Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

Contents lists available at ScienceDirect

Spectrochimica Acta Part B: Atomic Spectroscopy

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

Determination of metal impurity elements in lithium hexafluorophosphate

using inductively coupled plasma tandem mass spectrometry based on
reaction gas mixtures
Liang Fu a, *, Hualin Xie b, Jianhua Huang c, Xianhua Chen a, *, Lin Chen c
a
College of Materials Science and Engineering, Chongqing University, Chongqing 400045, PR China
b
College of Materials Science and Engineering, Yangtze Normal University, Fuling 408100, PR China
c
Hunan Academy of Chinese Medicine, Hunan University of Chinese Medicine, Changsha 410013, PR China

A R T I C L E I N F O A B S T R A C T

Keywords: An analytical method was developed for inductively coupled plasma tandem mass spectrometry (ICP-MS/MS)
Lithium hexafluorophosphate based on reaction gas mixtures to eliminate the spectral interferences. Lithium hexafluorophosphate (LiPF6) was
Inductively coupled plasma tandem mass dissolved in ultrapure water, and the presence of 18 metal impurity elements were determined directly by ICP-
spectrometry
MS/MS. In the MS/MS mode, the interference of argide ions could be quickly eliminated by adding H2 to the
Metal impurity elements
Reaction gas mixtures
reaction gas O2, while the addition of H2 to the reaction gas NH3/He promoted the formation of -NH3 from the
Spectral interference adducts -NH and -NH2. The reaction gas mixtures O2/H2 and NH3/He/H2 were used to eliminate the in­
terferences when using the mass shift method and on-mass method. The internal standard ions with similar mass
numbers and similar mass spectrometry behaviors were selected to correct for instrument instability, signal drift,
and matrix effects. The accuracy of the method was evaluated by comparative analysis using sector field
inductively coupled plasma mass spectrometry (SF-ICP-MS). The results show that the interference can be
completely eliminated by using mixtures of reaction gases mixtures, while achieving high sensitivities for the
analytes. The limits of detection (LODs) ranged from 0.30 to 63.8 ng L− 1. At 95% confidence level, no significant
difference was observed between the results obtained from ICP-MS/MS and SF-ICP-MS analyses, and the relative
standard deviation (RSD) was ≤5.67%. The developed method provides simple sample processing, good accu­
racy, and high precision, and it can be used for the interference-free determination of metal impurity elements in
LiPF6.

1. Introduction the surface of anode, increases the irreversible capacity of battery and
causes a worse cycling performance, especially the high-temperature
Energy crisis and environmental pollution continue to promote the cycling and high-temperature storage performances. H2O can react
development of a new energy industry with a huge demand for lithium- with LiPF6 to form acidic impurities, further inducing the decomposition
ion batteries. The electrolyte in lithium-ion batteries is mainly composed of LiPF6 to form HF, PF5, and fluorine containing phosphate impurities.
of lithium electrolytes, mixed organic solvents, and additives. Due to the They will cause the dissolution of cathode-active substances and exfo­
high voltage of lithium-ion battery and the active material of its negative liation of cathode material. Therefore, HF and H2O are generally
electrode, a nonaqueous electrolyte is required. The lithium hexa­ important indices of evaluating LiPF6 performance [4–6]. However, the
fluorophosphate (LiPF6), a nonaqueous electrolyte salt with the advan­ migration of metal impurity elements in LiPF6 and the corresponding
tages of greater solubility in organic carbonate solvents, high metal deposition have harmful degradation effects. Especially, the high
conductivity, and good electrochemical stability, is an important concentration of metal impurity elements not only decreases the battery
component of lithium battery electrolytes [1–3]. During the first charge/ reversible capacity, but also the metal impurity ions may also deterio­
discharge cycle, the formation of Li2O, LiF, and LiOH by H2O and HF on rate the solid electrolyte interphase (SEI), thus affecting the safety

* Corresponding author at: College of Materials Science and Engineering, Chongqing University, No.83, Shabei street, Shapingba District, Chongqing 400045,
China.
E-mail addresses: [email protected] (L. Fu), [email protected], [email protected] (X. Chen).

https://doi.org/10.1016/j.sab.2021.106217
Received 1 February 2021; Received in revised form 13 April 2021; Accepted 10 May 2021
Available online 12 May 2021
0584-8547/© 2021 Elsevier B.V. All rights reserved.
L. Fu et al. Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

performance of batteries [7–10]. The Chemical Industry Standards of the Table 1

People’s Republic of China (HG/T 4067–2015) imposes strict limits on Working parameters of ICP-MS/MS.
the content of metal impurity elements in LiPF6 electrolyte. Therefore, a Parameter Operating condition
theoretical basis of quality rating evaluation of LiPF6 can be obtained by
Preset plasma Low matrix
establishing a method for the fast determination of metal impurity ele­ RF power (W) 1550
ments in LiPF6. Plasma gas flow rate (L min− 1) 15.0
At present, inductively coupled plasma optical emission spectros­ Carrier gas flow rate (L min− 1) 0.85
copy (ICP-OES) and inductively coupled plasma mass spectrometry Auxiliary gas flow rate (L min− 1) 1.8
Sampling depth (mm) 8
(ICP-MS) are widely used for the rapid determination of impurity ele­ Makeup gas flow rate 0.15
ments in LiPF6 [5,8,11–13]. Dimethyl carbonate (DMC) was used to Spray chamber temperature (◦ C) 2
dissolve LiPF6 and the content of impurity elements was determined by Wait time offset (ms) 10
ICP-OES after wet digestion treatment. Because of the limits of detection Dwell times (ms) 100
Integration time (s) 1
(LODs), it could not satisfy the determination requirement for trace or
Sweeps/replicate 100
ultratrace elements [11]. The content of impurity elements in LiPF6 was Replicates 10
also directly measured by ICP-MS after dissolving in absolute ethanol. Selected mass at Q1 23, 24, 27, 39, 48, 51, 52, 55, 56, 59, 60, 63, 66,
The He collision mode and H2 reaction mode of the collision reaction cell 107, 111, 121, 137, 208
(CRC) were used to eliminate the spectral interference, and the LODs of Selected mass at Q2 23, 24, 27, 39, 150, 51, 68, 71, 72, 75, 111, 97,
83,107, 111, 121, 137, 208
analytes were significantly improved [12]. Although CRC technology Operating mode MS/MS
has become a general technique for eliminating spectral interference, Cell gas O2/H2 NH3/He/H2
the He collision mode has an elimination effect only for the interference Cell gas flow rate (mL min− 1) O2 0.30, H2 2.0 NH3/He 3.0, H2 2.0
of polyatomic ions, while it is not suitable for isobaric interferences or Octopole bias (V) -18 − 20
Energy discrimination (V) − 10 − 10
double charge interferences. In the He collision mode, all ions from the
plasma undergo collision resulting in an energy damage, which reduces
the sensitivity of the analytes and severely affects the determination of distillation.
trace elements [14,15]. The H2 reaction mode is suitable for the analysis A series of 0.0, 0.2, 0.5, 2.0, and 10.0 μg L− 1 calibration standard
of trace elements. However, in ICP-MS analysis, because all ions from solutions were prepared by adequate dilution of 1000 mg L− 1 single
plasmas enter CRC, new reaction product ion can be easily formed by the element standard solutions of Na, Mg, Al, K, Ti, V, Cr, Mn, Fe, Co, Ni, Cu,
reaction with H2 and new interferences are formed. Therefore, Zn, Ag, Cd, Sb, Ba, and Pb. An internal standard solution was prepared
improving the selectivity of the reaction in the CRC is an effective by adequate dilution of 10 mg L− 1 Li, Sc, Ge, Y, In, Tb, and Bi internal
manner to utilize the potential of reaction mode [16]. standard stock solutions (p/n 5183–4681), were purchased from Agi­
Inductively coupled plasma tandem mass spectrometry (ICP-MS/MS) lent, USA. A standard solution was prepared by mixing 50 mg L− 1 of Li,
is equipped with two tandem quadrupole mass analyzers (Q1, Q2) to 200 mg L− 1 of P, and 750 mg L− 1 of F in ultrapure water to simulate
eliminate the interferences. Only those ions with a specified mass-to- LiPF6 sample matrix solution. All solutions were prepared in 1% (v/v)
charge ratio (m/z) are allowed to enter the CRC, and other m/z ions nitric acid.
are excluded by Q1, significantly improving the selectivity of the re­
actions in CRC. Q2 selects only the target m/z ions from the CRC outlet
2.3. Sample pretreatment
and blocks other m/z ions from entering the detector. The controllability
of the reaction process and the consistency of reaction pathways provide
The sealed LiPF6 sample was opened in the glove box, and 0.5 g of
universality for the determination of trace and ultratrace elements in
LiPF6 was accurately weighed and put into a 50 mL high-density poly­
complex matrix samples [17–23]. In this study, a new strategy is pro­
ethylene (HDPE) plastic cup. Ultrapure water was added into the HDPE
posed to eliminate spectral interference using reaction gas mixtures O2/
cup to dissolve LiPF6, and then the mixture was transferred to a 100 mL
H2 and NH3/He/H2, providing a new method for the interference-free
HDPE volumetric flask till a constant volume was reached.
determination of metal impurity elements in LiPF6 by ICP-MS/MS.

2. Experimental 2.4. Experimental methods

2.1. Instrumentation and operating conditions Calibration standard solutions were determined by ICP-MS/MS.
MassHunter software was used to establish a calibration curve, and
Agilent 8800 ICP-MS/MS instrument (Agilent, USA) was configured the sample solution and blank solution were estimated simultaneously.
with an HF-resistant inert sample introduction kit from poly­ The 1 mg L− 1 internal standard was added to the sample solution using
fluoroalkoxy (PFA) to withstand the corrosion of HF acid and high an online addition method and a T-type kit to mix it with the sample
concentration of Li matrix produced by the reaction of LiPF6 with water. solution. The dilution factor when the internal standard was added
In addition, two quadrupole mass filters with a CRC located in between online was usually about 1/20; therefore, the actual concentration of the
were used. Table 1 shows the ICP-MS/MS operating conditions. Milli-Q internal standard solution entering the nebulizer was about 50 μg L− 1. In
deionizer (Millipore, USA) was used to obtain ultrapure water. To the MS/MS mode, the reaction gas mixture of O2/H2 was used to elim­
evaluate the accuracy of the developed method, an Element XR SF-ICP- inate the spectral interference for the determinations of Ti, Cr, Mn, Fe,
MS spectrometer (Thermo Fisher Scientific, USA) was used. The oper­ and Co. Besides, the reaction gas mixture of NH3/He/H2 was used to
ating conditions of SF-ICP-MS are shown in Table S1 (Appendix A). eliminate the spectral interference for the determination of Na, Mg, Al,
K, V, Ni, Cu, and Zn. The principle of spectral interference elimination by
2.2. Samples, standards, and reagents reaction gas mixtures is shown in Fig. 1.

LiPF6 samples were purchased from different sources in China 3. Results and discussion
(Chongqing, Shanghai, Shenzhen, and Nanjing). Ultrapure Milli-Q water
(resistivity higher than 18.2 MΩ cm, Millipore, USA) and 65% (w/w) 3.1. Mass spectrometry mode
suprapur nitric acid (Merck, Germany) was used throughout the ex­
periments. Nitric acid was additionally purified by sub-boiling LiPF6 matrix constituents, with Ar gas and H2O, would produce a

2
L. Fu et al. Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

Q1 I1 + CRC Q2 I2 +
I1 +: Interferences 1 O2 /H2 TiO+, CrO+,
I2 +: Interferences 2 MnO+, FeO+,
M+: Ti, Cr, Mn, Fe, Co M++O2 MO+
CoO+

I1 +: Interferences 1 NH3 /He/H2 Ni(NH3 ) 3 +,

I2 +: Interferences 2 M++NH3 M(NH3 )n + Fe(NH3 ) 2 +,
M+: Ni, Cu, Zn n=1, 2, 3 Zn(NH3 ) +

I 1 +: Interferences 1 NH3 /He/H2

I 2 +: Interferences 2 Na+, Mg+,
M+: Na, Mg, Al, K, V M++NH 3 M+ Al+, K+, V+

Fig. 1. Principle of eliminating spectral interference with reaction gas mixtures in the MS/MS mode.

36
large amount of interfering ions after ionization by plasma, affecting the Ar15NH+ [32], thereby improving the selectivity of the O2 reaction,
accuracy in the determination of analytes. Especially, the spectral which was conducive to the reaction between 52Cr+ and O2 and the
interference of (m/z) ≤ 80 amu had a more significant influence on the elimination of the interference. In the MS/MS mode, 1 μg L− 1 Cr stan­
determination of 23Na, 24Mg, 27Al, 39K, 48Ti, 51V, 52Cr, 55Mn, 56Fe, 59Co, dard solution was determined in the simulated sample solution by using
60
Ni, 63Cu, and 66Zn. The interference is caused by the middle mass O2 and O2/H2 as the reaction gases to study the mass spectrometry
isotopes of 107Ag, 111Cd, 121Sb, and 137Ba, and the heavy mass isotope of behavior of 52Cr+ with different reaction gases. Table 2 shows that
208
Pb mainly originated from the polyatomic ions formed by impurity although the increase in the yield of 52Cr+ was not obvious after the
elements in LiPF6. Because the contents of impurity elements in LiPF6 addition of H2, it decreased the BEC of 52Cr+ and increased the sensi­
were very low and the interference so formed was negligible, in this tivity of 52Cr+. The main interferences of 48Ti+ were 48Ca+, 32S16O+, and
31 17 +
experiment, Ag, Cd, Sb, Ba, and Pb were measured in the single quad­ P O . The signal intensity (48Ti16O+) obtained from the determina­
rupole (SQ, Q1 as a wideband mass filter) no-gas mode. Na, Mg, Al, K, Ti, tion of 48Ti+ by O2 mass shift method was over 10 times higher than that
V, Cr, Mn, Fe, Co, Ni, Cu, and Zn were measured in the MS/MS mode (Q1 by NH3/He mass shift method (48Ti(14NH3)+ 6 ) [18]. However,
48
Ca+,
32 16 + 31 17 + 48 16 + 32 16 +
as a unit (1 amu window) mass filter and Q2 also set to the single mass of S O , and P O can also react with O2 to form Ca O , S O2 ,
the target ion or reaction product ion) with CRC reaction mode in place and 31P17O16O+, respectively, which interfered with the determination
to eliminate the spectral interference. of 48Ti16O+. By adding a small amount of H2 in the reaction gas O2, the
O2 and NH3/He (v/v = 1/9, He was a buffer gas) were the most interfering ion 48Ca16O+, 32S16O+ 2 , and
31 17 16 +
P O O will further react
versatile and effective reaction gases for interference elimination and with H2 to form 48Ca16OH+, 32S16O2H+, and 31P17O16OH+, respectively,
widely used in ICP-MS/MS analysis [24–27], while H2 was widely used thus eliminating the interferences of 48Ti16O+. In this case, the contri­
as a reaction gas in the traditional analysis by ICP-MS with CRC [28–30]. bution of S and Ca to the interference of Ti is negligible owing to the low
As H2 was mainly used for the on-mass method to eliminate spectral concentrations of these elements. In contrast, P severely interferes with
interference, a small amount of H2 was added into the O2 and NH3/He the determination of Ti. In the MS/MS mode, 1 μg L− 1 Ti standard so­
systems to form reaction gas mixtures O2/H2 and NH3/He/H2 in these lution was determined in the simulated sample solution by using O2 and
experiments. The analysis mode was selected by comparing the ion yield O2/H2 as the reaction gases. Table 2 shows that the yield of
48 +
(the percentage of monitoring ion in all reaction product ions), sensi­ Ti decreased after the addition of H2, indicating the effective elimi­
tivity, and background equivalent concentration (BEC = ρsIb/(Is -Ib); ρs is nation of interference of 31P17O16O+ to 48Ti16O+. The BEC of 48Ti+
the concentration of analyte in the simulated sample solution, and Ib is decreased, and the sensitivity significantly increased. While 55Mn+,
56 +
the signal intensity of analyte in the blank solution; Is is the signal in­ Fe , and 59Co+ were all interfered by Ar ion, similar to 52Cr+, the
tensity of analyte in the simulated sample) of the monitoring ion for the application of reaction gas mixtures O2/H2 resulted in a lower BEC and
isotopes before and after adding H2. higher sensitivity compared to the single reaction gas O2. In this
The interference for 52Cr+ mainly came from 40Ar12C+, 36Ar16O+, experiment, Ti, Cr, Mn, Fe, and Co were determined in the O2/H2 re­
38 14 +
Ar N , and 36Ar15NH+ species formed by the argide. In the O2 re­ action mode to eliminate the interference.
action mode, the mass pair was set to Q1: 52 → Q2: 68, and the inter­ In the NH3/He reaction mode, the 23Na+, 24Mg+, 27Al+, and 39K+
ference was eliminated by estimating 52Cr16O+ [31]. In the O2/H2 ions hardly react with NH3 and the interferences were usually elimi­
reaction mode, the added H2 could quickly react with the species nated using the on-mass method [23]. In the MS/MS mode, the reaction
generated from argide ions, like 40Ar12C+, 36Ar16O+, 38Ar14N+, and between 51V+ and O2 was found to be an exothermic reaction,

Table 2
Ion yields, sensitivities, and BECs of analytes using O2 and O2/H2 as reaction gases in the MS/MS mode.
Isotope O2 O2/H2

Monitoring ion Yield Sensitivity BEC Monitoring ion Yield Sensitivity BEC
(%) (cps/(L μg− 1)) (ng L− 1) (%) (cps/(L μg− 1)) (ng L− 1)
48
Ti (1 μg L− 1) 48
Ti16O+ 70.2 134,000 7.93 48
Ti16O+ 58.5 90,100 3.08
52
Cr (1 μg L− 1) 52
Cr16O+ 17.6 179,000 31.3 52
Cr16O+ 19.5 233,000 19.2
55
Mn (1 μg L− 1) 55
Mn16O+ 16.3 259,000 12.9 55
Mn16O+ 15.6 312,000 8.34
56
Fe (1 μg L− 1) 56 16 +
Fe O 23.2 212,000 98.6 56 16 +
Fe O 25.0 349,000 71.3
59
Co (1 μg L− 1) 59
Co16O+ 24.5 263,000 3.15 59
Co16O+ 23.8 298,000 2.07

3
L. Fu et al. Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

spontaneously generating a large amount of 51V16O+ with a yield of 3.3. Optimization of reaction gas flow rate
63.3%, and it can be determined using O2 mass shift method [33]. In the
NH3/He reaction mode, the on-mass method can be used to determine 1 μg L− 1 of analyte standard solution was added to simulated sample
51 +
V by using 51V+ as the monitoring ion, where the yield of 51V+ solution for determination by ICP-MS/MS. In the MS/MS mode, the
(55.6%, Table 3) is slightly lower than that of 51V16O+. The determi­ interference elimination effects were investigated using different reac­
nation of trace amounts of V could be achieved by using either O2 or tion gas rates, and the results are shown in Fig. 2.
NH3/He as the reaction gas. Since the BEC of the NH3/He on-mass Fig. 2(a) shows the effects of interfering elimination via different H2
method was much lower than that of the O2 mass-shift method, the flow rates for the determination of Ti, Cr, Mn, Fe, and Co by H2 on-mass
NH3/He reaction gas was preferred for the determination of V. Because method and a fixed O2 flow rate (0.30 mL min− 1). As the H2 flow rate
the ions of 60Ni+, 63Cu+, and 66Zn+ could undergo a mass shift in NH3/ increased, the BECs of all elements decreased, indicating the decay of
He reaction mode to generate a large number of cluster ions, 60Ni interference. The BECs of the five elements reached the lowest value
(14NH3)+ 3,
63
Cu(14NH3)+ 2,
66
Zn(14NH3)+ were selected as monitoring with a H2 flow rate of 1.8–2.1 mL min− 1. When the cell gas flow rate was
ions [34]. In the on-mass method to eliminate interference, addition of a too high, the signal intensity of the analytes was severely damaged,
small amount of H2 into NH3/He mixture could quickly eliminate the which reduced the BEC of the analytes. With the increase in H2 flow rate,
interference of argide ions and cooperate with NH3 to eliminate the the concentration of H2 in CRC increased, and the high concentration of
interference. In the MS/MS mode, NH3/He and NH3/He/H2 were used H2 would collide with analytes to decrease the signal intensity of the
as the reaction gases to determine the 1 μg L− 1 of 23Na, 24Mg, 27Al, 39K, analytes, which reduced the BEC of the analytes. Fig. 2(b) shows the
and 51V in the simulated sample solution. Table 3 shows that for the effects of interfering elimination via different H2 flow rates for the
analyte ions using the on-mass method for interference elimination, the determination of Na, Mg, Al, K, V, Ni, Cu, and Zn by H2 as the reaction
yield did not change after the addition of H2. However, the BEC of an­ gas and a fixed NH3/He flow rate (3.0 mL min− 1). Fig. 2(b) shows that
alyte ions decreased, and the sensitivity increased. In the mass shift the BECs of Na, Mg, Al, K, V, Ni, Cu, and Zn were the lowest when the H2
method to eliminate interference, H2 can promote the addition of -NH flow rates were 1.8, 1.9, 2.0, 1.8, 1.9, 1.9, 2.0, and 1.7 mL min− 1,
and -NH2 to form -NH3 [35], favorable for the determination of 60Ni, respectively. In this study, the optimized H2 flow rate was selected as
63
Cu, and 66Zn. From the determination of 1 μg L− 1 of 60Ni, 63Cu, and 2.0 mL min− 1, which rendered a low BEC value of each element.
66
Zn in simulated sample solution, it was found that the use of reaction Fig. 2(c) shows the effects of interfering elimination via different O2
gas mixtures NH3/He/H2 not only increased the yield of analyte ions, flow rates for the determination of Ti, Cr, Mn, Fe, and Co by O2 as the
but also decreased the BEC of analyte ions, thus improving the sensi­ reaction gas and a fixed H2 flow rate (2.0 mL min− 1). As the O2 flow rate
tivity of analyte ions. Therefore, the NH3/He/H2 reaction mode was increased, the BECs of Ti, Cr, Mn, Fe, and Co gradually decreased to the
used in this experiment to eliminate interference in the determination of lowest value, and then the interferences were eventually eliminated.
Na, Mg, Al, K, V, Ni, Cu, and Zn used. When the O2 flow rate reached 0.30 mL min− 1, the BECs of the five el­
ements began to increase, suggesting that the O2 interference elimina­
3.2. Assignment of internal standard element tion capacity was close to the maximum value. In this study, the O2 flow
rate was set as 0.30 mL min− 1.
Different matrix components of the sample solution, calibration so­ Fig. 2(d) shows the effects of interfering elimination via different
lution, and blank solution lead to different transmission speeds, atomi­ NH3/He flow rates for the determination of Na, Mg, Al, K, V, Ni, Cu, and
zation efficiency, and signal intensities of the solutions, thereby Zn by using NH3/He as the reaction gas and a fixed H2 flow rate (2.0 mL
resulting in instrument instability, signal drift, and matrix effects. In the min− 1). The BECs of the eight elements decreased with increasing NH3/
present experiment, internal standard elements with similar behavior He flow rate. The BECs of Ni, Cu, and Zn reached the lowest under a
were selected for correction. In the no-gas mode of SQ, the analytes did specified NH3/He flow rate (2.6 mL min− 1), while the BECs of Na, Mg,
not participate in the mass shift reaction, and the selection of internal Al, K, and V still continued to decrease. When the NH3/He flow rate
standard elements was the same as that of ICP-MS. In the MS/MS mode, reached 3.0 mL min− 1, the BECs of the analytes began to increase,
there exists a mass shift reaction between analyzed elements in the CRC. indicating the complete elimination of interference. In this study, the
Based on the principle that internal standard elements with similar mass NH3/He flow rate was set at 3.0 mL min− 1.
numbers have similar mass spectrometry behaviors [36], 45Sc16O+ was
selected as the internal standard ion for the determination of Ti, Cr, Mn, 3.4. Linear relationship and LODs
Fe, and Co. 45Sc+ was selected as the internal standard ion for the
determination of Na, Mg, Al, K, and V; 45Sc(14NH3)+ 5 was selected as the The freshly prepared calibration standard solution series was directly
internal standard ion for the determination of Ni, Cu, and Zn. 115In+ was determined according to the proposed method. The range and correla­
selected as the internal standard ion for the determination of Ag, Cd, Sb, tion coefficient are shown in Table S2. The blank solution (ultrapure
and Ba; 209Bi+ was selected as the internal standard ion for the deter­ water) was analyzed 10 times using the established method, and the
mination of Pb. standard deviation of analyte was calculated. The LODs and limits of
quantification (LOQs) of the analytes were 3 times the standard

Table 3
Ion yields, sensitivities, and BECs of analytes using NH3/He and NH3/He/H2 as reaction gases in the MS/MS mode.
Isotope NH3/He NH3/He/H2

Monitoring ion Yield Sensitivity BEC Monitoring ion Yield Sensitivity BEC
(%) (cps/(L μg− 1)) (ng L− 1) (%) (cps/(L μg− 1)) (ng L− 1)
23
Na (1 μg L− 1) 23
Na+ 76.2 386,000 112 23
Na+ 77.5 414,000 83.7
24
Mg (1 μg L− 1) 24
Mg+ 73.1 253,000 16.2 24
Mg+ 72.6 286,000 12.0
27
Al (1 μg L− 1) 27 +
Al 65.8 329,000 43.0 27 +
Al 65.1 348,000 33.2
39
K (1 μg L− 1) 39 +
K 87.0 334,000 198 39 +
K 88.2 358,000 161
51
V (1 μg L− 1) 51 +
V 55.6 57,800 1.50 51 +
V 54.8 62,300 0.69
60
Ni (1 μg L− 1) 60
Ni(14NH3)+
3 4.83 12,400 18.3 60
Ni(14NH3)+
3 7.03 14,000 14.5
63
Cu (1 μg L− 1) 63
Cu(14NH3)+
2 6.50 8300 38.2 63
Cu(14NH3)+
2 9.20 10,200 30.1
66
Zn (1 μg L− 1) 66 14
Zn( NH3)+ 3.97 2100 92.6 66 14
Zn( NH3)+ 6.06 2560 71.0

4
L. Fu et al. Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

Fig. 2. Effect of different reaction gas flow rates to eliminate interference in the MS/MS mode. (a) H2, (b) H2, (c) O2, and (d) NH3/He.

deviation and 10 times the standard deviation, respectively [37]. Chongqing, Shanghai, Shenzhen, and Nanjing of China were determined
Table S2 shows that the analytes had a good linear relationship in the using the developed method. Each sample was repeatedly analyzed for
range of 0–10.0 μg L− 1, and the linear correlation coefficients were ≥ six times, and the results are shown in Table S4. The difference of metal
0.9996. The LODs of the analytes were in the range of 0.30–63.8 ng L− 1, impurity elements among different LiPF6 samples was significant: The
and the LOQs were in the range of 1.00–219 ng L− 1. contents of K, Ti, V, Cu, Zn, and Sb in sample A were obviously higher
than those in other samples, while the contents of Al, Fe, Ag, Cd, and Ba
3.5. Accuracy and precision of method were lower. The contents of Mn, Co, Ag, and Pb in sample B were higher
than those in other samples, while the contents of Mg, Ni, Cu, and Sb
Due to the lack of a standard reference material for LiPF6, the stan­ contents were lower. The contents of Al, Ni, Cd, and Ba in sample C were
dard solution was added to the sample solution in this experiment. The higher than those in other samples, while the contents of Na, K, Ti, V, Cr,
spiked solution was repeatedly analyzed six times using the developed and Sb were lower. The contents of Na, Mg, Cr, and Fe in sample D were
method. The spiked recovery and relative standard deviation (RSD) of higher than those in other samples, while the contents of Mn, Co, and Zn
analyte were calculated. SF-ICP-MS was used for comparative analysis. were lower. The metal impurity contents in the four LiPF6 samples were
In the SF-ICP-MS determination, based on qualitative scanning, three all lower than the standard specifications of the limit of ≤1 μg g− 1 in
resolutions were set as follows: 300 (low resolution, LR), 4000 (medium HGT4067–2015, satisfying the requirement for lithium-ion batteries and
resolution, MR), and 10,000 (high resolution, HR). Among them, LR production of lithium ion supercapacitors. The developed method is
mode was used to determine Ag, Cd, Sb, Ba, and Pb. MR mode was used applicable to the determination of metal impurity elements in LiPF6
to determine Na, Mg, Al, V, Ti, Mn, Co, Ni, Cu, and Zn, while HR mode products.
was used to determine K, Cr, and Fe. The results of two analytical
methods were statistically analyzed to evaluate the accuracy and pre­ 4. Conclusion
cision of ICP-MS/MS. The results are shown in Table S3. The spiked
recoveries of analytes were between 92.4 and 108% and the RSDs were By eliminating the interference with reaction gas mixtures, a new
≤ 5.67%. At the 95% confidence level, except Fe, the values determined method based on ICP-MS/MS was developed to determine 18 metal
for the other 17 elements had no significant difference with the results impurity elements in LiPF6. The LiPF6 samples were dissolved with ul­
obtained from SF-ICP-MS (P > 0.05), which verified the accuracy and trapure water and then directly injected for analysis. In the MS/MS
precision of the developed method. mode, by comparing the interference elimination effects of the reaction
gas mixtures systems (O2/H2, NH3/He/H2) and single reaction gas sys­
tems(O2, NH3/He) the reaction gas mixtures were selected to eliminate
3.6. Analysis of real samples
the spectral interference. The LODs of 18 metal impurity elements were
in the range of 0.30–63.8 ng L− 1. The spiked recoveries were 92.4%–
The LiPF6 samples (labeled as numbers A, B, C, and D) from

5
L. Fu et al. Spectrochimica Acta Part B: Atomic Spectroscopy 181 (2021) 106217

108%, and the RSD was ≤5.67%. The newly developed method has the [12] H. Liu, P. Zhang, H. Hu, Direct determination of impurity elements in high purity
lithium hexafluorophosphate dissolved with ethyl alcohol by inductively coupled
advantages of simple operation, fast analysis, and excellent accuracy
plasma mass spectrometry, Chin. J. Anal. Chem. 42 (2014) 913–917.
and precision, which is expected to provide technical support for the [13] L. Hanf, M. Diehl, L.S. Kemper, M. Winter, S. Nowak, Accessing copper oxidation
control and high-throughput determination of multiple metal impurity states of dissolved negative electrode current collectors in lithium ion batteries,
elements commonly found in high purity LiPF6. Electrophoresis 41 (2020) 1568–1575.
[14] S.D. Tanner, V.I. Baranov, D.R. Bandura, et al., Spectrochim. Acta B 57 (2002).
[15] N. Yamada, Kinetic energy discrimination in collision/reaction cell ICP-MS:
Author statement theoretical review of principles and limitations, Spectrochim. Acta B 110 (2015)
31–44.
[16] L. Balcaen, E. Bolea-Fernandez, M. Resano, F. Vanhaecke, Inductively coupled
Liang Fu: designed the experiment, supervised it, and participated in plasma – tandem mass spectrometry (ICP-MS/MS): a powerful and universal tool
the entire work. Hualin Xie: ICP-MS/MS and SF-ICP-MS experiments. for the interference-free determination of (ultra)trace elements -a tutorial review,
Jianhua Huang: Data statistic analysis. Xianhua Chen: writes manu­ Anal. Chim. Acta 894 (2015) 7–19.
[17] S.D. Fernandez, N. Sugishama, J.R. Encinar, A. Sanz-Medel, Triple quad ICPMS
scripts and revises manuscripts. Lin Chen: research design and data (ICPQQQ) as a new tool for absolute quantitative proteomics and
collection. All authors read and approved the final manuscript. phosphoproteomics, Anal. Chem. 84 (2012) 5851–5857.
[18] L. Balcaen, E. Bolea-Fernandez, M. Resano, F. Vanhaecke, Accurate determination
of ultra-trace levels of Ti in blood serum using ICP-MS/MS, Anal. Chim. Acta 809
Declaration of Competing Interest (2014) 1–8.
[19] S. Xing, M. Luo, N. Yuan, D. Liu, Y. Yang, X. Dai, W. Zhang, N. Chen, Accurate
determination of plutonium in soil by tandem quadrupole ICP-MS with different
All authors declare no conflict of interest. sample preparation methods, At. Spectrosc. 42 (2021) 62–70.
[20] E. Bolea-Fernandez, L. Balcaen, M. Resano, F. Vanhaecke, Overcoming spectral
overlap via inductively coupled plasma-tandem mass spectrometry (ICP-MS/MS).
Acknowledgments A tutorial review, J. Anal. At. Spectrom. 32 (2017) 1660–1679.
[21] E. Bolea-Fernandez, L. Balcaen, M. Resano, F. Vanhaecke, Potential of methyl
This study was supported by grants received from the Natural Sci­ fluoride as a universal reaction gas to overcome spectral interference in the
determination of ultratrace concentrations of metals in biofluids using inductively
ence Foundation of China (No. 81603400 and 81673585), National key
coupled plasma-tandem mass spectrometry, Anal. Chem. 86 (2014) 7969–7977.
R&D program of China (No. 2018YFC1703400 and 2018YFC1704400), [22] K. Boting, S. Treu, P. Leonhard, C. Heiß, N.H. Bings, First experimental proof of
Training Program for Excellent Young Innovators of Changsha (No. asymmetric charge transfer in ICP-MS/MS (ICP-QQQ-MS) through isotopically
kq1802017), China Postdoctoral Science Foundation (No. enriched oxygen as cell gas, J. Anal. At. Spectrom. 29 (2014) 578–582.
[23] M. Resano, M. Aramendía, F.V. Nakadi, E. Garcia-Ruiz, C. Alvarez-Llamas,
2019M652776), and Natural Science Foundation of Chongqing (No. N. Bordel, J. Pisonero, E. Bolea-Fernandez, T. Liu, F. Vanhaecke, Breaking the
cstc2019jcyj-msxmX0705). boundaries in spectrometry. Molecular analysis with atomic spectrometric
techniques, TrAC-Trend. Anal. Chem. 129 (2020) 115955.
[24] L. Fu, H. Xie, J. Huang, L. Chen, Rapid determination of trace elements in serum of
Appendix A. Supplementary data hepatocellular carcinoma patients by inductively coupled plasma tandem mass
spectrometry, Anal. Chim. Acta 1112 (2020) 1–7.
Supplementary data to this article can be found online at https://doi. [25] C. Walkner, R. Gratzer, T. Meisel, S.N.H. Bokhari, Multi-element analysis of crude
oils using ICP-QQQ-MS, Org. Geochem. 103 (2017) 22–30.
org/10.1016/j.sab.2021.106217. [26] A. Virgilio, R.S. Amais, C.D.B. Amaral, L.L. Fialho, D. Schiavo, J.A. Nóbrega,
Reactivity and analytical performance of oxygen as cell gas in inductively coupled
plasma tandem mass spectrometry, Spectrochim. Acta B 126 (2016) 31–36.
References
[27] Y. Zhang, Z. Pan, P. Jiao, J. Ju, T. He, T. Duan, H. Cai, Solvent extraction ICP-MS/
MS method for the determination of REE impurities in ultra-high purity Ce
[1] J. Liu, X. Li, Z. Wang, H. Guo, W. Peng, Y. Zhang, Q. Hu, Preparation and chelates, At. Spectrosc. 40 (2019) 167–172.
characterization of lithium hexafluorophosphate for lithium-ion battery [28] M.A. Dexter, H.J. Reid, B.L. Sharp, The effect of ion energy on reactivity and
electrolyte, T. Nonferr. Metal. Soc. 20 (2010) 344–348. species selectivity in hexapole collision/reaction cell ICP-MS, J. Anal. At.
[2] C. Shen, D. Xiong, L.D. Ellis, K.L. Gering, L. Huang, J.R. Dahn, Using the charge- Spectrom. 17 (2002) 676–681.
discharge cycling of positive electrode symmetric cells to find electrolyte/electrode [29] Q. Sun, D. Yuan, Z. Chen, M. Megharaj, R. Naidu, Reduction of polyatomic
combinations with minimum reactivity, J. Electrochem. Soc. 164 (2017) interferences during ion-chromatographic speciation of metal ions via their EDTA
A3349–A3356. complexes along with ICP-MS detection using an octopole reaction system,
[3] L.O. Valøen, J.N. Reimers, Transport properties of LiPF6-based Li-ion battery Microchim. Acta 169 (2010) 41–47.
electrolytes, J. Electrochem. Soc. 152 (2005) A882–A891. [30] J. Vidmar, P. Oprckal, R. Milacic, A. Mladenovic, J. Scancar, Investigation of the
[4] S.S. Zhang, A review on electrolyte additives for lithium-ion batteries, J. Power behaviour of zero-valent iron nanoparticles and their interactions with Cd2+ in
Sources 162 (2006) 1379–1394. wastewater by single particle ICP-MS, Sci. Total Environ. 634 (2018) 1259–1268.
[5] L. Terborg, S. Nowak, S. Passerini, M. Winter, U. Karst, P.R. Haddad, P. [31] L. Fu, H. Xie, S. Shi, Multielement analysis of Zanthoxylum bungeanum maxim.
N. Nesterenko, Ion chromatographic determination of hydrolysis products of Essential oil using ICP-MS/MS, Anal. Bioanal. Chem. 410 (2018) 3769–3778.
hexafluorophosphate salts in aqueous solution, Anal. Chim. Acta 714 (2012) [32] Y. Ogra, T. Hirata, Metallomics, Springer, Germany, 2017.
121–126. [33] C.D.B. Amaral, R.S. Amais, L.L. Fialho, D. Schiavo, T. Amorim, A.R.A. Nogueira, F.
[6] S. Nowak, M. Winter, Elemental analysis of lithium ion batteries, J. Anal. At. R.P. Rocha, J.A. Nobrega, A novel strategy to determine as, Cr, hg and V in
Spectrom. 32 (2017) 1833–1847. drinking water by ICP-MS/MS, Anal. Methods 7 (2015) 1215–1220.
[7] E. Peled, S. Menkin, Review-SEI: past, present and future, J. Electrochem. Soc. 164 [34] L. Fu, S. Shi, X. Chen, Ultra-trace metal elements analysis of high purity
(2017) A1703–A1719. tetramethylammonium hydroxide using inductively coupled plasma tandem mass
[8] B. Vortmann-Westhoven, M. Winter, S. Nowak, Where is the lithium? Quantitative spectrometry, Chin. J. Anal. Chem. 46 (2018) 107–112.
determination of the lithium distribution in lithium ion battery cells: investigations [35] P. Petrov, B. Russell, D.N. Douglas, H. Goenaga-Infante, Interference-free
on the influence of the temperature, the C-rate and the cell type, J. Power Sources determination of sub ng kg− 1 levels of long-lived 93Zr in the presence of high
346 (2017) 63–70. concentrations (μg kg− 1) of 93Mo and 93Nb using ICP-MS/MS, Anal. Bioanal. Chem.
[9] S. Nowak, M. Winter, The role of cations on the performance of lithium ion 410 (2018) 1029–1037.
batteries: a quantitative analytical approach, Acc. Chem. Res. 51 (2018) 265–272. [36] E. Bolea-Fernandez, A. Rua-Ibarz, M. Resano, F. Vanhaecke, To shift or not to shift:
[10] S.J. An, J. Li, C. Daniel, D. Mohanty, S. Nagpure, D.L. Wood, The state of adequate selection of an internal standard in mass-shift approaches using tandem
understanding of the lithium-ion-battery graphite solid electrolyte interphase (SEI) ICP-mass spectrometry (ICP-MS/MS), J. Anal. At. Spectrom. (2021), https://doi.
and its relationship to formation cycling, Carbon 105 (2016) 52–76. org/10.1039/D0JA00438C.
[11] Y. Fang, Z. Hao, Y. Song, C. Sun, J. Yu, L. Yu, Determination of twenty one [37] J. Uhrovcik, Strategy for determination of LOD and LOQ values-some basic aspects,
elements in lithium hexafluorophosphate by ICP-AES, Spectrosc. Spectr. Anal. 25 Talanta 119 (2014) 178–180.
(2005) 280–282.