www.jss-journal.

com Page 1 Journal of Separation Science

Acetone as a greener alternative to acetonitrile in liquid chromatographic fingerprinting

Cristiano Soleo Funari1,2,3, Renato Lajarim Carneiro4, Manish M. Khandagale1, Alberto José

Cavalheiro2, Emily F. Hilder1,*

1
Australian Centre for Research on Separation Science (ACROSS), School of Physical Sciences,

University of Tasmania, Hobart, Tasmania, Australia.

2
Institute of Chemistry, São Paulo State University, Araraquara, São Paulo, Brazil.
3
Faculty of Agricultural Sciences, São Paulo State University, Botucatu, São Paulo, Brazil.
4
Department of Chemistry, Federal University of São Carlos, São Carlos, São Paulo, Brazil.

Correspondence: Professor Emily Hilder, Australian Centre for Research on Separation Science

(ACROSS), School of Physical Sciences, University of Tasmania, Hobart, Tasmania, Australia.

Tel: +61 3 6226 7670

E-mail: Emily.Hilder@utas.edu.au

Abbreviations: C-CAD, Corona charged aerosol detector ; CCD, central composite design; DoE,

Design of experiments; EELalba, ethanolic extract of leaves of Lippia alba; GAC, green analytical

chemistry; EtOH, ethanol; GCFR, green chromatographic fingerprinting response.

Keywords: Corona charged aerosol detector /Green chromatography / Lippia alba / Metabolite

profiling/ Solvent replacement/

Received: 23-Nov-2014; Revised: 01-Feb-2015; Accepted: 02-Feb-2015

This article has been accepted for publication and undergone full peer review but has not been through the copyediting,
typesetting, pagination and proofreading process, which may lead to differences between this version and the Version of
Record. Please cite this article as doi: 10.1002/jssc.201401324.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 2 Journal of Separation Science

Abstract

A considerable amount of chemical waste from liquid chromatography analysis is generated

worldwide. Acetonitrile is the most employed solvent in liquid chromatography analyses since it

exhibits favorable physicochemical properties for separation and detection, but it is an unwelcome

solvent from an environmental point of view. Acetone might be a much greener alternative to replace

acetonitrile in reversed-phase liquid chromatography, since both share similar physicochemical

properties, but its applicability with ultraviolet absorbance based detectors is limited. In this work, a

reference method using acetonitrile and high-performance liquid chromatography coupled to an

ultraviolet photodiode array detector coupled to a corona charged aerosol detector system was

developed to fingerprint a complex sample. The possibility of effectively substituting acetonitrile with

acetone was investigated. Design of Experiments was adopted to maximise the number of peaks

acquired in both fingerprint developments. The methods with acetonitrile or acetone were successfully

optimised and proved to be statistically similar when only the number of peaks or peak capacity was

taken into consideration. However, the superiority of the latter was evidenced when parameters of

separation and those related to greenness were heuristically combined. A green, comprehensive, time-

and resource-saving approach is presented here, which is generic and applicable to other complex

matrices. Furthermore, it is in line with environmental legislation and analytical trends.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 3 Journal of Separation Science

Introduction

The debate on environmentally sustainable practices has been more active among practitioners in

the field of analytical chemistry and related areas in recent years [1–7]. It has been mainly pushed by

regulatory, safety and economical issues which arose from government and industrial sectors [8].

They understand that the impacts of an analytical process become important when it is used routinely

for QC of products and processes monitoring or even that individual unsustainable practices in

original research are not really carried out singly, but concomitantly by scientists and practitioners

worldwide. The term green analytical chemistry (GAC) was first employed by Namiesnik in 1999 [9].

Many definitions have been proposed for GAC, but the most comprehensive one has been attributed

to Lawrence, who defined it as “the use of analytical chemistry techniques and methodologies that

reduce or eliminate solvents, reagents, preservatives and other chemicals that are hazardous to human

health or the environment and that may also enable faster and more energy-efficient analyses without

compromising performance criteria” [7]. From this heuristic perspective the development of a new

method should be done taking into consideration aspects related with sustainability together with

aspects related to the efficiency of the analytical method in solving a specific problem [8, 10]. After

that, the concepts and principles of green chemistry gradually have been particularised to GAC [1,

11].

An important advance in GAC has been achieved with the proposal of its 12 principles by Gałuszka
et al. [11]. Briefly, they state that (1) direct analytical techniques should be applied to avoid sample
treatment, (2) small samples and a minimum number of samples are desirable, (3) in situ analysis
should be performed whenever possible, (4) integration of the analytical processes and operations to
save energy and solvents should be pursued, (5) automated and miniaturised methods should be
preferred, (6) derivatisation of samples should be avoided, (7) the generation of large volume of
analytical waste should be avoided and waste treatment should be performed, (8) multi-analyte, multi-
parameter approaches should be preferred over a single analyte, univariate approach, (9) the use of
energy should be minimised, (10) reagents/solvents obtained from renewable sources should be
preferred, (11) harmful reagents/solvents should be replace or eliminated and (12) the safety of the
analyst should be improved [11].

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 4 Journal of Separation Science

Among the systemic approaches addressed in chemical investigation of complex samples such as

crude extracts of plant organs, microorganisms, animals or biological fluids is the concept of a

metabolite profile or fingerprint. A fingerprint is a multi-analyte approach and as such it is in line with

the eighth principle of GAC outlined above [11]. It consists of a characteristic set of chromatographic

and/or spectroscopic signals acquired for a sample, and it is obtained from a comprehensive analytical

method that emphasises the chemical complexity of the sample by ideally detecting all compounds

present in it [12–14]. It can be useful in different types of investigation such as in the study of the

metabolite alterations of a given organism in response to an induced stress, in the case of

metabolomics, or to evaluate the identity and quality of plant-based materials as well as for QC of

their products, such as phytotherapeutics, nutraceuticals, cosmetics, and countless other natural

products useful to society [15–17]. This is a convenient alternative recommended by the World

Health Organization (WHO) to evaluate the consistency between batches in manufacturing processes

of standardised extracts, or to check the stability of finished products [12, 16, 18, 19]. This approach

has been gaining force by the understanding that the mere presence of a marker (or a number of

markers) in a complex sample used for therapeutic purposes does not necessarily represent their

overall biological properties [12, 20]. These may be based on synergistic effects between various

constituents (even those present in small amounts), acting on different receptors of an organism since

diseases are usually caused by multiple factors [12, 17, 20]. HPLC is the mostly used separation

technique for such applications. That is because it is suitable for almost any analyte in a sample

without the need of derivatisation (in line with the sixth principle of GAC outlined above) [11], is

easy to operate, can be fully automated and presents good resolution, reproducibility and selectivity

[12, 21]. The main drawback of HPLC is related to the impact caused by the disposal of the solvents

commonly used, notably acetonitrile and methanol (it is estimated that a typical analytical liquid

chromatograph employs about 1 L of organic solvent per day and that about 34 million litres of

chemical waste are generated each year worldwide) [6, 8, 12, 22]. Thus, the use of high amounts of

harmful organic solvents in HPLC mobile phases represents the biggest impact regarding

sustainability in a chromatography laboratory [8].

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 5 Journal of Separation Science

Although acetonitrile exhibits various advantages for LC applications, such as low viscosity,

miscibility with water and UV cutoff wavelength at 190 nm, it is an unwelcome solvent from the

sustainability point of view, contradicting the 7th, 11th and 12th principles of GAC outlined above [5,

11, 23–25]. That is because it is toxic to mammals has a half-life in water of 2–20 days and presents

acute and chronic toxicity to aquatic life [3, 26, 27]. Thus, its replacement with greener options is

highly desired in the current effort towards greening analytical chemistry methods and practices [3,

5–7, 11, 25–27].

The less toxic and biodegradable acetone has been much better ranked than acetonitrile in solvent

selection guides developed by pharmaceutical companies aiming to achieve more sustainable research

and processes [23, 24, 27]. According to the solvent selection guide developed by Pfizer, acetone is

classified as the second greenest solvent being surpassed only by water [24]. Acetone could be an

alternative to replace acetonitrile in many reversed-phase high-performance liquid chromatography

(RPLC) applications since both solvents share similar physical and chemical characteristics e.g. they

are fully miscible with water, hydrogen bond acceptor solvents and present some similar

solvatochromic properties [28], and similar viscosities [29]. Such replacement would fit principles

number 11 and 12 of GCA [11]. The main drawback of acetone compared to acetonitrile is that

acetone’s UV cut-off extends out to 330 nm restricting its usage with UV absorbance based detectors,

the most popular detectors in LC, to a narrow range of wavelengths [6, 29]. Nevertheless, the fast

popularisation of MS-based detectors in LC and the new generation of aerosol-based LC detectors

open new opportunities for the usage of acetone instead of acetonitrile. There are some reports on

satisfactory replacement of acetonitrile with acetone in RPLC employing mass spectrometer detectors

in the analysis of peptides [27, 29]. Recently, Hutchinson et al. investigated the potential replacement

of acetonitrile with acetone employing HILIC and corona charged aerosol detector (C-CAD) for the

separation of sugars [26]. These authors concluded that although acetonitrile led to higher column

efficiency and lower detection limits compared to acetone, the latter resolved the same number of

components as was possible with the former [26]. C-CAD has the ability of detecting all non-volatile

compounds in a sample even when they are devoid of a chromophore or are poorly ionisable

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 6 Journal of Separation Science

(prerequisites for UV- and MS-based detection, respectively) and is compatible with alternative,

greener volatile organic solvents such as acetone, ethanol and 2-propanol, and with elevated

temperature mobile phases [26, 30]. The main drawbacks of the C-CAD detector is that it is less

sensitive than UV- and MS-based detectors, it does not provide structural information of the detected

analytes and provides non-uniform response factors when gradient elution is performed due to

changes in the nebulisation efficiency as the mobile phase composition changes [26, 30]. However,

some strategies have been developed to normalise response. For example, an additional pump line can

be used to deliver solvent after the column in an inverse gradient to maintain solvent composition [26,

30]. Moreover, isocratic-temperature gradient separation can be used to replace conventional

replacement of solvent gradient, which can assist in uniform detection using C-CAD [30]. According

to Dolan et al., the variable temperature should be considered from the early stage of method

development as it can affect both the thermodynamic and kinetic of the process of separation [31].

Although energy is required to achieve high temperatures in the chromatographic system, which

opposes the ninth principle of GAC, it also allows reduction in the amount of organic solvent

employed in the mobile phase, which also fits the 7th, 11th and 12th principles of GAC outlined above

[11]. Thus, C-CAD detectors combined with high temperatures might be an interesting alternative for

chromatographic fingerprinting of very complex samples with no fully known composition or in

which non-chromophoric compounds should be expected. It is especially the case when the

replacement of a harmful solvent (such as acetonitrile) with a greener volatile one which presents a

high UV cut off (such as acetone) is desired and no quantification of analytes is required, as in

qualitative metabolite profiling.

Design of experiments (DoE) is a multivariate approach used in optimisation processes, such as in

the optimisation of analytical methods, where a response in intended to be maximised as function of

input parameters. It consists of a well-defined set of experiments statistically selected to evaluate

several variables at the same time, allowing the evaluation of the interactions between them. As a

result, it leads to a global optimised method stead of a local one obtained from an univariate approach

[12]. In the latter, one variable is changed at time (thus neglecting interactions between them) which

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 7 Journal of Separation Science

requires many experiments (and environmental resources, such as solvents and energy) when more

than two variables are considered to reach an acceptable separation [12, 32, 33]. That is the reason

why the univariate approach can be considered an inefficient trial-and-error process from both the

separation [32] and environmental point of view (it opposes the eighth principle of GAC, whereas the

multivariate approach fits it) [11].

In this work, the possibility of effectively replacing acetonitrile with acetone for fingerprinting a

complex plant extract with a real gain in terms of the greenness of the LC method was investigated.

For that, (i) an automated time and resource saving approach (DoE) was adopted to globally optimise

chromatographic methods based in both solvents, (ii) a HPLC coupled to a UV detector coupled to a

C-CAD detector system was used to allow comparison between detectors and to maximise the output

achieved from a certain amount of environmental resources, (iii) temperature was used as a variable

from the earlier stages of development and (iv) a more heuristic comparison of the developed methods

considering parameters related to separation and environmental parameters was performed. The

approach was maintained as generic as possible to be applicable to other complex matrices. This

approach offers an alternative to traditional LC method development approaches and shows that it is

possible to make the analysis of complex samples greener with no loss of separation performance.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 8 Journal of Separation Science

2 Materials and methods

2.1 Chemicals and reagents

Acetonitrile and acetone (Honeywell Burdick & Jackson, USA) and absolute ethanol (Scharlau, Spain

and J. T. Baker, USA) used were HPLC grade. The AcOH (Merck, Germany) was AR (ACS) grade.

2.2 Plant material

Leaves of the plant were collected from a private garden in Araraquara city (state of São Paulo,

Brazil) in 2012. A voucher specimen (CSF 2) was identified as Lippia alba (Mill.) N.E.Br. by Dr

Roseli B. Torres and deposited in the “Herbarium of the Agronomic Institute of Campinas” (state of

São Paulo, Brazil).

2.3 Extraction and concentration

Twenty one grams of fresh leaves of L. alba were covered with liquid nitrogen and ground using

mortar and pestle. The ground material was then transferred to a glass flask wrapped with aluminium

foil and extracted by maceration with three aliquots of 110 mL of EtOH at approximately 25 °C, with

constant stirring. The fluid extract was filtered and then concentrated under reduced pressure at 35°C,

yielding 531 mg of the ethanolic extract of L. alba (EELalba).

2.4 Pre-treatment using SPE

The C18 stationary phase (Waters Sep-Pak C18, 820 mg, 55–105 µm) was activated with 7.0 mL of

EtOH followed by equilibration with 7.0 mL of 15:85 H2O/EtOH v/v. It was then loaded with ca. 80

mg of the extract solubilised in 500 µL of the 15:85 H2O/EtOH v/v. The elution was developed with

3.0 mL of the 15:85 H2O/EtOH v/v to eliminate very low polarity compounds [10]. The eluate was

fully dried under N2. Before a set of analysis, it was solubilised in 7:3 H2O/EtOH v/v and filtered with

a 0.22 µm polyethersulfone filter to yield concentration of 20 or 80 mg/mL.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 9 Journal of Separation Science

2.5 HPLC analyses

In the method developments, a Dionex 3300 UHPLC system (Thermo Scientific, Australia) equipped

with binary solvent manager, an autosampler, column thermostat and a variable UV-Vis detector was

used. A C-CAD (Thermo Scientific, Australia) was placed in line with the UV-Vis detector. The N2

flow was fixed at 35± 0.5 psi at ambient temperature as recommended by the manufacturer.

Separations were obtained employing a C18 column (Waters XBridge BEH, 150 x 4.6 mm; 3.5 µm). A

micro channel pre-column heater (Thermo Scientific, Australia) was installed to ensure thermal

equilibration of the eluent. Chromatographic data were processed using the Chromeleon

Chromatographic Data system (version 6.8) and OriginPro 8 (OriginLab, USA). Statistical analyses

were performed using Matlab 2011a (Mathworks, USA), OriginPro 8 and Excel 2007 (Microsoft,

USA.) software. The conditions of the analyses of the experimental designs performed during method

optimisation are presented in Table 1 and Tables S2 and S4 (supporting information), whereas the

optimised methods are indicated in the caption of Figure 1 and in Section 3. An initial isocratic

elution that corresponds to a column dead volume (D0=1.54 mL) was performed before any gradient

elution presented in this work. After each run the HPLC column was cleaned with 10 mL of the

organic solvent to elute potential non-eluted compounds, thus avoiding interferences in subsequent

analyses and miss statistical interpretation. Column equilibration volume was 25 mL as indicated for

the manufacturer, whilst sample injection volume was fixed at 20 µL. Sample concentration was 20

mg/mL for the fractional factorial designs, but it was increased to 80 mg/mL when CCD designs were

performed to allow the detection of very minor compounds. Any signal with area up to 0.01 pA*min

was considered as a peak, since this value was slightly greater than the maximum peak values

observed in blank runs.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 10 Journal of Separation Science

3 Results and discussion

3.1 Development of a reference method with acetonitrile

Initially, a reference method for the separation of an ethanolic extract of leaves of Lippia alba

(EELalba) employing acetonitrile as organic solvent was developed using DoE and an automated

HPLC system. The overall number of peaks (n) was the response considered for optimisation purpose

[20]. In the first step, the influence of five variables in the process of separation was investigated by

means of a two-level fractional factorial design (Tables S1 and S2, supporting information). This type

of design might be enough to discern between relevant and irrelevant factors whereas performing only

a statistically selected part of a full factorial design [34]. In other words, it can be time and

environmental resources (such as energy and solvents) saving approach whereas allowing the

screening of a higher number of factors in a relatively few number of runs likened to a full factorial

design [10, 34].

In the range evaluated, the initial percentage of acetonitrile, gradient time and concentration of

acetic acid in A (X1, X2 and X4, respectively, Table S1, supporting information) were shown to be

statistically meaningful according to normal probability plots

(http://www.itl.nist.gov/div898/handbook/toolaids/pff/E-Handbook.pdf). The other factors tested (X3

and X5 for acetonitrile, Table S1) proved to not be statistically significant in the range evaluated.

Thus, in the optimisation step, the relevant variables were enclosed in a three-factor central

composite design (CCD) to maximise the total number of peaks. Alternatively, a full three-level

factorial design might be used at this step, but it would require a higher number of experiments than

the three-factor CCD whilst providing similar outcomes [34]. Thus, from the green chemistry

perspective the three-factor CCD should be preferred because it would save time, solvent and energy

[10].

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 11 Journal of Separation Science

The levels of variables X1, X2 and X4 are shown in Table S3 (supporting information) whereas the

CCD itself and the outcomes obtained are shown in Table 1. On the other hand, X3 and X5 were fixed

at 30°C and 0.85 mL/min, respectively. The lower level tested for the variable temperature (30°C)

was preferred since it means less consumption of energy in the process of separation compared to

higher temperatures, and it is a major issue in green chemistry and GAC [11].

From this set of experiments the mathematical model at a confidence level of 95% (Eq. 1) was

found to be as follows:

yˆ1  170.5 26.6 x2  6.3 x4 (Eq. 1)

 2.0  2.4  2.4

with an R2 of 0.89 and 89.29% of the explained variance. This model (Eq. 1) indicated that the best

result (highest number of peaks, n) should be achieved with variables X2 and X4 at level +1.68 (60 min

and 1% acetic acid in A, respectively). As the level of X1 proved to be not significant in the range

evaluated in this step, it was kept in the lower level (–1.68, which means 5% of acetonitrile) to reduce

the acetonitrile consumption during the analyses. Although it refers to the lowest percentage of

acetonitrile in the gradient elution, it is important because the column used here requires

approximately 16 column dead volumes (25 mL) of the initial mobile phase for equilibration

according to the manufacturer, thus, impacting in the overall environmental performance of the

method as discussed later in this manuscript. Thus, the predicted optimum point was experimentally

tested leading to 220 ± 6.1 peaks (n=7). The full chromatographic conditions and a chromatogram

with the optimised method are shown in Fig. 1a. This value surpassed all results observed in the

original CCD and was very similar to the optimum response predicted by the model (224).

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 12 Journal of Separation Science

3.2 Development of a substitution method with acetone

Once the reference method with acetonitrile was optimised, the next step was to optimise a

replacement method with acetone instead of acetonitrile no loss in separation performance, but with a

gain in environmental performance. Thus, the influence of six variables, including an addition of up to

50% of ethanol in acetone (Table S4, supporting information), was initially investigated in a 2 IV6–2

fractional factorial design (Table S5, supporting information).

The variables gradient time, temperature of analysis and concentration of acetic acid in A (X2, X3 and

X6, respectively, Table S4) appeared as the most statistically significant factors according to normal

probability plots. Thus, a three-factor CCD with X2, X3 and X6 was carried out to optimise a method

using acetone instead of acetonitrile. Table S6 (supporting information) indicates the variable levels

whist Table 1 shows the experiments performed as well as the responses achieved. Variables X1 and

X5 (Table S4, supporting information) were fixed at 5% of acetone v/v and 0.85 mL/min in the new

set of experiments, respectively. As the percentage of ethanol in B (X4, Table S4, supporting

information) proved to not be statistically significant in the range evaluated, it was fixed at 0% of

EtOH, which meant that B was fixed as 100% of acetone. This decision was taken to allow a more

direct comparison between acetonitrile and acetone since both share similar physicochemical

properties, as previous discussed.

The mathematical model built from this CCD as function of X2, X3 and X6 at a confidence level of

95% (Eq. 2) was found to be as follows:

yˆ 2  163.5 19.9 x2  7.6 x(Eq. 3.8 x22  4.2 x62  7.1 x2 x3

3  2)
 2.1 1.6 1.6 1.5 1.5  2.0

with an R2 of 0.94 and 94.21% of the explained variance. Eq. 2 indicated that the best result should be
achieved with variables X2, X3 and X6 at levels +1.63, –1.63 and 0, respectively. In other words, with
gradient time of 60 min, temperature of analysis at 30 °C and 0.5% of acetic acid in H2O v/v as A.
Thus, the predicted optimum point was experimentally tested leading to 219 ± 5.1 peaks (n=7). The
full chromatographic conditions and a representative chromatogram with the optimised method are
shown in Fig. 1b. This average value surpassed all previous experimental points observed in the
original CCD (Table 1) and proved to be very close to the optimum point predicted by the model (217
peaks).

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 13 Journal of Separation Science

The same performance of separation and detection observed with acetonitrile was achieved with

acetone when data were recorded employing C-CAD detector. The number of detected peaks acquired

with the optimised methods with both organic solvents were statistically similar (220.0±6.1 and 219 ±

5.1, respectively). The metabolite profiles obtained with both solvents proved to be very similar (Fig.

1), with the peaks (including the most prominent ones) being well spread from the dead time (t0=1.81

min) to the end of the analysis (61.8 min). This highlight the quality of the methods of separation

developed here based in the C-CAD detector and well planned statistical designs.

Regarding peak shapes, no significant differences were visually observed between acetonitrile and

acetone, with symmetric peaks being observed for both (Fig. 1). The average peak-width values were

also similar: 0.154 ± 0.008 for acetonitrile and 0.157 ± 0.009 for acetone. The peak capacity (PC) was

calculated according to Dolan et al. for each replicate by equation 3 [33]:

PC = 1 + (tG/W) (Eq. 3)

where tG is the gradient time, and W is the average peak-width.

As both optimised method had the same tG (60 min) and led to similar average peak-width, the PC

should be expected to be also statistically similar. Indeed, the PC observed for acetonitrile and acetone

were: 391.4 ± 20.7 and 384.2 ± 21.4 for acetonitrile and acetone, respectively.

The chromatograms recorded using the UV detector at 350 nm from both acetonitrile or acetone based

optimised methods are shown in supporting information. This detector responded well for compounds

eluted until ca. 35 min, whereas it was unable to detect mostly of the metabolites eluted after that (the

same behaviour was observed for data acquired at 240 and 254 nm with acetonitrile and 330 with both

organic solvents, see supporting information).

The relative lack of sensitivity of C-CAD detector compared to the UV detector was compensated

by its detection of many non-chromophoric compounds present in our sample at 80 mg/mL (and 20

µL injection volume). As a result, the former allowed the establishment of a much more

comprehensive qualitative fingerprint of the sample.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 14 Journal of Separation Science

In another type of application, LC–C-CAD based methods developed in analytical scale could be

scaled up aiming to the isolation of compounds for posterior identification by spectrometric

techniques, such as NMR spectroscopy and MS. This is a usual procedure in natural products

chemistry field and in this type of application the advantage of employing acetone stead of acetonitrile

should be even bigger from the environmental point of view, since (i) higher consumptions of

solvents are observed in semi-preparative and preparative separations compared to analytical scale

and (ii) the amount of energy required to separate acetone from the desired compound should be

lower than that required to separate acetonitrile (as their boiling points at atmospheric pressure are 56

and 82°C, respectively).

These findings reveal the relevance of the strategy of employing LC coupled to an UV-based

detector in line with a C-CAD detector when a sample previously known to contain chromophoric and

non-chromophoric compounds (or a totally unknown sample) is under investigation. That is because

whereas the latter allows the development of a much more comprehensive metabolite profile based in

a less impacting organic solvent, the former can provide structural information of the chromophoric

compounds which absorbs above the UV cut-off of the solvents employed (330 nm for acetone and

210 for EtOH, for example) if a photodiode array detector (PDA-UV/vis) is available.

Although some authors claim that acetone is incompatible with UV-based detectors, our data

suggest that this solvent could be used to optimise a less comprehensive fingerpint HPLC method for

EELalba based on the outputs recorded with the UV detector employed.

It is possible to suppose that good methods could be developed with acetone and UV detector for

samples containing analytes which absorb above 330 nm or even that acetone might be employed in

the first dimension in 2D-LC analyses without any limitation of this type [6]. In 2D-LC application

the UV detector is usually placed only in the second dimension in which another green organic

solvent with lower UV cut off (e.g. ethanol which presents UV cut off at 210 nm) could be employed

as a mobile phase component [6].

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 15 Journal of Separation Science

3.3 Expanding the comparison between acetonitrile and acetone based methods with

environmental parameters

As discussed above, the methods with acetonitrile and acetone proved to be statistically similar

when parameters related to quality of separation were considered. However, a state of art comparison

should take into consideration not only parameters related to the separation efficiency of the methods,

but also those related to their environmental impacts [8]. For that purpose, we previously proposed a

hybrid metric considering both types of parameters when methods using mobile phases with different

compositions and/or flow rates were under investigation, as follows (Eq. 4) [10]:

GCFR / HPLC-EAT (Eq. 4)

where GCFR is the green chromatographic fingerprinting response used at that time to optimise the

methods[10] and HPLC-EAT is an environmental score calculated for the methods based in safety,

health and environment (SHE) impacts [22].

Since this metric was design to be flexible by allowing the replacement of the response used during

the method optimisation (nominator in Eq. 4), here we have replaced the response GCFR with the

total number of peaks (n) used to develop the methods outlined above.

n / HPLC-EAT (Eq. 5)

Eq. 5 will assume higher values (thus indication a better method) by enhancing n whist decreasing

the SHE impact score given by HPLC-EAT (for the latter the lower, the better, since they indicate a

less impacting method). As the n acetonitrile and acetone bases methods were observed to be 220 and

219, respectively, and the HPLC-EAT scores of them calculated using the free software provided by

Gaber et al. [22] were found to be 131.7 and 100.8., Eq. 5 led to a final score of 1.7 and 2.2,

respectively, thus highlighting the superiority of the latter over the former. That is because the acetone

based method consumed less environmental resources (lower SHE) while providing similar

chromatographic output.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 16 Journal of Separation Science

It is important to notice that the component consumption of energy during the process of separation

that was used by Funari et al. [35] as a parameter to compare chromatographic methods from a more

comprehensive understanding of chromatographic performance was not taken into consideration here.

That was because (i) the consumptions of energy were similar in the analyses with acetonitrile and

acetone (same run time and same temperature of analysis) and (ii) the cumulative energy demands

(CED) relative to the amount of organic solvents consumed during the analyses are negligible in

analytical scale when compared to the consumption of energy observed for the HPLC system itself

[35].

It is also possible to deduce that the acetone-based method is greener than the acetonitrile-based one

from a life-cycle assessment (LCA) perspective. Whereas both method showed the same process

efficiency while consuming the same amount of organic solvents and energy, acetone itself is

classified as greener than acetonitrile by LCA [23]. If no equivalence in performance of separation

were observed for acetonitrile and acetone based methods, this deduction would be not possible and a

calculation of LCA should be performed [36, 37]. A more efficient process implies less feedstock,

less energy and less waste [38]. If the acetonitrile-based method was more efficient than the acetone

based one (which is not the case here), it might counteract the worse LCA and SHE scores of

acetonitrile and acetone when analysed as final products. This point is well summarised by the

definition of Jessop, who states that “the greenest solvent is the solvent that makes your process or

product have the least environmental impact”

(http://www.researchgate.net/profile/Philip_Jessop/topics).

Both methods were also compared by using the Analytical Eco-Scale metric [39]. The acetonitrile
and acetone based methods got total score per analysis of 61 and 65, respectively (Table S7,
supporting information). Thus, the green superiority of the latter was confirmed (the higher the score,
the greener the analysis for this metric [39], just the opposite of HPLC-EAT, where the lower the
score, the greener the analysis [22]). Just like the HPLC-EAT metric, the analytical eco-scale does not
take into consideration parameters of separation performance, but a final score achieved by the
multiplication of number of peaks (or peak capacity or another response where the higher the value,
the better) by the analytical eco-scale should combine separation and environmental performances
should lead to a more comprehensive metric. For the acetonitrile and acetone-based methods
developed in this work, the scores were 13 420 and 14 235, respectively, thus confirming the
superiority of the latter over the former for the same reason explained above for Eq. 5.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 17 Journal of Separation Science

4 Concluding remarks

A comprehensive approach was proposed in this work as an alternative to traditional method

development by considering both parameters of separation and environmental parameters from early

stages of research. From this, it was possible to demonstrate the application of nine of the twelve

principles of GAC proposed by Gałuszka et al. [11]. A minimum quantity of sample to afford

sufficient detectability in the C-CAD detector was achieved (second principle), an integration of the

analytical processes and operations to save energy and reduction of solvents was contemplated by

coupling a HPLC system to two detectors in series (fourth principle), system automisation was

applied to accelerate method developments and a miniaturised technique (SPE) was used for sample

preparation (fifth principle), no sample derivatisation was necessary during the investigation (sixth

principle), no large volume of chemical waste was generated (seventh principle) due to the selection

of optimised experimental designs which allowed simultaneous changes in up to six variables in a

reduced number of experiments and time, multi-analyte and multi-parameter approach were adopted

(eighth principle), although temperatures up to 90°C were tested in this work as a potential tool to

increase separations and to reduce solvent consumption, it was possible to achieve optimum results

for both methods by operating close to room temperature (30°C), thus minimizing energy

consumption (ninth principle). Finally, the harmful acetonitrile was replaced with the much greener

acetone (11th principle) with no loss of separation performance while increasing the safety of the

analyst (12th principle).

This work highlights the importance of using statistics to develop analytical methods and to better

compare performances of solvents and also the importance of updating the concept of “performance”

in LC.

C-CAD was shown to be suitable for fingerprinting complex crude extracts since its relative lack of

sensitivity compared to UV detection was compensated by the injection of more concentrated samples

and by the detection of non-chromophoric compounds, which resulted in a more comprehensive

fingerprint of the sample compared to the chromatograms obtained with UV.

This work was kept as generic as possible to be applicable to other complex matrices.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 18 Journal of Separation Science

Acknowledgements

The authors would like to thank the São Paulo Research Foundation (FAPESP, grant #013/07600-3)

and the Australian Research Council (ARC) for financial support. CSF is the recipient of the grant

#012/15877-7, and EFH is the recipient of an ARC Future Fellowship. The authors also acknowledge

the assistance of Dr Roseli B. Torres, of the Agronomic Institute of Campinas for the plant

identification, and of Prof André Gonzaga dos Santos, of the Faculty of Pharmaceutical Sciences of

the São Paulo State University (FCF-UNESP).

The authors have declared no conflict of interest.

References

[1] Koel, M., Kaljurand, M., Pure Appl. Chem. 2006, 78, 1993–2002.

[2] Armenta, S., Garrigues, S., Guardia, M. d. l., TrAC, Trends Anal. Chem. 2008, 27, 497–511.

[3] Kaljurand, M., Koel, M., in: Guardia, M. d. l., Garrigues, S. (Eds.), Challenges in Green

Analytical Chemistry, Royal Society of Chemistry, Cambridge 2011, pp. 16 8–198.

[4] Tobiszewski, M., Mechlinska, A., Namiesnik, J., Chem. Soc. Rev. 2010, 39, 2869–2878.

[5] Welch, C. J., Wu, N., Biba, M., Hartman, R., Brkovic, T., Gong, X., Helmy, R., Schafer, W., Cuff,

J., Pirzada, Z., Zhou, L., TrAC, Trends Anal. Chem. 2010, 29, 667–680.

[6] Plotka, J., Tobiszewski, M., Sulej, A.M., Kupska, M., Gorecki, T., Namiesnik, J., J. Chromatogr.

A 2013, 1307, 1–20.

[7] Sandra, P., Vanhoenacker, G., David, F., Sandra, K., Pereira, A. S., LCGC Europe 2010, 23, 242–

259.

[8] Raynie, D., Majors, R., LCGC Europe 2011, 24, 78–91.

[9] Namiesnik, J., Environ. Sci. Pollut. Res. 1999, 6, 243–244.

[10] Funari, C. S., Carneiro, R. L., Andrade, A. M., Hilder, E. F., Cavalheiro, A. J., J. Sep. Sci. 2014,

37, 37–44.

[11] Gałuszka, A., Migaszewski, Z., Namieśnik, J., TrAC, Trends Anal. Chem. 2013, 50, 78–84.

[12] Tistaert, C., Dejaegher, B., Vander Heyden, Y., Anal. Chim. Acta 2011, 690, 148–161.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 19 Journal of Separation Science

[13] Jiang, Y., David, B., Tu, P., Barbin, Y., Anal. Chim. Acta 2010, 657, 9–18.

[14] Valcárcel, M., Cárdenas, S., Simonet, B. M., Carrillo-Carrión, C., J. Chromatogr. A 2007, 1158,

234–240.

[15] Wolfender, L. L., Nat. Prod. Commun. 2009, 10, 1417–1430.

[16] Gao, H., Wang, Z., Li, Y., Qian, Z., Front. Med. 2011, 5, 195–202.

[17] Funari, C. S., Castro-Gamboa, I., Cavalheiro, A. J., Bolzani, V. d. S., Quim. Nova 2013, 36,

1605–1609.

[18] Alaerts, G., Dejaegher, B., Smeyers-Verbeke, J., Vander Heyden, Y., Comb. Chem. High

Throughput Screening 2010, 13, 900–922.

[19] Ji, Y. B., Xu, Q. S., Hu, Y. Z., Vander Heyden, Y., J. Chromatogr. A 2005, 1066, 97–104.

[20] Vander Heyden, Y., LCGC Europe 2008, 21, 438–443.

[21] Zhong, X. K., Li, D. C., Jiang, J. G., Curr. Med. Chem. 2009, 16, 3064–3075.

[22] Gaber, Y., Törnvall, U., Kumar, M. A., Ali Amin, M., Hatti-Kaul, R., Green Chem. 2011, 13,

2021–2025.

[23] Henderson, R. K., Jiménez-González, C., Constable, D. J. C., Alston, S. R.,Inglis, G. G. A.,

Fisher, G., Sherwood, J., Binks, S. P., Curzons, A. D., Green Chem. 2011, 13, 854.

[24] Alfonsi, K., Colberg, J., Dunn, P. J., Fevig, T., Jennings, S., Johnson, T. A., Kleine, H. P.,

Knight, C., Nagy, M. A., Perry, D. A., Stefaniak, M., Green Chem. 2008, 10, 31–36.

[25] Welch, C. J., Brkovic, T., Schafer, W., Gong, X., Green Chem. 2009, 11, 1232–1238.

[26] Hutchinson, J. P., Remenyi, T., Nesterenko, P., Farrell, W., Groeber, E., Szucs, R., Dicinoski, G.,

Haddad, P. R., Anal. Chim. Acta 2012, 750, 199–206.

[27] Fritz, R., Ruth, W., Kragl, U., Rapid Commun. Mass Spectrom. 2009, 23, 2139–2145.

[28] Johnson, A. R., Vitha, M. F., J. Chromatogr. A 2011, 1218, 556–586.

[29] Keppel, T. R., Jacques, M. E., Weis, D. D., Rapid Commun. Mass Spectrom. 2010, 24, 6–10.

[30] Khandagale, M. M., Hutchinson, J. P., Dicinoski, G. W., Haddad, P. R., J. Chromatogr. A 2013,

1308, 96–103.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 20 Journal of Separation Science

[31] Dolan, J. W., J. Chromatogr. A 2002, 965, 195–205.

[32] Klein, E. J., Rivera, S. L., J. Liq. Chromatogr. Relat. Technol. 2000, 23, 2097–2121.

[33] Dolan, J. W., Snyder, L. R., Djordjevic, N. M., Hill, D. W., Waeghe, T. J., J. Chromatogr. A

1999, 857, 1–20.

[34] Ferreira, S. L., Bruns, R. E., da Silva, E. G., Dos Santos, W. N., Quintella, C. M., David, J. M.,

de Andrade, J. B., Breitkreitz, M. C., Jardim, I. C., Neto, B. B., J. Chromatogr. A 2007, 1158, 2–14.

[35] Funari, C. S., Carneiro, R. L., Cavalheiro, A. J., Hilder, E. F., J. Chromatogr. A 2014, 1354, 34–

42.

[36] Mercer, S. M., Andraos, J., Jessop, P. G., J. Chem. Educ. 2012, 89, 215–220.

[37] Guinee, J. B., Handbook on Life Cycle Assessment, Kluwer Academic Publishers, Dordrecht

2002.

[38] Anastas, P., Eghbali, N., Chem. Soc. Rev. 2010, 39, 301–312.

[39] Gałuszka, A., Migaszewski, Z. M., Konieczka, P., Namieśnik, J., TrAC, Trends Anal. Chem.

2012, 37, 61–72.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 21 Journal of Separation Science

Figure 1. HPLC–C-CAD fingerprints of EELalba; Column: XBridge BEH, 150 x 4.6 mm; 3.5 µm.

Mobile-phase components: acetonitrile (B) and 1.0% AcOH in H2O (A) (top chromatogram); and

acetone (B) and 0.5% AcOH in H2O (A) (bottom chromatogram). Elutions: isocratic at 5% of B for

1.8 min. followed by a gradient elution from 5 to 100 of B until 61.8 min. Flow rate: 0.85 mL/min;

Temperature of analysis: 30°C; Sample concentration: 80 mg/mL. Injection volume: 20 µL.

Enlargements from 45 to 60 min are shown over each chromatogram.

This article is protected by copyright. All rights reserved.

www.jss-journal.com Page 22 Journal of Separation Science

Table 1
Three factor central composite design with acetonitrile and acetone and respective responses.
Factorsa),b) Number of peaks using:
Runs
X1 (X2) X2 (X3) X4 (X6) Acetonitrile (y1) Acetone (y2)
1 –1 –1 –1 139 141
2 +1 –1 –1 146 184
3 –1 +1 –1 192 134
4 +1 +1 –1 195 162
5 –1 –1 +1 154 140
6 +1 –1 +1 148 197
7 –1 +1 +1 207 141
8 +1 +1 +1 192 156
9 –1.683 0 0 171 112
10 1.683 0 0 184 189
11 0 –1.683 0 113 172
12 0 1.683 0 211 151
13 0 0 –1.683 142 144
14 0 0 1.683 176 155
CP1c) 0 0 0 171 161
c)
CP2 0 0 0 170 159
c)
CP3 0 0 0 185 164
CP4c) 0 0 0 177 162
c)
CP5 0 0 0 167 172
a) The design used to acetone is assessed by considering the factors indicated between brackets.

b) X1: initial percentage of acetonitrile v/v; X2: gradient time; X3: temperature of analysis; X4/X6:
concentration of acetic acid in A v/v. The non-codified levels for each solvent are indicated in Tables
S3 and S6.

c) Central point followed by the replicate number.

This article is protected by copyright. All rights reserved.