willems, R.A. Lontie, W.A. Seth-Paul, Anal. Chim. Acta 51
(1970) 553.
5. Conclusion [10] G.J. Willems, C.J. De Ranter, Anal. Chim. Acta 68 (1974) 111.
[11] M.T. Ramirez, A. Morales-Perez, A. Rojas-Hernandez, I. Gonzalez,
J. Electroanal. Chem. 410 (1996) 203.
In this work different aspects of the reaction between DPC [12] P. Salinas-Hernandez, A. Rojas-Hernandez, M.T. Ramirez-Silva,
and Cu(II) in basic media have been reviewed. Experiments Spectrochim. Acta A 59 (2003) 2667.
confirm that there is a catalytic oxidation of DPC to DPCO [13] E. van Dalen, S. Balt, Anal. Chim. Acta 25 (1961) 507.
by Cu(II). This reaction proceeds to the total oxidation of [14] R.W. Turkington, F.M. Tracy, Anal. Chem. 30 (1958) 1699.
DPC to DPDO. Kinetics experiments show that the order of [15] R.E. Stoner, W. Dasler, Anal. Chem. 32 (1960) 1207.
the reaction with respect to Cu(II), DPC and OH is 1 (for
[16] J.L. Gerlach, R.G. Frazier, Anal. Chem. 30 (1958) 1142.
[17] P. Krumholz, H. Watzek, Monatsh 70 (1937) 437.
each reactant). [18] G.H. Ayres, F.L. Johnson, Anal. Chim. Acta 23 (1960) 448.
The results were applied to develop an automated method- [19] H.R. Geering, J.F. Hodgson, Anal. Chim. Acta 36 (1966) 537.
ology for the determination of Cu(II) at ultratrace level. The [20] W. Stumm, J.J. Morgan, Aquatic chemistry, Chemical Equilibria and
automated system allows one the in-line conditioning of the Rates in Natural Waters, Wiley-Interscience, New York, 1996.
[21] J.A. Dean, Analytical Chemistry Handbook, McGraw-Hill, New
sample for the elimination of matrix effects in natural and York, 1995.
mineral waters. Moreover, the analytical performance is sim- [22] D.L. Massart, B.G.M. Vandeginste, L.M.C. Buydens, S. De Jong,
ilar to the well-known GFAAS technique, but shows better P.J. Lewi, J. Verbeke, Handbook of Chemometrics and Qualimetrics
reproducibility and sample throughput. (Part A and B), Elsevier Science B.V, Amsterdam, 1997.