Spectrochimica Acta Part B 129 (2017) 42–48

Contents lists available at ScienceDirect

Spectrochimica Acta Part B

journal homepage: www.elsevier.com/locate/sab

High purity polyimide analysis by solid sampling graphite furnace atomic

absorption spectrometry
Rafael F. Santos, Gabriel S. Carvalho, Fabio A. Duarte, Rodrigo C. Bolzan, Erico M.M. Flores ⁎
Departamento de Química, Universidade Federal de Santa Maria, 97105-900 Santa Maria, RS, Brazil

a r t i c l e i n f o a b s t r a c t

Article history: In this work, Cr, Cu, Mn, Na and Ni were determined in high purity polyimides (99.5%) by solid sampling graphite
Received 21 October 2016 furnace atomic absorption spectrometry (SS-GFAAS) using Zeeman effect background correction system with
Received in revised form 14 January 2017 variable magnetic field, making possible the simultaneous measurement at high or low sensitivity. The following
Accepted 16 January 2017
analytical parameters were evaluated: pyrolysis and atomization temperatures, feasibility of calibration with
Available online 20 January 2017
aqueous solution, linear calibration range, sample mass range and the use of chemical modifier. Calibration
Keywords:
with aqueous standard solutions was feasible for all analytes. No under or overestimated results were observed
Graphite furnace atomic absorption spectrometry and up to 10 mg sample could be introduced on the platform for the determination of Cr, Cu, Mn, Na and Ni. The
Trace elements relative standard deviation ranged from 3 to 20%. The limits of detection (LODs) achieved using the high sensi-
Polymer analysis tivity mode were as low as 7.0, 2.5, 1.7, 17 and 0.12 ng g−1 for Cr, Cu, Mn, Na and Ni, respectively. No addition
Contaminants of chemical modifier was necessary, except for Mn determination where Pd was required. The accuracy was eval-
Polymers uated by analyte spike and by comparison of the results with those obtained by inductively coupled plasma op-
Solid sampling tical emission spectrometry and inductively coupled plasma mass spectrometry after microwave-assisted
Electrothermal atomic absorption spectrometry
digestion in a single reaction chamber system and also by neutron activation analysis. No difference among the
results obtained by SS-GFAAS and those obtained by alternative analytical methods using independent tech-
niques. SS-GFAAS method showed some advantages, such as the determination of metallic contaminants in
high purity polyimides with practically no sample preparation, very low LODs, calibration with aqueous stan-
dards and determination in a wide range of concentration.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction modify the electrical properties of PI that is undesirable to further appli-

Polyimide (PI) is a polymeric material that presents important char-
acteristics, such as thermo-oxidative stability, high mechanical strength,
resistance to solvents and electrical properties [1–3]. This polymer has
been widely employed for applications at high temperature [4] as di-
electric for electronic components [5], frameworks [6], membranes for
gas separation [7], composites [8], crystal liquid displays [9], electrolu-
minescent diodes [10], encapsulating implant devices [11], among
others.
Some works have shown the influence of elemental impurities (e.g.,
Cr, Cu, Mn, Na and Ni) in polymers that even at trace levels can catalyze
oxidation processes causing irreversible decomposition, affecting the
technological and thermal properties as well as the performance of
the final product [12,13]. For PI coatings, Bellucci et al. [14] studied the
corrosion of this polymer due to the presence of elemental impurities.
In addition to the oxidation process, the presence of metallic ions may
⁎ Corresponding author.
E-mail address: ericommf@gmail.com (E.M.M. Flores).
cation as semiconductor and dielectric material [13,15,16].
According to the association of Semiconductor Equipment and Ma-
terials International (SEMI), a well defined control of elemental impuri-
ties in polymers used for semiconductor manufacturing as PI is required.
In this sense, SEMI recommends a protocol (SEMI F48-0600) with sev-
eral steps for sample preparation and further elemental determination
by spectrometric techniques [17]. Inductively coupled plasma optical
emission spectrometry (ICP-OES) and inductively coupled plasma
mass spectrometry (ICP-MS) are recommended by SEMI F48-0600,
but these techniques require an efficient sample preparation step in
order to avoid/minimize interferences [18]. However, most sample
preparation methods have some disadvantages, such as prolonged
total analysis times and risk of contamination or analyte losses, mainly
for open systems [18–20]. In addition, the requirement of high amounts
of concentrated acids can increase the limits of detection (LODs) [18,19].
It is important to mention that the method described by SEMI F48-0600
for the determination of Cr, Cu, Mn, Na and Ni is not recommended for
polymers where LODs lower than 1 μg g−1 are required as in the case for
PI [17].

http://dx.doi.org/10.1016/j.sab.2017.01.005
0584-8547/© 2017 Elsevier B.V. All rights reserved.
 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48 43

Alternatively to analytical techniques that requires a sample prepa- Table 1

ration step, methods based on solid sampling (SS) are considered as a Instrumental parameters applied for the determination by SS-GFAAS, ICP-OES and ICP-MS.

powerful and reliable approach for trace elements determination due Analyte SS-GFAAS ICP-OES ICP-MS
to the possibility of achieving low LODs, low reagent consumption, rel- (nm) (m/z)
atively low sample preparation and small sample mass required for the Wavelength Spectral HCL current
(nm) bandpass (nm) (mA)
analysis [21–23]. Some techniques based on the direct determination
52
such as neutron activation analysis (NAA) [24], slurry sampling graphite Cr 357.9 0.8 4.0 205.552 Cr
63
Cu 324.8 0.8 2.0 327.396 Cu
furnace atomic absorption spectrometry [25], particle-induced X-ray 55
Mn 279.5 0.2 7.0 257.610 Mn
emission and Rutherford backscattering spectrometry [26], have been Na 589.0 0.5 3.0 589.592 ND
used for elements determination in PI samples. In addition, several Ni 232.0 0.2 5.0 232.003 58
Ni
works have shown the feasibility of solid sampling graphite furnace ND = not determined.
atomic absorption spectrometry (SS-GFAAS) as one of the most power-
ful analytical technique for elemental determination in solid samples
due to its simplicity, low LODs and low time consumption [27–31]. Re- between zero (switched off) and maximum magnetic field applied
cent publications have demonstrated the feasibility of SS-GFAAS for the (0.8 T) [38]. For samples with high analyte content, the absorbance
determination of Cr, Cu, Mn, Na and Ni in various advanced materials, was measured applying a variable magnetic field between 0.6 T and
such as raw material for polyurethane [30], fluoropolymers [31], alumi- 0.8 T, providing a reduction on the sensitivity [38].
num nitride [32], pharmaceutical grade barium sulfate [33], and yttrium Pyrolytically coated graphite tubes and platforms (Analytik Jena)
oxyorthosilicate [34]. However, the use of this technique for the deter- were used throughout. A microbalance (model M2P, Sartorius,
mination of these analytes in PI samples, has not yet been described in Göttingen, Germany) with resolution of 1 μg was used for sample
literature, although it is the most simple and low cost technique com- weighing. The optimized heating programs for all analytes determined
paring to those cited above. by SS-GFAAS are shown in Table 2.
Due to the low elements concentration in PI, it is evident that there is Polyimide residues obtained after pyrolysis step were analyzed in an
still a need for accurate methods for its analysis. However, PI is very infrared spectrometer (model Spectrum 400 FT-IR/FT-NIR,
hard to digest, due its chemical inertness, being SS-GFAAS an efficient PerkinElmer-SCIEX, Waltham, EUA) with deuterated triglycine sulfate
alternatively for the analysis of this polymer. In this sense, SS-GFAAS (DTGS) detector and KBr beam splitter. Spectra were collected from
with background correction system based on Zeeman effect was applied 4000 to 650 cm−1, with 16 scans. This instrument is equipped with an
in the present work for the determination of Cr, Cu, Mn, Na and Ni. This attenuated total reflection (ATR) universal sampling accessory supplied
background correction system allows the simultaneous monitoring of with a top plate of ZnSe crystal.
different sensitivities conditions (high and low sensitivity), being possi- For results comparison, PI samples were digested in an ultra-high
ble to determine elements at high or low concentration, without chang- pressure single reaction chamber (SRC) system (model UltraWave™,
ing the wavelength or gas flow-rate during atomization. Therefore, the Milestone, Sorisole, Italy) equipped with five quartz vessels (maximum
suitability of low or high sensitivity, as well as the pyrolysis and atomi- volume, pressure and temperature of 40 mL, 199 bar and 300 °C,
zation curves for sample and aqueous standard solution, the linear cal- respectively).
ibration range, the influence of sample mass and the use of chemical An ICP-OES instrument (model Spectro Cirus CCD, Spectro Analytical
modifier was investigated. The accuracy was evaluated by results com- Instruments, Kleve, Germany) with a cross-flow nebulizer coupled to a
parison with those obtained by spike recovery tests and determination double pass Scott type spray chamber was used for determination of
by ICP-OES and ICP-MS after sample digestion and also by NAA. analytes in digested samples. The radiofrequency (RF) power was set
at 1400 W and the argon flow-rate was set at 12, 1.0 and
2. Experimental 1.00 L min− 1 for principal, auxiliary and nebulizer gas, respectively.
Chromium, Cu, Mn and Ni were also determined by ICP-MS using a
2.1. Instrumentation and operating conditions model Elan DRC II™ spectrometer (PerkinElmer-SCIEX, Thornhill, Cana-
da) equipped with a concentric nebulizer (Meinhard Associates, Gold-
The determination of Cr, Cu, Mn, Na and Ni was carried out using a en, USA), a cyclonic spray chamber (Glass Expansion, Inc., West
graphite furnace atomic absorption spectrometer (model AAS ZEEnit Melbourne, Australia) and a quartz torch with a quartz injector tube
60, Analytik Jena, Jena, Germany) equipped with a transversely heated (2 mm i.d.). The RF power was set at 1300 W and the argon flow-rate
graphite tube atomizer and a background correction system based on was set at 15, 1.2 and 1.08 L min−1 for principal, auxiliary and nebulizer
Zeeman effect. This correction system allows changing the magnetic gas, respectively.
field strength from 0 to 1.0 T. The spectrometer was operated with the Results for Cr, Mn and Na were also compared with those obtained
main analytical lines of the analytes for high and low sensitivity by NAA analysis performed by the Nuclear and Energy Research Insti-
modes. Hollow cathode lamps (Narva, Berlin, Germany) were used as tute (IPEN, São Paulo, Brazil). Sample masses from 40 to 450 mg were
specific radiation sources. The spectrometer was equipped with a solid irradiated using a neutron flow of 3 × 1012 N cm2 s−1. Gamma activity
sample introduction device for the insertion of the graphite platform
into the graphite tube (manual solid sampling system, model SSA-5, Table 2
Analytik Jena). The instrumental parameters used for determinations Heating program for the determination of Cr, Cu, Mn, Na and Ni in polyimide by SS-GFAAS.
by SS-GFAAS, ICP-OES and ICP-MS are described in Table 1. The mini-
Step Cr Cu Mn Na Ni
mum and maximum (min/max) magnetic field strength corresponding
Pyrolysis
to high sensitivity mode were 0 and 0.8 T, while to low sensitivity mode
Temperature (°C) 1400 900 1300 1000 1100
they were 0.6 and 0.8 T, respectively. The background correction system
employed in this work provides a special dynamic mode, allowing the Atomization
Temperature (°C) 2500 2200 2400 1900 2500
extension of linear calibration range [35–38]. The use of dynamic
Heating rate (°C s−1) 3000 3000 2000 2000 3000
mode presents some advantages due to the possibility of simultaneous Hold time (s) 10 10 8 10 8
analyte determination at low and high concentration, without using
Drying: 110 °C, ramp of 5 °C s−1, hold time 30 s, Ar flow-rate 2.0 L min−1.
any alternative analytical line or applying furnace gas flow-rates [35– Pyrolysis: hold time 40 s, Ar flow-rate 2.0 L min−1.
38]. For low analytes concentration, the high sensitivity mode was Atomization: Argon flow-rate 0 L min−1.
used and the absorbance was obtained by the difference on absorbance Clean out: 2500 °C, ramp of 3000 °C s−1, hold time 5 s, Ar flow-rate 2.0 L min−1.
44 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48
measurements were performed using a Canberra hyperpure Ge detec-
tor (GX 2020) coupled Integrated Signal Processor (Model 1510) and
MCA System 100, both from Canberra. The radioisotopes monitored
and the energy applied were: 51Cr (320.1 keV), 56Mn (846.6 keV) and
24
Na (1368.5 keV). Cu and Ni were not determined by NAA. Comparison
between the analytes concentration obtained by all methods was per-
formed by analysis of variance (ANOVA), with a confidence level of
95%, using the software GraphPad InStat Software Inc. (Version 3.00,
1997).
2.2. Samples and reagents
Water was distilled, deionized and further purified in a Milli-Q sys-
tem (Millipore Corp., Bedford, USA). Analytical grade nitric acid (65%,
Merck, Darmstadt, Germany) was distilled in a sub-boiling system
(model duoPUR 2.01E, Milestone, Bergamo, Italy). Calibration with
standard solutions for SS-GFAAS was prepared by dilution of single-ele-
ment stock solutions (1000 mg L−1 of Cr, Cu, Mn, Na and Ni, Merck).
Standard solutions for the determination of analytes by ICP-MS were
prepared by dilution of a 10 mg L− 1 multi-element stock solution
(SCP33MS, SCP Science, Canada). All reagents used were of analytical
grade or better. Argon 99.996% (White Martins-Praxair, São Paulo, Bra-
zil) was used for GFAAS and for plasma-based instruments, as well as for
pressurization of SRC system.
Two PI samples were used in this work. A high purity polyimide
powder (PI powder, Evonik, Austria) was analyzed for initial experi-
ments and for optimization of the proposed method. This sample has
a particle size up to 50.3 μm, according to the manufacturer. The SS-
GFAAS method was also applied to a PI film (DuPont, USA), commonly
used for electronic applications. This film was cut in small pieces using
a scissors (about 10 × 10 mm) and weighed directly into the graphite
platform.
2.3. SS-GFAAS procedure
Before the first daily measurement, the platform and graphite tube
were cleaned by five heating cycles at 2500 °C for 5 s. Then, blank mea-
surements were carried out using the same heating program described
for solid sample analysis. This procedure was repeated until the blanks
became insignificant (integrated absorbance was lower than 0.002 s).
Blanks were evaluated by simulating the same steps used for the sam-
ples (weighing, platform transportation and modifier addition, if appli-
cable). Pyrolysis and atomization curves were established for the
analytes using PI powder, by weighting into the graphite platform, as
well as for standard solution by pipetting mono-elemental standard.
As satisfactory results were obtained for Cr, Cu, Na and Ni, the use of
Fig. 1. Effect of pyrolysis temperature on residual sample mass (%). Error bars represent the standard deviation (n = 3).
Pd as chemical modifier was evaluated only for Mn determination. In
this sense, Pd was added directly to the sample after the weighing
step, by the addition of a Pd(NO3)2 solution (5 μg). Calibration curves
were made by pipetting the respective mono-elemental standard solu-
tion directly on the platform.
For accuracy evaluation, about 300 mg PI powder was digested with
6 mL of concentrated HNO3 (Merck) in the ultra-high pressure SRC sys-
tem. According to the manufacturer recommendation, it was necessary
to add inside the reaction chamber, a mixture containing 120 mL of
water and 5 mL H2O2 (30%, Merck), before microwave irradiation. The
quartz vessels containing the sample and HNO3 were placed in a rack
that was inserted into the chamber of the microwave oven. Before
heating, the chamber was sealed and pressurized with 40 bar of argon.
The heating program was applied according to the manufacturer rec-
ommendation: i) 10 min of ramp up to 150 °C, ii) hold at 150 °C for
10 min, iii) 10 min of ramp up to 250 °C, iv) hold at 250 °C for 15 min.
After cooling (30 min), final digests were diluted up to 25 mL for further
determination of analytes by ICP-OES and ICP-MS.
3. Results and discussion
3.1. Evaluation of matrix changes during heating program
The thermal behavior of PI was initially evaluated by simulation of
the pyrolysis step [27,32]. The aim of this evaluation was to obtain use-
ful information about the matrix changes after the pyrolysis as well as to
evaluate the mass loss during this step. Therefore, about 5 mg sample
was weighted in the platform and heated at temperatures from 200 to
2000 °C during 40 s. The mass loss behavior is shown in Fig. 1.
As can be seen in Fig. 1, no significant loss of sample mass as well as
no visual change was observed up to 700 °C. From the infrared (IR)
spectrum of original PI (Fig. 2A), it was possible to identify characteristic
absorption bands of succinimide and carbonyl at 1777 and 1714 cm−1,
respectively. This IR spectrum is similar to those spectra for the residual
PI matrix after pyrolysis at 700 °C, suggesting that there are no molecu-
lar changes at this temperature. As shown in Fig. 1, PI degradation starts
at 800 °C, where about 30% of sample mass was lost due to the thermal
decomposition, also evidenced visually by gaseous degradation prod-
ucts. At 900 °C, the mass loss was about 40% and the characteristic ab-
sorption bands for succinimide and carbonyl become weaker (Fig. 2C).
For pyrolysis temperatures higher than 1000 °C, the mass loss was not
significantly changed, remained constant (about 50%) and resulted in
black residues. As can be seen in Fig. 2D, the IR spectrum did not show
characteristic absorption bands that could be identified. In this sense,
pyrolysis temperatures lower than 800 °C were considered unsuitable,
because it could affect the atomization step, by causing non-spectral
 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48
100
95
90
85
80
75
Transmittance (%)
70
65
4000 3500 3000 2500 2000 1500 1000 500 4000 3500 3000 2500 2000 1500 1000 500
100
95
90
85
80
75
70
65
4000 3500 3000 2500 2000 1500 1000 500 4000 3500 3000 2500 2000 1500 1000 500
Wavenumber (cm-1)
Fig. 2. Infrared spectra of (A) original sample and (B), (C) and (D) after pyrolysis at 800, 900 and 1000 °C, respectively.
interferences due to the vaporization of gaseous decomposition prod-
ucts from the sample.
3.2. Pyrolysis and atomization curves
A suitable way to optimize the heating program for instrument cal-
ibration and further determination by SS-GFAAS is based on the evalu-
ation of pyrolysis and atomization profiles [39]. Therefore, aqueous
standard solution and solid sample (PI powder) were compared with
each other and used to establish the pyrolysis and atomization temper-
atures for Cr, Cu, Mn, Na and Ni. In order to avoid/reduce the back-
ground absorbance, the initial temperature evaluated for pyrolysis
was 900 °C for all analytes. In this sense, the integrated absorbance for
Cr, Mn, Na and Ni was normalized to 1 mg of sample mass (real masses
ranged from 0.8 to 1.2 mg). Due to the low Cu concentration the sample
mass for this element was normalized to 5 mg (masses ranged from 4.4
to 5.2 mg). The pyrolysis and atomization curves were plotted using
normalized integrated absorbance while for the aqueous standard solu-
tion integrated absorbance was used. The pyrolysis temperatures in
Table 2 are related to the maximum temperatures that could be
achieved for both aqueous standard solution and solid sample without
analyte losses. In Fig. 3, are shown the pyrolysis and atomization curves
for Cr, Cu, Mn, Na and Ni.
The maximum pyrolysis temperature for Cr, Cu, Mn, Na and Ni in
aqueous solution without analyte losses was 1500, 1000, 1400, 1100
and 1300 °C, while for solid sample it was 1500, 1100, 1500, 1300 and
1200 °C, respectively. It is important to mention that Mn losses were ob-
served during pyrolysis (from 900 °C, data not showed), while for the
standard solution no analyte losses were observed at this temperature.
Therefore, the use of Pd chemical modifier was used for Mn in order to
reduce the analyte losses at higher pyrolysis temperatures. As a com-
promise pyrolysis step for both solid samples and standard solutions
were set at 1400, 900, 1300, 1000 and 1100 °C for Cr, Cu, Mn, Na and
Ni, respectively.
Some drawbacks were observed for Cr determination, probably due
to the formation of relative stable carbides (Cr3C2) at temperatures
about 1300 °C [40]. On the other hand, oxides or carbides of Cu, Mn,
Na and Ni are not thermostable, requiring lower temperatures for
45
100
95
90
85
80
75
A 70 B
65
100
95
90
85
80
75
C 70 D
65
decomposition when compared to Cr [41]. The minimum atomization
temperature required to achieve the maximum integrated absorbance
for Cr, Cu, Mn, Na and Ni for aqueous solutions was 2500, 1900, 2000,
1800 and 2100 °C, while for solid sample it was 2400, 2100, 2400,
1900 and 2500 °C, respectively. Therefore, the atomization tempera-
tures for Cr, Cu, Mn, Na and Ni were set at 2500, 2200, 2400, 1900 and
2500 °C. The optimized conditions for the determination of Cr, Cu, Mn,
Na and Ni by SS-GFAAS are shown in Table 2. Using these conditions,
background absorbance were relatively low (peak height was lower
than 0.4) for all analytes and the non-specific absorbance signals were
efficiently corrected by the background correction system (Appendix
A. Supplementary data).
3.3. Calibration and linearity
The feasibility of instrument calibration with aqueous standard solu-
tion for further determination of analytes in solid samples is an impor-
tant advantage of SS-GFAAS [42]. In this sense, aqueous standard
solutions were added (4 to 30 μL), by manually pipetting onto the graph-
ite platform used for solid samples. The linear calibration range, charac-
teristic mass and also LODs (calculated considering the maximum
sample mass that could be used for each analyte) are shown in Table 3.
Calibration was made using aqueous standard solutions, at low and
high sensitivity modes, and the linear correlation coefficient achieved
(R²) for all analytes was better than 0.995. The LODs (3σ, n = 10) for
high sensitivity mode were 7.0, 2.6, 1.7, 17 and 0.12 ng g−1 while for
the low sensitivity mode, the LODs were 14, 11, 25, 33 and 1.2 ng g−1
for Cr, Cu, Mn, Na and Ni, respectively. These low LODs allowed the de-
terminations in PI at very low concentrations and can be considered
suitable for routine quality control for this polymer. As expected, the
characteristic mass obtained for high sensitivity mode were lower
than those obtained using low sensitivity mode. Another important re-
sult that should be highlighted is the extended linearity of the calibra-
tion curves that allows the analyses at low or high concentration.
Using low sensitivity mode, the characteristic masses were 31, 4, 11,
17 and 10 times higher than at high sensitivity mode for Cr, Cu, Mn, Na
and Ni, respectively. For high sensitivity mode, the characteristic masses
were similar to those reported in previous works: 8.1, 7.4, 2.5, 0.58 and
46 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48

2.0
2.0 Table 3
Linear calibration range, characteristic mass and LOD for SS-GFAAS.
1.5
1.5

Parameter High sensitivitya Low sensitivityb

1.0
1.0
Cr
Linear range (pg) 200 to 2000 2000 to 30,000
Integrated absorbance (s) or normalized integrated absorbance (s mg-1)

0.5
0.5 Characteristic mass (pg) 8.1 250
LOD (ng g−1) 7.0 14
Cr
0
0.0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600 Cu
0 1000 1200 1400 1600 1800 2000 2200 2400 2600 Linear range (pg) 120 to 2000 250 to 20,000
0.6
0.6
Characteristic mass (pg) 7.4 31
LOD (ng g−1) 2.6 11

0.4
0.4
Mn
Linear range (pg) 50 to 1000 100 to 3000
Characteristic mass (pg) 2.5 26
0.2
0.2
LOD (ng g−1) 1.7 25

Cu Na
0
0.0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
Linear range (pg) 20 to 800 50 to 3400
0 1000 1200 1400 1600 1800 2000 2200 2400 2600 Characteristic mass (pg) 0.58 9.8
0.6
0.6 LOD (ng g−1) 17 33

Ni
0.4
0.4 Linear range (pg) 0.50 to 1 3.0 to 38
Characteristic mass (pg) 12 120
LOD (ng g−1) 0.12 1.2
0.2
0.2 a
High sensitivity mode (0.0/0.8 T).
b
Low sensitivity mode (0.6/0.8 T).
Mn
0
0.0
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
0 1000 1200 1400 1600 1800 2000 2200 2400 2600
4.0
4.0

3.4. Influence of sample mass

3.0
3.0

Depending on the sample mass, the analyte concentration can be

2.0
2.0
under or overestimated due to the influence of residual matrix during
the atomization step [43]. In this sense, the sample mass range that
1.0
1.0
could be introduced in the atomizer under the optimized conditions
0
0.0
Na was investigated for each analyte. Therefore, sample masses ranging
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
0 1000 1200 1400 1600 1800 2000 2200 2400 2600 from 0.2 to 10 mg were weighed directly onto the platform, considering
0.8
0.8
analytes concentration within the linear calibration range (Table 3). It is
important to mention that the residual matrix remained on the platform
0.6
0.6
after the cleaning step of heating cycle was removed before the next
measurement using a polypropylene spatula. For the evaluation of sam-
0.4
0.4
ple mass (Fig. 4), sets of measurements for each analyte were divided in
0.2
0.2 mass intervals of 1 mg and their respective mean concentration were
Ni compared by analysis of variance with a confidence level of 95%
0
0.0 (Tukey-Kramer multiple comparison test).
800 1000 1200 1400 1600 1800 2000 2200 2400 2600
0 1000 1200 1400 1600 1800 2000 2200 2400 2600 According to Fig. 4, the sample mass did not have any influence on
Temperature (oC) the determination of Cr, Cu, Mn, Na and Ni by SS-GFAAS. Therefore, up
to 10 mg sample mass could be applied for these determinations, with-
Fig. 3. Pyrolysis and atomization curves (● normalized absorbance for polyimide sample; out under or overestimated results. Sample masses higher than 10 mg
○ integrated absorbance for standard solution). For Cr, Na and Ni the low sensitive mode were not evaluated because this is the maximum loading capacity of
was used, while for Cu and Mn the high sensitivity mode was applied. Palladium (5 μg) the graphite platform. For Cr, Na and Ni evaluation, the spectrometer
was used as chemical modifier for Mn determination. The integrated absorbance for Cr,
was operated at low sensitivity mode, while for Cu and Mn, the high
Mn, Na and Ni was normalized to 1 mg sample while for Cu it was normalized to 5 mg.
Error bars represent the standard deviation (n = 3). sensitivity mode was chosen as showed in Fig. 4.
In general, the RSDs were below 2.9, 8.5, 3.7, 3.9 and 8.0% for Cr, Cu,
Mn, Na and Ni, respectively. These values can be considered as suitable
12 pg for Cr, Cu, Mn Na and Ni, respectively [26,28]. It is important to for solid PI analyses at very low concentration. It is important to men-
note that low sensitivity mode was used for Cr, Na and Ni, while Cu tion that despite the use of the low sensitivity mode for the determina-
and Mn were determined in high sensitivity mode. In addition, high tion of Cr, Na and Ni, the high sensitivity mode can also be used for the
blank values were obtained for Na (0.056 ± 0.061 s, n = 10) when determination of these analytes at lower concentration in PI samples. If
high sensitivity mode was applied. For the low sensitivity mode, blank necessary, for analytes at high concentration, the use of a low sensitivity
values for Na were negligible (0.0067 ± 0.0074 s, n = 10). condition is an important advantage of the proposed method, allowing
Another important advantage of the proposed method was the possi- the determination in a wide range of concentration.
bility of extending the linear calibration range. The use of dynamic mode
allows the analyses in a relatively high range of analytes concentration
without changing the wavelength or by applying furnace gas flow during 3.5. Figures of merit and analysis of samples
analyte atomization. The low LODs allowed the determination, especially
for Cu and Na that were at very low concentration and were not deter- The accuracy of SS-GFAAS was evaluated by results comparison with
mined in digested samples by ICP-OES and ICP-MS. those obtained by ICP-OES, ICP-MS (after digestion of 300 mg sample)
 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48 47

5
5,0
4. Conclusions

4
4,0
A solid sampling approach for the determination of Cr, Cu, Mn, Na
and Ni in high purity polyimide was developed. Although the proposed
3,0
3 method is unable to perform simultaneous multi-element analysis, it
could be considered as a good alternative when compared to conven-
2,0
2 tional methods requiring previous digestion for liquid introduction.
Concentration (µg g-1)

Moreover, it practically avoids the use of high amount of toxic and cor-
1,0
1 rosive reagents. The sample mass did not influence the analyses and up
to 10 mg sample mass could be analyzed without over or
0,000 underestimated results, being an important advantage, especially for
0 22 44 66 88 10
10
trace analysis. In addition, SS-GFAAS is not time consuming and the
0,6
0.6 risk of contamination is reduced. Calibrations curves were made with
aqueous standard solution and the use of Pd (5 μg) as chemical modifier
was required only for Mn. Another advantage of this method is the very
0,4 low LODs achieved (lower than 12 ng g−1 for all analytes), allowing the
0.4
determination at trace levels, combined with the extension of calibra-
tion range (dynamic mode), allowing the determination at low and
high concentration.
0,2
0.2

Acknowledgements
0,0
0 0 2 4 6 8 10 The authors are grateful to Conselho Nacional de Desenvolvimento
0 2 4 6 8 10 Científico e Tecnológico (CNPq), Coordenação de Aperfeiçoamento de
Sample mass (mg) Pessoal de Nível Superior (CAPES) and Fundação de Amparo à Pesquisa
do Estado do Rio Grande do Sul (FAPERGS) for supporting this study.
Fig. 4. Influence of polyimide mass for the determination of Cr ( ), Cu ( ), Mn ( ), Na ( )
and Ni ( ). Each symbol corresponds to individual measurement. For Cr, Na and Ni the
low sensitive mode was applied, while for Cu and Mn the high sensitivity mode was Appendix A. Supplementary data
used. Palladium (5 μg) was used as chemical modifier for Mn determination
Supplementary data to this article can be found online at http://dx.
doi.org/10.1016/j.sab.2017.01.005.

References

and also by NAA. The results for PI powder and film by SS-GFAAS are
shown in Table 4.
No statistical differences were observed between the results using
SS-GFAAS, ICP-OES and ICP-MS for Cr, Mn and Ni. On the other hand,
this comparison was not possible for Cu and Na, which were lower
than the LODs. Therefore, in order to evaluate the accuracy of the SS-
GFAAS method, results were also compared with those obtained by
NAA. As can be seen in Table 4, no statistical difference was observed
for the results for Cr and Mn. However, the Na concentration deter-
mined by NAA was lower than the LOD (0.4 μg g−1). Taking into account
that all techniques used for results comparison were not feasible for the
determination of Cu and Na in PI samples, the accuracy was evaluated
by recovery tests (n = 15) using SS-GFAAS. Recoveries were 103 ±
15% for Cu and 101 ± 14% for Na.
For PI powder, no significant difference was observed between the
results for the determination of Cr, Mn and Ni by SS-GFAAS, ICP-OES,
ICP-MS and NAA. For the PI film, the results for ICP-OES and ICP-MS
were lower than the LOD, impairing the comparison with results obtain-
ed by SS-GFAAS.
[1] D.J. Liaw, K.L. Wang, C.Y. Huang, K.R. Lee, J.Y. Lai, C.S. Ha, Advanced polyimide mate-
rials: syntheses, physical properties, and applications, Prog. Polym. Sci. 37 (2012)
907–974.
[2] H.B. Xiang, Z. Huang, L.Q. Liu, L. Chen, J. Zhu, Z.M. Hu, J.R. Yu, Structure and proper-
ties of polyimide (BTDA-TDI/MDI co-polyimide) fiber obtained by wet-spinning,
Macromol. Res. 19 (2011) 645–653.
[3] A.B. Franzier, Recent applications of polyimide to micromachining technology, IEEE
Trans. Electr. Insul. 42 (1995) 442–448.
[4] B.J. Kim, E. Meng, Review of polymers MEMS micromachining, J. Micromech.
Microeng. 26 (2016) 1–21.
[5] M.H. Chen, J.H. Yin, R. Jin, L. Yao, B. Sua, Q.Q. Lei, Dielectric and mechanical proper-
ties and thermal stability of polyimide-graphene oxide composite films, Thin Solid
Films 584 (2015) 232–237.
[6] Y. Wang, Q. Gao, Q. You, G. Liao, H. Xia, D. Wang, Porous polyimide framework: a
novel versatile adsorbent for highly efficient removals of azo dye and antibiotic,
React. Funct. Polym. 103 (2016) 9–16.
[7] Y. Liu, T.S. Chung, R. Wang, D.F. Li, M.L. Chng, Chemical cross-linking modification of
polyimide/poly(ether sulfone) dual-layer hollow-fiber membranes or gas separa-
tion, Ind. Eng. Chem. Res. 42 (2003) 1190–1195.
[8] A.I. Kupchinshin, B.G. Taipova, A.A. Kupchishin, B.A. Kozhamkulov, Study on the
physical and mechanical properties of composites based on polyimide and polycar-
bonate, Mech. Compos. Mater. 51 (2015) 115–118.
[9] A. Baba, F. Kaneko, K. Shinbo, K. Kato, S. Kobayashi, Evaluation of liquid crystal mol-
ecules on polyimide LB films using attenuated total reflection measurement, Thin
Solid Films 327 (1998) 353–356.

Table 4
Results for the determination of Cr, Cu, Mn, Na and Ni in polyimide samples by SS-GFAAS, ICP-OES, ICP-MS after microwave digestion and by NAA. Concentration in μg g−1 (mean ± SD).

SS-GFAAS
ICP-OES ICP-MS NAA
Analyte PI film PI powder PI powder PI powder PI powder

Cr 0.105 ± 0.006 3.78 ± 0.11 3.79 ± 0.12 3.83 ± 0.04 3.90 ± 0.20
Cu 0.010 ± 0.001 0.082 ± 0.007 b1.3 b1.0 ND
Mn 0.034 ± 0.004 0.328 ± 0.012 0.335 ± 0.010 0.354 ± 0.025 0.42 ± 0.05
Na 7.34 ± 0.51 0.335 ± 0.013 b0.59 ND b0.4
Ni 0.554 ± 0.036 1.76 ± 0.14 1.75 ± 0.15 1.87 ± 0.08 ND

ND = Not determined.
48 R.F. Santos et al. / Spectrochimica Acta Part B 129 (2017) 42–48
[10] E.I. Maltsev, M.A. Brusentseva, V.A. Kolesnikov, V.I. Berendyaev, B.V. Kotov, A.V.
Vannikov, Bright blue-green electroluminescence from aromatic polyimides, Appl.
Phys. Lett. 71 (1997) 3480–3482.
[11] A.C. Patil, N.V. Thakor, Implantable neurotechnologies: a review of micro and
nanoelectrodes for neural recording, Med. Biol. Eng. Comput. 54 (2016) 23–44.
[12] T.J. Kem, R.A. McIntyre, Influence of transition metal-doped titanium (IV) dioxide on
the photo degradation of polystyrene, Polym. Degrad. Stab. 91 (2006) 3010–3019.
[13] W. Schnabel, Polymer Degradation - Principles and Practical Applications, Macmil-
lan, New York, 1981.
[14] F. Bellucci, L. Nicodemo, T. Monetta, M.J. Kloppers, Latanision, A study of corrosion
initiation on polyimide coatings, Corros. Sci. 33 (1992) 1203–1226.
[15] E. Sacher, Dieletric properties of polyimide film. ll, IEEE Trans. Electr. Insul. 14
(1979) 85–92.
[16] E.M.M. Flores, E.I. Muller, F.A. Duarte, P. Grinberg, R.E. Sturgeon, Determination of
trace elements in fluoropolymers after microwave-induced combustion, Anal.
Chem. 85 (2013) 374–380.
[17] Semiconductor equipment and materials international – SEMI F48-0600, test meth-
od for determining trace metals in polymer materials, http://www.semi.org 2009
(accessed 01.07.2016).
[18] E.M.M. Flores, Microwave-Assisted Sample Preparation for Trace Element Determi-
nation, Elsevier, Amsterdam, 2014.
[19] P.A. Mello, C.K. Giesbrecht, M.S. Alencar, Determination of sulfur in petroleum coke
combining closed vessel microwave-induced combustion and inductively coupled
plasma-optical emission spectrometry, Anal. Lett. 9 (2008) 1623–1632.
[20] J.V. Maciel, C.L. Knorr, E.M.M. Flores, Feasibility of microwave-induced combustion
for trace element determination in Engraulis anchoita by ICP-MS, Food Chem. 145
(2014) 927–931.
[21] P. Barth, J. Hassler, I. Kudrik, V. Krivan, Determination of trace impurities in boron
nitride by graphite furnace atomic absorption spectrometry and electrothermal va-
porization inductively coupled plasma optical emission spectrometry using solid
sampling, Spectrochim. Acta B 62 (2007) 924–932.
[22] L.F. Rodrigues, J.C.P. Mattos, V.L. Dressler, D.P. Pozebon, E.M.M. Flores, Determination
of cadmium, cooper and lead in alumina based catalysts by direct solid sampling
graphite furnace atomic absorption spectrometry, Spectrochim. Acta B 62 (2007)
933–938.
[23] M. Resano, M. Aramendiab, M. Belarra, A. High-resolution continuum source graph-
ite furnace atomic absorption spectrometry for direct analysis of solid samples and
complex materials: a tutorial review, J. Anal. Atom. Spectrom. 29 (2014) 2229–2250.
[24] E. Sacher, The determination of trace inorganic contaminants in polyimides, IEEE
Trans. Electr. Insul. 18 (1983) 369–373.
[25] V. Krivan, B. Koch, Determination of Ca, Cu, Fe, K, Na and Si in polyimides for micro-
electronics by electrothermal atomic absorption spectrometry involving samples
dissolution in organic solvents, Anal. Chem. 67 (1995) 3148–3153.
[26] M. Jaskola, M. Korman, A. Stolarz, Estimation of the impurity levels in polyimide foils
and the life-time of the foils irradiated by charged projectiles, Nucl. Instrum.
Methods A 590 (2008) 176–180.
[27] J.C.P. Mattos, E.M.M. Flores, V. Krivan, Trace element analysis of silicon nitride pow-
ders by direct solid sampling graphite furnace atomic absorption spectrometry, J.
Anal. Atom. Spectrom. 23 (2008) 931–937.
[28] R.C. Bolzan, D.P. Moraes, J.C.P. Mattos, V.L. Dressler, E.M.M. Flores, Arsenic determi-
nation in pharmaceutical grade barium sulfate using direct sampling electrothermal
atomic absorption spectrometry, J. Braz. Chem. Soc. 21 (2010) 686–693.
[29] P.A. Mello, L.F. Rodrigues, M.A.G. Nunes, J.C.P. Mattos, E.I. Muller, V.L. Dressler,
E.M.M. Flores, Determination of metal impurities in carbon nanotubes by direct
solid sampling electrothermal atomic absorption spectrometry, J. Braz. Chem. Soc.
22 (2011) 1040–1049.
[30] L.F. Rodrigues, J.C.P. Mattos, R.C. Bolzan, E.M.M. Flores, F.A. Duarte, Determination of
trace elements in raw material for polyurethane production using direct sampling
graphite furnace atomic absorption spectrometry, J. Anal. Atom. Spectrom. 29
(2014) 324–331.
[31] B.M. Soares, R.F. Santos, R.C. Bolzan, E.I. Muller, E.G. Primel, F.A. Duarte, Simulta-
neous determination of iron and nickel in fluoropolymers by solid sampling high-
resolution continuum source graphite furnace atomic absorption spectrometry,
Talanta 160 (2016) 454–460.
[32] J.C.P. Mattos, L.F. Rodrigues, E.M.M. Flores, V. Krivan, Determination of trace impuri-
ties in aluminum nitride by direct solid sampling graphite furnace atomic absorp-
tion spectrometry, Spectrochim. Acta B 66 (2011) 637–643.
[33] R.C. Bolzan, L.F. Rodrigues, J.C.P. Mattos, V.L. Dressler, E.M.M. Flores, Chromium de-
termination in pharmaceutical grade barium sulfate by solid sampling electrother-
mal atomic absorption spectrometry with Zeeman effect background correction,
Talanta 74 (2007) 119–124.
[34] N. Laczai, L. Kovács, Á. Péter, L. Bencs, Solid sampling determination of lithium and
sodium additives in microsamples of yttrium oxyorthosilicate by high-resolution
continuum source graphite furnace atomic absorption spectrometry, Spectrochim.
Acta Part B 117 (2016) 8–15.
[35] M.T.C. de Loos-Vollebregt, L. Galan, Extended range Zeeman atomic absorption spec-
trometry based on a 3-field a.c. magnet, Spectrochim. Acta B 41 (1986) 825–835.
[36] M.T.C. de Loos-Vollebregt, Extension of the dynamic range in a.c. Zeeman electro-
thermal atomic absorption spectrometry, Spectrochim. Acta B 48 (1993)
1505–1515.
[37] B.V. L'vov, L.K. Polzik, P.N. Fedorov, Extension of the dynamic range in Zeeman graphite
furnace atomic absorption spectrometry, Spectrochim. Acta B 47 (1992) 1411–1420.
[38] A.A.S. Operating Manual, ZEEnit 60/65 – Atomic Absorption Spectrometer, Analytik
Jena, 2001.
[39] B. Welz, Atomic absorption spectrometry, VCH, second ed., 1985.
[40] N.S. Thomaidis, E.A. Piperaki, Effect of chemical modifiers on the kinetic parameters
characterizing the electrothermal atomization of chromium, Spectrochim. Acta B 55
(2000) 611–627.
[41] A.B. Volynsky, Graphite atomizers modified with high-melting carbides for electro-
thermal atomic absorption spectrometry. II. Practical aspects, Spectrochim. Acta B
53 (1998) 1607–1644.
[42] M.A. Belarra, M. Resano, F. Vanhaecke, L. Moens, Direct solid sampling with electro-
thermal vaporization/atomization: what for and how? TrAC Trends Anal. Chem. 21
(2002) 828–839.
[43] M.A. Belarra, C. Crespo, M.P. Martinez-Garbayo, J.R. Castillo, Direct determination of
metals in solid samples by graphite-furnace atomic absorption spectrometry: does
sample mass influence the analytical results? Spectrochim. Acta B 52 (1997)
1855–1880.