Investigation of the atomization mechanism of gold nanoparticles in graphite furnace atomic absorption spectrometry

Accepted Manuscript Investigation of the atomization mechanism of gold nanoparticles in graphite furnace atomic absorption spectrometry

Anja Brandt, Kerstin Leopold PII: DOI: Reference:

S0584-8547(18)30267-2 doi:10.1016/j.sab.2018.10.004 SAB 5534

To appear in:

Spectrochimica Acta Part B: Atomic Spectroscopy

Received date: Revised date: Accepted date:

11 June 2018 22 August 2018 1 October 2018

Please cite this article as: Anja Brandt, Kerstin Leopold , Investigation of the atomization mechanism of gold nanoparticles in graphite furnace atomic absorption spectrometry. Sab (2018), doi:10.1016/j.sab.2018.10.004

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT Investigation of the atomization mechanism of gold nanoparticles in graphite furnace atomic absorption spectrometry Anja Brandt and Kerstin Leopold* Institute of Analytical and Bioanalytical Chemistry, Ulm University, Ulm, Germany ; * Email: [email protected]

Keywords

T

Graphite furnace atomic absorption spectrometry; Nanoparticle analysis; Atomization

SC RI P

mechanism, Signal interpretation Abstract

Recently, graphite furnace atomic absorption spectrometry (GFAAS) has been introduced as a new tool to distinguish between silver ions and nanoparticles using

NU

graphite furnace atomic absorption spectrometry (GFAAS) by evaluation of the newly presented parameters atomization delay (tad) and atomization rate (kat). Moreover, sizing of NPs in aqueous suspensions by GFAAS measurement was shown by

MA

several authors. However, the atomization mechanism of NPs in GFAAS and possible differences to ionic salts introduced into the graphite furnace has not yet

ED

been investigated. In this work, therefore we study the newly introduced parameters and further peak characteristics, like full width at half maximum (FWHM), peak asymmetry and appearance time (tAP) and their concentration-dependent trends

PT

applying ionic gold Au(III) standards as well as 5- to 100-nm-sized gold nanoparticles (AuNPs), respectively. Interpretation of the data helps to enlighten the atomization

CE

mechanism and kinetic of the atom release process of AuNPs in comparison to ionic Au(III). Ionic Au(III) shows a rising trend in tad with increasing gold concentrations,

AC

whereas tad is nearly constant for AuNPs over this concentration range. On the other hand, AAS peaks of ionic Au(III) reveal constant tA P, while AuNPs show a shift in appearance time. Moreover, peak asymmetry differs for ionic Au(III) in comparison to AuNPs. These differences suggest different atomization mechanisms involved in the evaporation of gold atoms introduced into the graphite furnace as either ionic Au(III) solution or AuNP hydrosol.

1

Introduction

Metal nanoparticles are used in a wide field of industrial and consumer products due to their unique chemical and physical properties. Nanoparticles are applied in e.g.

ACCEPTED MANUSCRIPT cosmetics, clothes, domestic and medical devices etc. [1,2]. Gold nanoparticle (AuNP) application is discussed and tested in medicine as anti-cancer drug, drug delivery and imaging agent [3] and as catalyst e.g. in CO oxidation [4]. Because of the increasing use of metal nanoparticles, there is a need of robust analytical methods for determination and characterization of nanoparticles for quality assurance of preparations as well as in real-world samples where NPs occur due to usage and disposal of NP-containing products. For the later purpose, low analyte concentrations

T

in demanding matrices are to be determined and therefore analytical method

SC RI P

development is more challenging. So far, no standardized methods are established for NP analysis in real-world samples. Most commonly applied analytical techniques for size fractionation and determination of NPs in real samples are based on chromatographic separation, like e.g. field flow fractionation [5] and followed by

NU

element-specific detection, e.g. single particle inductively coupled plasma mass spectrometry [6,7]. Only recently graphite furnace atomic absorption spectrometry (GFAAS) has been discussed as a potentially useful technique in this regard. Gagné

MA

et al. [8] were the first who reported on detection of ionic silver and silver nanoparticles (AgNPs) in biological samples using GFAAS. The authors showed in

ED

their work that with increasing nanoparticle size an increasing atomization temperature is required. Thereafter, our group developed a new evaluation strategy of the transient absorbance signals introducing two new parameters, “atomization

PT

delay“ (tad) and “atomization rate” (kat), for discrimination of ionic Ag and AgNPs in food [9,10]. This strategy was successfully applied also for detection of AuNPs and

CE

ionic Au in aqueous suspensions as well as for distinguishing [11] and sizing [12] AuNPs. Lately, this could also be shown for AgNPs [13]. Hence, GFAAS has a high

AC

potential for rapid and easy size and concentration determination of metal nanoparticles.

However, i n order to enlighten the mechanism behind the observed behavior the atomization process of metal NPs in GFAAS has to be investigated in detail. For ionic metal solutions applied in GFAAS this has been comprehensively studied in the last three decades of the past century. Thereby, concentration-depending changes in maximum and appearance time of the signals were reported as well as other peak characteristics, like peak width and symmetry. With regard to the kinetic order for the atom release process McNally and Holcombe [14] examined these characteristics for gold absorbance signals coming from aqueous Au(III) standard solutions. For AuNPs

ACCEPTED MANUSCRIPT or any other metal NPs such investigations have not yet been reported. Therefore, i n this work, concentration-dependent peak characteristics of 5- to 100-nm-sized AuNPs were studied in order to gain insights into the atom release mechanism during atomization step in GFAAS.

2.1

Experimental Preparation of suspensions and solutions

T

2

SC RI P

All suspensions and solutions were prepared using ultra pure water (UPW) from a Milli-Q system (Millipore, Billerica, USA). Commercially available citrate-stabilized Au NP sized 5, 20, 60 and 100 nm (BBI Solutions, Cardiff, UK) were used for all investigations. Gold concentrations of these stock hydrosols were determined after dissolutions of an aliquot in aqua regia by GFAAS to be 51.5 mg L-1, 58.5 mg L-1,

NU

54.8 and 42.9 mg L-1, respectively. Sizes of AuNPs were controlled by evaluation of n>100 NPs depicted in transmission electron microscopy images (JEM-1400, Joel,

MA

Germany). The found mean sizes are 6 ± 1 nm, 19 ± 3 nm, 58 ± 7 nm and 100 ± 12 nm. The stock suspensions were vortexed for 30 seconds prior to taking aliquots for dilution with ultra pure water (UPW) to the desired concentrations. All NP

ED

suspensions were prepared in volumetric polypropylene (PP) flasks. Ionic Au(III) solution was prepared in volumetric glass flask by dilution of standard solution (Merck

PT

KGaA, 1000 mg Au L-1, Darmstadt, Germany) and stabilization with 2 M hydrochloric acid (HCl, p.a., VWR Chemicals, France). For purification, glass and PP flasks were

CE

filled with aqua regia for 24 h. Afterwards glass flasks were rinsed three times and then filled up with UPW until next use. For PP flasks following cleaning procedure

AC

was applied: rinsing three times with UPW, hydrochloric bath (22 vol .-%) for 24 h, rinsing three times with UPW, nitric acid bath (39 vol.-%, HNO3 , AnalaR Normapur®, VWR Chemicals, France) for 24 h, rinsing three times with UPW, filling up with UPW until use. 2.2

GFAAS measurements

A high-resolution continuum source atomic absorption spectrometer ContrAA 600 (Analytik Jena AG, Jena, Germany) equipped with a graphite furnace atomization unit and SSA 600 auto sampler for solid sampling containing a liquid dosing unit was used for GFAAS measurements. As purge and protective gas Argon with a purity of 99.996 % (MTI, Neu-Ulm, Germany) was used. For atomic absorption measurement

ACCEPTED MANUSCRIPT the most sensitive line of Au (242.795 nm) was used and center pixel 101 was evaluated for all peak characteristics. For all experiments pyrolytic graphite -coated solid tubes without dosing holes (Analytik Jena AG) were used. The furnace temperature program is given in Table 1. For background correction iterati ve baseline correction algorithm was used. For each measurement 15 µL of AuNP suspension or ionic Au(III) solution were pipetted on a pyrolytic graphite-coated platform (Analytik Jena AG), which was transferred into the graphite tube. The solutions and

from 75 to 675 pg.

SC RI P

T

suspensions have Au concentrations from 5-45 µg L-1, thus total Au amount ranged

Temperature (°C)

Heating rate -1 (°C s )

Hold (s)

1

Drying I

80

6

20

2

Drying II

90

3

20

3

Drying III

110

5

10

4

Pyrolysis I

350

50

20

5

Pyrolysis II

500

300

10

6

Auto-zero

500

0

5

7

Atomization

2000

1500

3

8

Cleaning

500

4

ED

MA

Description

2450

PT

Step No

NU

Table 1: Graphite furnace program of HR-CS GFAAS for Au analysis

2.3

Signal interpretation strategy

CE

Six characteristics of the obtained atomic absorbance signals were evaluated for all measurements in this work. In addition to the typical parameters evaluated in AAS,

AC

namely peak height, i.e. maximum absorbance, and peak area, i.e. absorbance integrated over time, these are: 

Atomization delay (tad), i.e. the time period from starting the atomi zation step until the peak maximum is found in seconds;



Atomization rate (kat ), calculated as the slope of the polynomial fitted curve at the first inflection point in 1/seconds;



Appearance time (tAP), i.e. the time from starting the atomization process until the absorbance signal exceeds the 10criterion in seconds;



Full width at half maximum (FWHM) of the fitted peak in seconds;

ACCEPTED MANUSCRIPT 

Asymmetry factor (AF) of the fitted peak calculated as the ratio between the width of the falling part (F) and the width of the rising part (R) at half maximum of the main peak of the absorbance signal.

Figure 1 illustrates these parameters for better understanding.

a)

0.20

0.15

0.15

0.10

0.10

0.05 10s 0.00 0

0.20

0.4

tAP

kat

0

1 1.3 1.6 1.9 -0.4

0.05 4

b)

0.10 ½ height 0.05

0.00 1 2 3 Time (s)

FWHM = R+F

0.15

T

tad

SC RI P

Absorbance

0.20

R F

AF = F/R

0.00

0

1 2 3 Time (s)

4

c)

0

1

2 3 Time (s)

4

Figure 1: Illustration of the additionally evaluated peak characteristics of an exemplary gold AAS st

NU

signal: a) Atomization delay (t ad) and appearance time (t AP); b) Atomization rate (k at ), inset: 1 derivative; c) Full width at half maximum (FWHM) and asymmetry factor (AF).

MA

In other contexts, tAP, FWHM and symmetry of absorbance peaks resulting from dissolved metal salts have been studied in GFAAS before [14–17]. While tad and kat have been introduced only recently by our group in order to investigate metal NPs by

ED

GFAAS [10]. Besides, no clear definition can be found in literature on how to determine tAP as the time or temperature whe n atomization is first observed. In this

PT

work, 3 different approaches, i.e. 3-, 10- and 10 % of peak height, were tested in order to find a criterion for defining the appearance of the signal, which is applicable

CE

to both Au species and the complete tested concentration range. Thereby, 10- criterion (analogue to definition of limit of quantification, LOQ) provided the most

AC

robust detection of signal appearance applicable to all species and at all investigated Au concentrations. Consequently, tA P was calculated as the time interval from starting the atomization process until the absorbance signal has reached a value that is calculated from the mean absorbance of a blank sample (empty graphite platform) plus 10 times its standard deviation. Moreover, this is the first study in which dependence on concentration and NP size of all of these peak characteristics is evaluated systematically for aqueous Au(III) solutions and AuNP suspensions.

ACCEPTED MANUSCRIPT 3

Results and Discussion

3.1

Peak height and peak area

Figure 2 presents absorbance given as peak height and peak area, respectively, for ionic Au(III) and AuNPs. Both parameters show for all gold species as expected a

MA

NU

SC RI P

T

linear increase with increasing Au concentration following Lambert-Beer’s Law.

Figure 2: a) Peak height and b) Peak area for ionic Au(III) and AuNPs (5 to 100 nm) in a concentration -1

range from 5 to 45 µg L . Error bars represent ± one standard deviation with n ≥ 4. Found linear

ED

correlations are summarized in Table 2.

Evaluating peak height, slope and interception of the found linear regressions show a

PT

slightly decreasing trend with increasing NP size with the exception of 5-nm-sized AuNPs (see Table 2). However, these differences are statistically not significant, so

CE

that correct determination of Au concentration in AuNP hydrosols can be performed using calibration data obtained with aqueous Au(III) standard solutions and

AC

evaluating peak height.

Table 2: Correlation of concentration with peak height and peak area, respectively, for ionic Au(III) and AuNPs sized 5, 20, 60, or 100 nm, respectively. Errors are ± expanded uncertainty (P = 95 %, N ≥ 14). -1

Au species

Correlation between x = concentration (µg L ) and y = peak height y = peak area (Abss) -5

AuNP 5 nm

-5

y = (338 ± 28) * 10 x + 0.043 ± 0.008 2 (R = 0.922)

y = (240 ± 10) * 10 x + 0.015 ± 0.003 2 (R = 0.979)

-5

-5

AuNP 20 nm

y = (417 ± 30)* 10 2 (R = 0.979)

x + 0.021 ± 0.009

y = (254 ± 11) * 10 x + 0.007 ± 0.003 2 (R = 0.993)

-5

AuNP 60 nm

y = (360 ± 30) * 10 x + 0.020 ± 0.009 2 (R = 0.972)

y = (245 ± 15) * 10 x + 0.008 ± 0.004 2 (R = 0.986)

-5

AuNP 100 nm

y = (331 ± 27) * 10 x + 0.014 ± 0.008 2 (R = 0.974)

y = (224 ± 15) * 10 x + 0.009 ± 0.004 2 (R = 0.981)

-5

-5

ACCEPTED MANUSCRIPT -5

ionic Au(III)

y = (384 ± 30) * 10 x + 0.016 ± 0.009 2 (R = 0.976)

-5

y = (310 ± 21) * 10 x + 0.011 ± 0.006 2 (R = 0.982)

In contrast, evaluation of peak area reveals significantly higher sensitivity of ionic Au(III) than AuNP suspensions, while again a decreasing trend for the slopes with increasing NP size was observed. However, here too, the difference between the slopes of the differently sized AuNPs are - except between 20 and 100 nm - not

T

significant. Incomplete atomization could be a reason for the lower peak area of NPs.

SC RI P

However, when applying higher atomization temperature or longer atomization time the same differences in peak area for AuNPs and ionic Au(III) were observed. Furthermore, a second atomization step performed directly after the first confirms that all Au is atomized for ionic Au(III) as well as for AuNPs. Another reason for the difference in sensitivity could be the presence of hydrochloric acid in the ionic Au(III)

NU

solution, while the AuNP hydrosols are suspended in ultra pure water. Therefore, hydrochloric acid was added to the AuNP hydrosols in order to match the matrix of

MA

ionic Au(III). However, this does not lead to any change in the peak areas for AuNPs in comparison to pure water as matrix. Anyway, peak area and height depend on the temperature program used for pyrolysis and atomization as well as on the furnace

ED

geometry and graphite surface. By changing these parameters in order to optimize quantification of Au, lower peak heights for ionic Au (III) than for AuNP were

PT

obtained, whereas peak area becomes similar for the different Au species. In conclusion, the herein found results regarding peak height and area are valid only for

CE

the applied furnace settings as given in see Table 1, whereas furnace conditions for optimal quantification of Au (in contrast to optimal sizing and speciation of Au) are

3.2

AC

different and may obviously lead to other findings. Atomization delay and atomization rate

The determined tad and kat values for all measurement series are presented in Figure 3. As already described in our previous work [12], tad of ionic Au(III) increases with rising concentration following a logarithmic trend, while the atomization delay of AuNPs is more or less constant over the observed concentration range. The increase of tad in Au(III) measurement was attributed to the formation of Au aggregates during drying and pyrolysis step of the graphite furnace program. Au(III) ions are able to migrate on the graphite platform, conglomerate at imperfections of the surface and thus form small aggregates [18–20]. These aggregates, i.e. in-situ formed NPs, grow

ACCEPTED MANUSCRIPT larger with increasing Au concentration and hence more energy is needed to atomize them. At a concentration of 45 µg L-1 Au(III) tad reaches the same value as 5-nmsized AuNPs. At concentrations higher than that tad remains constant (see Figure S1 (Appendix), suggesting

that equilibrium between aggregate

formation and

evaporation i.e. atomization is reached. At a concentration as low as 5 µg L-1 a preceding small peak occurs before the distinct main peak in the absorbance signal of Au(III) (see Figure 4a), which indicates the presence of single Au atoms. With

T

higher concentrations this first peak disappears, implying that all Au atoms form

SC RI P

aggregates (see Figure 5a). On the other hand, this pre-peak becomes dominant when the pyrolysis step is omitted and thus migration of Au atoms on the graphite surface is limited to a minimum before atomization (see Figure 4b). This is in good agreement with the observation that migration of metal atoms on graphite surface

NU

starts at a temperature of about 300°C [18]. Furthermore, the broad shoulder occurring at t~3s for all Au species disappears completely under these conditions. This tailing peak has been described in literature extensively and is attributed to the

MA

migration of gold into in-situ formed pyrolytic graphite causing a delay in atomization [21–23]. Omitting pyrolysis step - in which in this study mainly the citrate -coating of

ED

the NPs is decomposed - apparently also inhibits this process. Anyway, as to be expected, signals are shifted to later times with the modified furnace program, because atomization starts directly after drying, i.e. at a temperature of 150°C

PT

instead of 500°C. Nevertheless, size-dependency of tad is observed with both furnace programs: Signals for Au(III) appear first followed by 5-nm AuNPs and finally 100-nm

CE

AuNPs. Moreover, the time shifts between 1st and 2nd Au(III) peak are the same for both furnace programs (t=0.7s) as well as for the tad differences between 5- and

AC

100-nm sized AuNPs (t=0.5s). These results indicate the robustness and reproducibility of the approach and indicate that possible incomplete decomposition of the coating does not affect the observed trends in tad.

2.4 a)

2.3 2.2

tad (s)

2.1 2.0 1.9 1.8 1.7

T

1.6 0

10 20 30 40 Concentration (µg L-1)

50

2.3 b)

45 µg/L 35 µg/L 25 µg/L 15 µg/L 5 µg/L

2.1

tad (s)

2.0

45 µg/L 35 µg/L 25 µg/L 15 µg/L 5 µg/L

NU

2.2

1.9 y = 0.15ln(x) + 1.5026

y = 0.15ln(x) + 1.5026

0.8 c)

PT

0.7 0.6

0.8

kat (s-1)

0.6 0.5 0.4

0.4 0.3 0.2

0.3

0.1

0.2

0

0.1

0

10

0 0

10

20

AuNP 5 nm AuNP 20 nm AuNP 60 nm AuNP 100 nm ionic Au(III)

CE

0.5

0.7

100

ED

10 NP size (nm)

MA

1.8 10 100 1.7 NP size (nm) 1.6 1

20

AuNP 5 nm AuNP 20 nm AuNP 60 nm AuNP 100 nm ionic Au(III)

30

Concentration (µg

30

40

Concentration (µg

50

SC RI P

1.5

AC

at

ACCEPTED MANUSCRIPT

40

50

L-1)

L -1)

Figure 3: a & b) Atomization delay (t ad) and c) Atomization rate (k at ) for ionic Au(III) and AuNPs (5 to -1

100 nm) in a concentration range from 5 to 45 µg L . Error bars represent ± one standard deviation with n ≥ 4. Found linear correlations are summarized in Table S 1-3 (Appendix).

0.05 ACCEPTED MANUSCRIPT AuNP 5 nm 0.04 0.05

a)

2.25

0.02

0.03

0.02

AuNP 5 nm

0.04 0.02

AuNP 100 nm

2.40

0.03 0.01

2.97

0.02 0 0

0.01

0.89

0

0.01

0 0

1

2 3 Time (s)

4

5

0

2.04

1

2 3 Time (s)

SC RI P

0.01

Au(III)

T

0.04

Absorbance

Absorbance

0.04

1.60

Au(III)

0.03

1.30

1.78

0.03

b)0.05

Absorbance Absorbance

0.05

AuNP 100 nm

0 1 0

2 1 3 2 4 3 Time (s) Time (s)

4

5

5 4

5

Figure 4: Averaged absorption signals (n≥3) of Au(III), 5-nm AuNPs, and 100-nm AuNPs, respectively, -1

NU

(c(Au)= 5 µg L ) and corresponding t ad values applying (a) the graphite furnace program as given in Table 1 and (b) a modified temperature program without pyrolysis where atomization starts immediately after drying (Changes to temperature program in detail: Step no. 4 was set to 150°C,

MA

-1

heating rate: 50°C s , hold time: 10 s; Step no.5 was omitted; Step no.6 was set to 150°C)

In contrast to tad of Au(III), tad values of AuNPs are less affected by gold

ED

concentration. While 5-nm-sized AuNPs show a constant tad value of 1.76 ± 0.02 s over the investigated concentration range, the larger AuNPs seem to slightly

PT

decrease in tad by about 0.08 s with increasing concentration from 5 to 45 µg L -1 (see Figure 3). In our previous investigation [12], in contrast, a slight increase in

CE

atomization delay for 20-nm-sized AuNPs was observed. We therefore assume that these minimal changes in tad do not reflect a concentration dependency, but are more

AC

likely to be attributed to slight temperature drifts of the graphite furnace during the measurement day. Hence, we conclude that citrate-stabilized AuNPs do not in-situ form larger aggregates in the graphite furnace. Moreover, the logarithmic correlation between atomization delay and AuNP size, which can be used for size calibration [12], was again confirmed in these measurement series and is shown here to be independent of Au concentration (see Figure 3b). The found correlations for each concentration are given in the supporting information in Table S2 (Appendix). Using those to calculate the size of the in-situ formed aggregates of the ionic Au(III) results in 1.72 ± 0.48 nm at 5 µg L-1 and 8.47 ± 3.06 nm at 45 µg L-1, which is in good agreement with literature [19].

ACCEPTED MANUSCRIPT Figure 3c) presents the atomization rates kat for ionic Au(III) and AuNPs. As to be expected kat increases with increasing concentration for all investigated gold species. Moreover, kat values are generally higher for smaller NPs and lower for larger NPs. This effect can be explained by the particle number concentration of a 5-nm-NP suspension, which is about 4 orders of magnitude higher than that of a 100-nm-NPs suspension at the same metal concentration. I.e. a larger total surface area occurs and accordingly more Au atoms can be released at once under the same conditions,

T

resulting in a higher kat for smaller NPs. Intriguingly, kat values of ionic Au(III) – or in-

SC RI P

situ formed aggregates of 2-8 nm in diameter – are in the range of the 100-nm-sized AuNPs, i.e. do not follow this trend at all. An explanation might be that the aggregates formed from ionic Au(III) have a different shape than comparable AuNPs from citrate-stabilized suspensions. These are typically spheres while the shape of

NU

the in-situ formed aggregates from ionic Au(III) is unknown and is most probably more or less semi-spherical. Such droplet-like shapes have significantly lower

Appearance time, full width at half maximum and asymmetry factor

ED

3.3

MA

surface area and may therefore show different atom release mechanism.

Beside the above-mentioned atomization delay and rate further characteristics of the peak shape may give an indication of kinetic order of the atom desorption process.

PT

McNally and Holcombe [19] had described in their study that the absorbance signals in GFAAS follow different changes in shape for different release orders when

CE

increasing the gold concentration. These authors have also simulated absorbance signals for different orders of release with increasing concentrations. For zero-order

AC

or “pseudo zero-order” constant appearance time and generation rates at the increasing part of the signal are observed together with a shift of the absorbance maxima to higher times. In contrast, the absorbance signal of a first-order process shows no time shift of the peak maximum with increasing concentration, but a shift of appearance time to earlier times. The characteristics of absorbance signals with release orders between 1 and 0 are in-between these. Moreover, FWHM increases with increasing order of release. For atomization of gold from an ionic Au(III) standard with initial masses ranging from 0.06 to 2 ng McNally and Holcombe [19] found a release order of ⅓. The same order of release was found in this study for Au(III) by comparing the simulated and the experimentally determined absorbance

ACCEPTED MANUSCRIPT signal. Interestingly, however, when comparing the absorbance signals of ionic Au(III) with those of AuNP it is obvious, that they show different behaviour with increasing concentration (Figure 5a,b). The maxima of the absorbance signals of ionic Au(III) are shifted to later times (higher tad, see Figure 3a) with increasing concentration and the appearance time stays constant (Figure 5c). In contrast, tad of AuNPs remains constant (Figure 3a) while tA P decreases significantly with increasing concentration (Figure 5c). tA P of AuNPs follow a logarithmic decrease with very good

T

correlation coefficients R2 = 0.959 - 0.988 (see Table S1, Appendix) while ionic

SC RI P

Au(III) has a constant appearance time of tA P = 1.13 ± 0.02 s. Increasing the Au concentration of NP suspensions leads also to a higher number of particles in the graphite furnace, while in the case of the ionic Au(III) an increasing concentration results in this concentration range only in larger aggregates. Thus, in the NP

NU

suspension total surface area rises with higher concentration and so the probability that an atom is desorbed becomes higher, i.e. appearance time decreases. Correlation of NP size with tAP is presented in Figure 5d). tA P increases linearly with

MA

the logarithmic size of NPs, however, due to the high concentration dependency of this parameter tA P is – in contrast to tad – less useful for size calibration. In

ED

conclusion, the nearly constant tad and the decreasing tAP of AuNPs signals indicate a first-order release mechanism. In contradiction to this, however, FWHM shows higher values for ionic Au(III) than for AuNPs (see Figure 5e). Moreover, the absorption

PT

signals of a first-order release mechanism are supposed to be symmetric, but signals of AuNPs sized 20, 60 and 100 nm are asymmetric (see Figure 5f). Table 3

CE

summarizes the peak characteristics observed in this study compared to the theoretical/simulated ones for different desorption orders.

AC

Table 3: Peak characteristics with increasing concentration and corresponding kinetic order of the atom desorption process.

Theoretical/simulated behaviour of Au(III) (according to McNally and Holcombe [19]) Kinetic tad tAP FWHM AF order

Parameters observed in this work Au Species tad

tAP

FWHM

AF

0

Increase

Constant

Lower

≠1

Au(III)

Increase

Constant

Higher

>1

1

Constant

Decrease

Higher

=1

AuNPs

Constant

Decrease

Lower

<1

ACCEPTED MANUSCRIPT

tAP

Au (III)

0.2 Absorbance

Absorbance

0.004

0 0.5

1.5

0

2.0

1

2 3 Time (s)

4

5

1.8 1.7

2.2 0.1 2.1

0 1

tAP (s) tAP (s)

1.5

1.51.5 1.41.4

45 µg/L 35 µg/L 25 µg/L 15 µg/L 5 µg/L

1

2 3 Time (s)

4

5

y = 0.15ln(x) + 1.5026 10 NP size [nm]

100

1.2 1.1 1.0 0 0.70

10 20 30 40 Concentration (µg L-1)

e)

0.7

1.4

0.6

1.3

ED

0.65

1.3 1.3 1.2 1.2 1.1 1.1 1.0 1.0 50 1 1 0.8

MA

1.3

0.5

25 µg/L µg/L 35 µg/L 35 µg/L 45 µg/L 45 µg/L

0.55

PT

0.50 0.45

0.30 0

10 20 30 40 Concentration (µg [µg LL-1-1]) Concentration

AC

0.35

CE

0.40

100 100

10(nm) NP size

f)

NP size (nm)

1.2

AuNP nm AuNP 5 5nm AuNP AuNP20 20nm nm AuNP AuNP60 60nm nm AuNP 100 AuNP 100nm nm ionic Au(III)

0.4

AuNP1.1 5 nm

0.3AuNP 20 nm AF

0.60

10

5 µg/L 5 µg/L 15 µg/L µg/L

NU

1.4

2

2.0

1.61.6

kat (s-1)

tAP (s)

1.6

FWHM [s]

0.004

2.0 1.7 2.0 d) 1.9 1.9 1.6 1.81.8 1 1.71.7

c)

1.9

2.3

1.9 0 0 1.8

0

tAP

Au NP 60 nm

tad [s]

0.1

0.008

T

0.2

b)

0.008

SC RI P

a)

1.0

0.2AuNP 60 nm

ionic Au(III)

AuNP0.9 100 nm

0.1ionic Au(III) 0.8

0 0 0.7 10

20

30

40

50

L-1)

50

0.6 Concentration (µg 0 10 20 30 40 Concentration (µg Concentration (µg L-1) L-1)

50

Figure 5: Exemplary absorption spectra of a) Ionic Au(III) and b) 60-nm-sized AuNPs and peak characteristics with increasing Au concentrations; c,d) Appearance time; e) Full width at half maximum; f) Asymmetry factor. Error bars represent ± one standard deviation with n≥2. Found linear correlations are summarized in Table S 1-3 (Appendix).

Moreover, it has been suggested to use the order of release to determine the shape of aggregates formed in the graphite furnace [19]. A release order of ⅓, as observed for ionic Au(III), indicates hemispherical droplets with a contact angle of 90°, which is in good agreement with our assumptions described in section 3.2. Concentration dependency of tAP and tad values for AuNPs, in contrast, follow the behaviour of a 1st

ACCEPTED MANUSCRIPT order release kinetic. Typically, this is expected for evaporation of dispersed metal atoms from the graphite platform having strong metal-graphite interaction. However, this is not the case here, since the shift of tad and tA P with increasing NP size clearly indicates that the introduced NPs stay intact until enough energy is provided to vaporize the particles at once without prior disaggregation into individual atoms. This behaviour can also explain the signal asymmetry for AuNPs. While evaporation of dispersed metal atoms with strong metal-graphite interaction leads to symmetric

T

signal shape (AF=1), for AuNPs an AF < 1 was observed. The symmetry of the

SC RI P

absorbance signal derives from supply and loss of free atoms in the optical pathway, i.e. from vaporization of the analyte and diffusion of the atomic vapour. Though diffusion immediately starts after vaporization, the diffusion rate is assumed to stay constant for a given furnace and thus atomic vapour generation is dominant in the

NU

arising part of the signal, while in the falling part diffusion dominates. When the absorbance signal drops down to a value of 1/e all analyte is already evaporated and only diffusion takes place [15,16]. In case of AuNPs with AF<1, the time in which

MA

diffusional loss of free atoms is dominant (decreasing signal part) is much shorter than the time of the increasing signal part, when vaporization of atoms is dominant.

ED

At the same time, signal width (measured as FWHM) is lower for AuNPs than for Au(III). Since diffusion rate of free atoms in the given furnace geometry can be assumed to be independent from the introduced Au species, this indicates that

PT

evaporation rate of AuNPs is higher than for Au(III). I.e. in case of Au(III) more time is needed to reach the state when only diffusion occurs and no atoms are vaporized

CE

anymore. This can be explained by the presence of in-situ generated agglomerates in a wider size range and thus larger aggregates are atomized later while smaller

AC

aggregates - the main mass of the analyte - is already atomized. This also reflects that the parameter of ‘atomization delay’, i.e. a defined point of time, gives modal size information. Moreover, when comparing release of Au(III) and AuNPs it has to be considered that Au(III) ions have to be reduced prior to atomization, while AuNPs are already in oxidation state zero. In conclusion, different release mechanism for ionic Au(III) and AuNPs are clearly indicated by the presented data. While for Au(III) all signal characteristics and concentration dependent trends are in agreement with literature data determining a release order of ⅓, the release order of AuNPs cannot be clearly determined with the present data set.

ACCEPTED MANUSCRIPT 4

Conclusion

This study gives interesting insights in the atom release mechanism of ionic Au(III) and AuNPs in GFAAS. Concentration-dependent measurements of ionic Au(III) and 5- to 100-nm-sized AuNPs were performed and different parameters like atomization delay (tad ) and rate (kat), appearance time (tA P), full with at half maximum (FWHM), and asymmetry factor (AF) were evaluated. Thereby, clearly different characteristics of the according absorbance signals of ionic Au(III) and AuNP have been observed.

T

Firstly, quantification of AuNP suspensions against ionic Au( III) solution is only

SC RI P

correct using peak height for evaluation, since peak area depends on the gold species, i.e. is lower for AuNPs of larger sizes. Secondly, there is a logarithmic increase of tad of ionic Au(III) with increasing concentration through the in-situ formation of larger Au aggregates while it remains constant for AuNPs. Moreover,

NU

comparison of kat of ionic Au(III) with AuNPs indicates that the in-situ formed aggregates have a smaller total surface area in contrast to AuNPs introduced into the graphite furnace. Furthermore, tAP of ionic Au(III) is concentration-independent and a

MA

release order of ⅓ for ionic Au(III) solution was determined. In contrast, tad of AuNPs is not concentration-dependent while tAP is shifted to lower times with increasing

ED

concentration. This indicates a first-order release process for AuNPs. However, asymmetry of the absorbance signals of both gold species and generally lower FWHMs for AuNPs were observed. This is in contradiction to simulated GFAAS

PT

signals of first-order release process. Hence, the release mechanism of AuNPs is obviously different from that of Au(III), however, further investigations are necessary

CE

to clearly determine its release order.

AC

Funding

This work was supported by the German Research Foundation [Deutsche Forschungsgemeinschaft; DFG; grant number LE 2457/8-1].

Appendix A. Supplementary data Supplementary data to this article can be found online at doi:... References [1]

A. Nel, Toxic Potential of Materials at the Nanolevel, Science. 311 (2006) 622 –627.

doi:10.1126/science.1114397.

ACCEPTED MANUSCRIPT [2]

S.

Greßler,

R.

Fries,

Lebensmittelkontaktmaterialien

-

Nanosilber

Produkte,

in

Kosmetika,

gesundheitliche

und

Hygieneartik eln regulatorische

und

Aspekte,

Bundesminestrium für Gesundheit, Österreich, 2010. [3]

E.C. Dreaden, A.M. Alkilany, X. Huang, C.J. Murphy, M.A. El-Sayed, The golden age: gold

nanoparticles for biomedicine, Chem Soc Rev. 41 (2012) 2740–2779. doi:10.1039/C1CS15237H. [4]

M.-C. Daniel, D. Astruc, Gold nanoparticles: assembly, supramolecular chemistry, quantum -

size-relat ed properties, and applications toward biology, catalysis, and nanotec hnology, Chem. Rev.

T.K. Mudalige, H. Qu, S.W. Linder, An improved methodology of asymmetric flow field flow

SC RI P

[5]

T

104 (2004) 293–346.

fractionation hyphenated with inductively coupled mass spectrometry for the determination of size distribution of gold nanoparticles in dietary supplements, J. Chromatogr. A. 1420 (2015) 92–97. doi:10.1016/j.chroma.2015.09.091. [6]

M. Witzler, F. Küllmer, A. Hirtz, K. Günther, Validation of Gold and Silver Nanoparticle

NU

Analysis in Fruit Juices by Single -Particle ICP -MS without Sample Pretreatment, J. Agric. Food Chem. 64 (2016) 4165–4170. doi:10.1021/acs.jafc.6b01248.

T.P.J. Linsinger, R. Peters, S. Weigel, International interlaboratory study for sizing and

MA

[7]

quantification of Ag nanoparticles in food simulants by single-particle ICPMS, Anal. Bioanal. Chem. 406 (2014) 3835–3843. doi:10.1007/s00216-013-7559-9. F. Gagné, P. Turcotte, C. Gagnon, Screening test of silver nanoparticles in biological samples

ED

[8]

by graphite furnace-atomic absorption spectrometry, Anal. Bioanal. Chem. 404 (2012) 2067–2072.

[9]

PT

doi:10.1007/s00216-012-6258-2.

N.S. Feichtmeier, N. Ruchter, S. Zimmermann, B. Sures, K. Leopold, A direct solid sampling

analysis method for the detection of silver nanoparticles in biological mat rices, Anal. Bioanal. Chem.

[10]

CE

408 (2016) 295–305. doi:10.1007/s00216-015-9108-1. N.S. Feichtmeier, K. Leopold, Detection of silver nanoparticles in parsley by solid sampling

AC

high-resolution–continuum source atomic absorption spectrometry, Anal. Bioanal. Chem. 406 (2014) 3887–3894. doi:10.1007/s00216-013-7510-0. [11]

M. Resano, E. Garcia-Ruiz, R. Garde, High-resolution continuum source graphite furnace

atomic absorption spectrometry for the monitoring of Au nanoparticles, J Anal Spectrom. 31 (2016) 2233–2241. doi:10.1039/C6JA00280C. [12]

K. Leopold, A. Brandt, H. Tarren, Sizing gold nanoparticles using graphite furnace atomic

absorption spectrometry, J Anal Spectrom. 32 (2017) 723–730. doi:10.1039/C7JA00019G. [13]

T. Panyabut, N. Sirirat, A. Siripinyanond, Use of electrothermal atomic abso rption

spectrometry for size profiling of gold and silver nanoparticles, Anal. Chim. Acta. 1000 (2018) 75 –84. doi:10.1016/j.aca.2017.09.032. [14]

Y. Xiu-ping, N. Zhe-ming, Y. Xiao-tao, H. Guo-qiang, An approach to the determination of the

ACCEPTED MANUSCRIPT kinetic paramet ers for atom formation in electrothermal atomic absorption spectrometry, Spectrochim. Acta Part B At. Spectrosc. 48 (1993) 605–624. doi:10.1016/0584-8547(93)80064-2. [15]

D. Rojas, W. Olivares, A method for the determination of the kinetic order and energy o f the

atom formation process in electrothermal atomization atomic absorption spectromet ry (E TA -AAS), Spectrochim. Acta Part B At. Spectrosc. 47 (1992) 387–397. doi:10.1016/0584-8547(92)80033-D. [16]

N.S. Thomaidis, E.A. Piperaki, Determination of the kinet ic parameters for the electrothermal

atomization of gold with and without chemical modifiers, Spectrochim. Acta Part B At. Spectrosc. 54

V. Slaveykova, S. Manev, D. Lazarov, A pplication of the Kelvi n equation to vaporization of

SC RI P

[17]

T

(1999) 1303–1320. doi:10.1016/S0584-8547(99)00078-6.

silver and gold in electrothermal atomic absorption spectrometry, Spectrochim. Acta Part B At. Spectrosc. 50 (1995) 1725–1732. doi:10.1016/0584-8547(95)01334-2. [18]

R.W. Fonseca, J. McNally, J. A. Holcombe, Mechanisms of vaporization for silver and gold

using electrot hermal atomization, Spectrochim. Acta Part B At. Spectrosc. 48 (1993) 79 –89.

[19]

NU

doi:10.1016/0584-8547(93)80009-J.

J. McNally, J.A. Holcombe, Existence of Microdroplets and Dispersed Atoms on the Graphite

[20]

MA

Surface in Electrothermal Atomizers, Anal. Chem. 59 (1987) 1105–1112. E. Ganz, K. Sattler, J. Clarke, Scanning tunneling microscopy of Cu, Ag, Au and Al adatoms,

small clusters, and islands on graphite, Surf. Sci. 219 (1989) 33–67. E. Iwamoto, M. Itamot o, K. Nishioka, S. Imai, Y. Hayashi, T. Kumamaru, E ffects of Conditions

ED

[21]

for Pyrolysis of Ascorbic Acid as a Chemical Modifier on the Vaporization Mechanism of Gold in

doi:10.1039/a701659j.

V. Majidi, R.G. Smith, R.E. Bossio, R.T. Pogue, M.W. McMahon, Observation of pre -

CE

[22]

PT

Electrothermal Atomic Absorption S pectrometry, J. Anal. At. Spectrom. 12 (1997) 1293 –1296.

atomization events on electrothermal at omizer surfaces, Spectrochim. Acta Part B At. Spectrosc. 51 (1996) 941–959. doi:10.1016/0584-8547(96)01477-2. J.G. Jackson, R.W. Fonseca, J.A. Holcombe, Migration of Ag, Cd and Cu into highly oriented

AC

[23]

pyrolytic graphite and pyrolytic coated graphite, Spectrochim. Acta Part B At. Spectrosc. 50 (1995) 1837–1846. doi:10.1016/0584-8547(95)01377-6.

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5