Journal of Solid State Electrochemistry (2021) 25:2849–2859

https://doi.org/10.1007/s10008-021-05060-3

ORIGINAL PAPER

Formation of lithiated gold and its use for the preparation of reference
electrodes — an EQCM study
Christopher Behling1,2 · Karl J. J. Mayrhofer1,2 · Balázs B. Berkes1

Received: 29 June 2021 / Revised: 28 September 2021 / Accepted: 29 September 2021 / Published online: 11 October 2021
© The Author(s) 2021, corrected publication 2022

Abstract
Lithiated gold wires can be used to build reference electrodes with outstanding potential stabilities over several days and
even over the course of one year. These electrodes are well suited for investigations in the context of lithium-ion batteries
(LIBs). In this work, a detailed procedure for the preparation of such electrodes with tailored mechanical properties, which
can be fitted gastight into electrochemical cells using commercially available fittings, is given. The electrochemical lithiation
process is studied using the electrochemical quartz crystal microbalance (EQCM) technique, and the differences in lithiation
of wire type and thin film type gold electrodes are discussed. All experiments were carried out with two different electrolytes,
namely, a ­LiPF6 and a lithium bis(trifluoromethane sulfonyl) imide (LiTFSI)-based electrolyte, and we conclude that for a
higher lithiation rate and long-term stability, the use of LiTFSI-based electrolyte in the preparation phase is beneficial. The
EQCM data provides a better insight in the analysis of film formation processes, like the buildup of the solid electrolyte
interphase (SEI) during the lithiation, the rate of deposition of metallic lithium, or additional information on the kinetics of
Li-Au alloy formation.

Keywords Gold-lithium alloy · SEI formation · LiPF6 · LiTFSI · Lithiation · Reference electrodes

Introduction development of sophisticated measurement methods. Studies

in full battery cells are in that sense sometimes limiting, as
Lithium-ion batteries (LIBs) represent the most important they always contain overlapping information of both, anodic
achievement among modern electrochemical energy stor- and cathodic, processes. In contrast, three electrode elec-
age devices that fundamentally changed our society through trochemical cells, which employ a reference electrode with
portable electronics and electric vehicles. The ever-growing well-defined and stable electrode potential, can separate the
demand for higher energy density LIBs requires new characterization of cathode and anode active materials and
ways and technologies to meet the needs for their future thus provide more straightforward insights into half reac-
development. An important step in battery performance tions. However, the choice of the right reference electrodes
improvement is gaining deeper understanding of underly- in nonaqueous systems is not obvious. Due to the large num-
ing electrochemical processes, accompanied by thorough ber of applied solvents, solvent mixtures, and electrolyte
electrochemical testing and characterization, as well as the salts, it is hardly possible to find a commercial reference
electrode fitting all the experimental requirements. Aqueous
This paper is dedicated to Prof. Inzelt on the occasion of his 75th electrochemistry in this respect is in many cases simpler, and
birthday. companies offer many possibilities for commercially avail-
able, diverse reference electrodes. An excellent compilation
* Balázs B. Berkes of both aqueous and nonaqueous reference electrodes can
[email protected]
be found in the Handbook of Reference Electrodes edited
1
Helmholtz Institute Erlangen-Nürnberg for Renewable by G. Inzelt et al. [1].
Energy (IEK-11), Forschungszentrum Jülich, Cauerstr. 1, In case of LIB research, the Li/Li+ system, a piece of
91058 Erlangen, Germany lithium wire or foil in contact with a lithium ion contain-
2
Department of Chemical and Biological Engineering, Fried ing electrolyte as half-cell, is probably the most widely used
rich-Alexander-Universität Erlangen-Nürnberg, Cauerstr. 1, reference electrode. But in many cases, this half-cell serves
91058 Erlangen, Germany

13
Vol.:(0123456789)
2850 Journal of Solid State Electrochemistry (2021) 25:2849–2859

as both the counter and the reference electrode at the same described in [13], although in this case not a whole wire was
time. This might be problematic even with well-established lithiated, but only the tip of an insulated gold wire which was
cell designs, such as coin type cells, Swagelok T cells, or exposed to the electrolyte (the cross sectional area). Most of
pouch/coffee bag cells due to the known issue with lithium the remaining gold lithiation experiments reported utilize
dendrite formation, especially in consecutive charge/discharge thin films, sputtered either onto a Cu substrate [20, 21] or
experiments. Another problem of using Li as counter and ref- onto a W wire [14].
erence electrode simultaneously has been demonstrated by During the first gold wire lithiations with a ­LiPF6 con-
LaMantia et al. [2]. At high current densities, the behavior of taining electrolyte, we experienced many peculiarities with
Li becomes stochastic; i.e., the potential difference between huge discrepancies between the reported current–time
lithium and the working electrode (WE) changes due to the curves and our own results. Switching from ­LiPF6 to lithium
polarizability of lithium. To overcome these problems, lithium bis(trifluoromethane sulfonyl) imide (LiTFSI) just partially
might be used as the reference electrode by inserting it into solved the problem, so we decided to investigate the phe-
the separator or in between two separator sheets [3–5]. This nomena deeper, utilizing the electrochemical quartz crystal
insertion of lithium metal can, however, be problematic due microbalance (EQCM) technique. It has the advantage that
to the mechanical properties of the metal. It is soft, malleable, it employs thin Au films deposited onto quartz, thereby con-
and easy to deform. Therefore, it is usually attached to a cop- forming to previous studies, while also being a very sensitive
per, nickel, or stainless-steel wire. The size of the so obtained technique for surface mass changes. Due to big differences
reference electrodes, however, is relatively big and can disturb in the construction and composition of the SEI in the two
the electric field between the cathode and anode and falsify applied electrolytes, this latter aspect seemed to be espe-
sensitive measurements, like impedance spectroscopy, via cially beneficial.
non-ideal positioning relative to the other electrodes [6, 7]. The EQCM is mainly a surface mass sensitive method,
An alternative solution to lithium is the use of cathode widely used in investigations including metal and alloy dep-
materials, like ­Li4Ti5O12 or ­LiFePO4 [2, 8] or lithium metal osition, underpotential deposition (UPD) or electroplating
alloys [9, 10] for the preparation of reference electrodes for just to mention a few. A book chapter written by G. Inzelt
nonaqueous electrolytes. And the use of so-called point type gives a good overview of the method through selected appli-
reference electrodes helps to significantly improve the qual- cations by the author [22]. The mass changes occurring dur-
ity and reproducibility of sensitive measurements. These ing deposition processes together with the simultaneously
reference electrodes are composed of a thin wire with an measured electrochemical data can be very helpful for the
electronic insulator on its surface, being exposed to the elec- elucidation of reaction mechanisms via identification of the
trolyte only at its end. Intensive studies showed that metal species accumulated on the surface. The method has also
wires forming an alloy with lithium show superior stability been used for investigations of LIB systems. In the 1990s,
compared to in situ deposited thin lithium films [11–13]. it was used to study the behavior of battery electrolyte solu-
However, through a continuous growth of the solid elec- tions on Au electrodes [23], for the study of the deposi-
trolyte interphase (SEI), such thin films can be delithiated, tion and stability of SEI forming Li compounds in the UPD
limiting the lifetime of the reference electrode. region of lithium [24] or to monitor the mass change of elec-
Recently, gold has attracted the attention of researchers in the trodeposited Li on Ni-coated crystals [25]. Nowadays, it is
field of LIBs as a possible reference electrode after lithiation [13, mainly used to investigate intercalation processes at both the
14]. Both cited works describe the use of gold for the preparation cathode and anode or formation of SEI on graphite, usually
of micrometer sized wire reference electrodes. In earlier stud- together with dissipation monitoring taking into account the
ies, as a noble metal gold was rather used as a model system for porous structure of the layers [26, 27]. However, to the best
fundamental investigations, like the reduction or oxidation reac- of our knowledge, it has so far never been used to monitor
tions of battery electrolytes [15], however, its electrochemical alloy formation between Li and Au during long-term gal-
lithiation has also been reported as early as 1971 [16]. vanostatic Li deposition that is of interest in relation with
For the real-time monitoring of dissolution processes the preparation of our PEEK embedded lithiated gold wire
in nonaqueous electrolytes, we have recently developed an reference electrodes (AuLi-Ref).
electrochemical flow cell [17–19]. One particular challenge As it has been very well described in the preface of [1]:
for battery research with this approach was the reference “In order to produce reliable experimental data, reference
electrode, which also herein needs to feature reproducible electrodes have to fulfill many, partly contradictory condi-
and stable electrode potentials, a well-defined shape, and tions, which practically never can be met to full satisfac-
mechanical stability so that a tight installation into close tion. For the choice and application of a particular reference
vicinity of the WE is possible. We decided for a gold wire electrode it is of utmost importance to know its properties
reference electrode mounted into a PEEK capillary as a as detailed as possible.” Therefore, we decided to dedicate
housing for the wire. The lithiation process of a gold wire is a detailed study to the preparation of macroscopic sized Au/

13
Journal of Solid State Electrochemistry (2021) 25:2849–2859 2851

Li reference electrodes that can be used in a huge variety of to a copper wire served as the counter electrode (CE), while
electrochemical cells to study nonaqueous systems, espe- the gold wires or EQCM-crystals were used as the WEs.
cially LIBs.
Calibration
Experimental
The calibration of the EQCM was carried out using a spi-
ral Ag wire as the counter and reference electrode and
Materials
0.1 mol·L−1 ­AgClO4 added into the different electrolytes.
Silver was chosen as the calibration system because its redox
The battery electrolytes ­LiPF6 (1 mol·L−1 in EC:EMC, 3:7
potential lies within the stability windows of the electrolytes,
by wt.) and LiTFSI (1 mol·L−1 in EC:EMC, 3:7 by wt.) used
while also not being able to form alloys with the used Au
for the experiments were bought from E-Lyte Innovations
crystals. During the calibration, constant current densities of
GmbH (Münster, Germany). The silver salt ­AgClO4 (anhy-
200, 20, and 2 µA·cm−2 were applied consecutively between
drous) which was used for the calibration of the EQCM was
the Au quartz and the Ag wire for 2 min each while moni-
obtained from Alfa Aesar (by Thermo Fisher GmbH, Kan-
toring the frequency change. First negative currents were
del, Germany). Silver (99.9%; 0.25 mm outside diameter,
applied to deposit silver, then positive currents were used
OD), gold (99.9%; 0.25 mm OD), and platinum (99.99%;
to redissolve the deposited Ag atoms. The thereby calcu-
0.25 mm OD) wires as well as lithium foil (99.8%; 0.5 mm)
lated calibration factor is obtained as the mean mass per Hz
were from MaTecK (MaTecK Material-Technolgie &
change of all applied current densities (ng·Hz−1) and is used
Kristalle GmbH, Jülich, Germany). The glassy carbon rods
for the conversion of frequency changes to mass values in
(SIGRADUR® G; 1 mm OD) used for potential stability
the subsequent experiments.
measurements of the AuLi-Ref were purchased from HTW
Hochtemperatur-Werkstoffe GmbH (Thierhaupten, Ger-
many). All crystals for the EQCM, namely, Au-Ti (5 MHz,
100 nm Au on thin Ti adhesion layer) and Au-Cr (5 MHz, Results and discussions
100 nm Au on thin Cr adhesion layer), were obtained from
MicroVacuum Ltd. (Budapest, Hungary). The PEEK elec- Preparation and testing of the PEEK embedded
trode bodies (LF-1.6; Leak-Free Reference Electrode Junc- lithiated gold wire reference electrodes
tion, 1.6 mm OD, 1.0 mm ID), which were used for the
assembly of the reference electrodes, were ordered from Based on previous lithiation experiments from the literature
Innovative Instruments, Inc. (Tampa, FL, USA). [11, 14], 200 µA·cm−2 current density was used to lithiate
a gold wire with 250 µm outer diameter. The first experi-
Instruments ments were carried out in a commercial battery electrolyte
(1 mol·L−1 ­LiPF6 in EC:EMC, 3:7 by wt.). The lithiation
All experiments were performed inside an Ar filled glove- time has been chosen to be 72 h. However, the typical E − t
box (M. BRAUN Inertgas-Systeme GmbH, Garching, Ger- curves, containing the characteristic plateaus, reported in the
many) with water and oxygen levels < 1 ppm. Electrochemi- literature for thin film electrodes could not be reproduced.
cal measurements were conducted with either an SP-300 or Switching to a LiTFSI based electrolyte, which is widely
a VSP-300 potentiostat by Biologic Sciences Instruments used for lithiation experiments [14, 20, 21], and trying dif-
(Seyssinet-Pariset, France). The EQCM device (eQCM-I ferent current densities only slightly changed the shapes of
Mini) was obtained from MicroVacuum Ltd. (Budapest, the E − t curves, and well-pronounced plateaus could not be
Hungary). observed. The equilibrium electrode potentials (equilibrium
open circuit voltages, OCV) of the wires, however, were
Cells very close to the potentials reported in [13]. Figure 1a shows
typical chronopotentiometric (CP) curves obtained with gold
EQCM measurements were carried out in a temperature wires in LiTFSI-based electrolyte and the subsequent OCV
controlled cell of the type “QSHE-open” at 25 °C (Micro- values for 48 h. Even if the lithiation curves slightly differed,
Vacuum Ltd.) using a modified lid. Lithiation experiments the equilibrium potentials of the lithiated gold wires con-
of gold wires were performed in homemade 2-electrode cells verged to a stable value in all cases (Fig. 1b). For a gold wire
of either polypropylene or glass. These cells were built by having a radius of r and a length of l immersed into the solu-
modification of chromatography sample vials of varying vol- tion, the ratio of the amounts of lithium and gold (nLi/nAu)
umes depending on the desired immersed length of the wires after a lithiation time t with a constant current density i,
into the electrolyte. In all cells, a piece of lithium foil fixed assuming 100% faradaic efficiency (FE) and neglecting the

13
2852 Journal of Solid State Electrochemistry (2021) 25:2849–2859

Fig. 1  a Five CP curves for

the lithiation of gold wires in
1 mol·L−1 LiTFSI containing
EC:EMC (3:7 by wt.) with
200 µA·cm−2 current density
applied over 72 h and b OCV
curves of five Au wires lithiated
as in a all converging to a stable
equilibrium electrode potential

cross sectional area of the thin wire, can be calculated by the with an aqueous solution containing ­H2SO4 to maintain a
following formula: wet junction. In order to avoid cross reactions of the acidic
aqueous contents of the body with the later introduced elec-
nLi i × 2r𝜋 × l × t MAu 2 × i×M Au
= × 2 = ×t =k×t trolyte and lithiated wire, a thorough cleaning procedure
nAu F r 𝜋 × l × 𝜌Au F × r × 𝜌Au
was developed. This procedure consists of three stages with
where F is the Faraday constant and MAu and ρAu are the three different solvents. In the first stage, the body is repeat-
molar mass and density of gold, respectively. In our case, edly purged with water using a syringe and cleaned inside
k = 3.4 × 10−6 s−1, i.e., after 72 h lithiation, the ratio of the a water-filled beaker in an ultrasonication bath outside the
two metals in the wire is close to 1 (0.9). Nevertheless, this glovebox; after that, the same steps are repeated in the sec-
ratio is enough to keep the potential of the AuLi-Ref created ond stage, this time with EtOH. After the second stage, the
from such a wire constant for an extended period of time, as body is set to rest in a EtOH filled vial. In the third stage, the
will be demonstrated below. However, its potential is still body is again repeatedly purged and cleaned in the sonica-
ca. + 310 mV positive to Li, which will be different for thin- tion bath, this time with target solvent mix of EC:EMC (3:7
film electrodes with a much higher nLi/nAu ratio. by wt.). After the final cleaning, the body is placed inside
Such lithiated Au wires offer many advantages for the a sealed vial filled with EC:EMC and later introduced into
preparation of REs: (i) their size can be flexibly varied and the glovebox for assembly of the final electrode. Before the
with that the size of the RE; (ii) they can be placed in a lithiated gold wire is introduced into the body, the body is
Luggin-capillary, as demonstrated below, in an easy man- filled with the target electrolyte depending on the electrode’s
ner; (iii) they offer good mechanical stability unlike thin desired field of use (1 mol·L−1 LiTFSI or 1 mol·L−1 ­LiPF6
Li wires; (iv) gold has excellent electrical conductivity and in EC:EMC, 3:7 by wt.) (Fig. 2a). After that, the wire is
can be directly connected to the cables of the potentiostat introduced into the body (Fig. 2b). Finally, a wire end ferrule
at the free end of the wire; and (v) after proper lithiation its is placed on top of the body and wire. The metal part of the
potential stays very stable even for long experimental times. ferrule is crimped onto the wire to ensure electric contact,
In the following part, we discuss the preparation of PEEK while the cavities between PEEK body and wire ferrule are
housed AuLi-Refs from the wires. sealed using a two-component epoxy resin to prevent evapo-
For the lithiation, the gold wire is placed in a simple ration of the electrolyte and ensure mechanical stability of
homemade cell filled with LiTFSI electrolyte (1 mol·L−1; the electrode (Fig. 2c). The electrode is then placed in a
EC:EMC, 3:7 by wt.). While the gold wire is operated as 1.5-mL vial filled with the target electrolyte to prevent the
WE, a piece of lithium foil which is attached to a copper junction from drying out. After complete hardening of the
wire is used as the CE. The wire is lithiated with a constant epoxy resin (ca. 24 h), the electrode can be used.
current of 200 µA·cm−2 according to Zhou et al. and Simon In order to test the long-term stability of the refer-
et al. [11, 14]. To ensure sufficient lithiation and avoid future ence electrode, it has been applied in a three electrode
potential instabilities (as is demonstrated below), the lithi- cell configuration containing a glassy carbon WE, a Pt
ation is carried out for 72 h. After successful lithiation, the wire CE and 10 mmol·L−1 ferrocene (ferrocene is bis-(η-
wire is transferred into a PEEK electrode body of 1.6 mm cyclopentadienyl)iron(II)) dissolved in 1 mol·L−1 LiTFSI
outer and 1 mm inner diameter equipped with a leak free containing EC:EMC (3:7 by wt.). The reason of using fer-
junction at the bottom end. The PEEK electrode body is rocene is first that the ferrocene (Fc)/ferrocenium (­ Fc+)
intended for aqueous use and therefore is shipped filled couple as a reference redox system is recommended for

13
Journal of Solid State Electrochemistry (2021) 25:2849–2859 2853

Fig. 2  Preparation steps of

1.6 mm PEEK housed AuLi-Ref a) b) c)
from lithiated gold wires

reporting electrode potentials in nonaqueous media [28]; (1:1 by vol.) solvent mixture and concluded that the Fc/
therefore, the value of the equilibrium electrode potential Fc+ redox couple is reversible even in this highly concen-
of the Fc/Fc+ couple should be reported for the newly cre- trated electrolyte.
ated RE. Second and luckily, this compound demonstrates Figure 3a demonstrates this excellent reversibility with a
good solubility, invariant redox potentials, and excellent cyclic voltammogram obtained in 10 mmol·L−1 Fc contain-
chemical and electrochemical reversibility in the studied ing LiTFSI-based electrolyte. The long-term stability has
battery electrolytes. It has been shown to be a potentially been checked by carrying out cyclic voltammetric experi-
useful redox reagent for the chemical overcharge protec- ments every 1 h in the above-described cell. The half wave
tion of rechargeable lithium and lithium-ion batteries and potential E1/2 has been determined from each CV curve from
therefore has been studied in detail by Laoire et al. [29]. the anodic (Epa) and cathodic (Epc) peak potentials using
The authors studied the kinetics of oxidation of Fc to ­Fc+ a homemade software with regard to the (huge number ) of
in the non-aqueous lithium ion conducting electrolyte CVs. The E1/2 values were estimated as 1∕2 Epa + Epc . Fig-
composed of a solution of 1 mol·L−1 ­LiPF6 in EC:EMC ure 3b shows the change of E1/2 directly after preparation

a) b) c)

Fig. 3  a Cyclic voltammetric response of ferrocene (10 mmol·L−1 in 1 mol·L−1 LiTFSI containing EC:EMC (3:7 by wt.) with 100 mV·s−1 scan
rate and b, c stability of potential over time

13
2854 Journal of Solid State Electrochemistry (2021) 25:2849–2859

of the AuLi-Ref (dark green curve) and after 1 year of use CP lithiation curves showed the characteristic shape, known
(dark red curve). The value of E1/2 shifted ca. 100 mV in from the literature in both electrolytes. Figure 4 shows the
one year. The change of E1/2 within the timeframe of 50 h results of measurement sequences consisting of a CP part with
can be estimated by deriving the E1/2(t) function and is dis- 200 µA·cm−2 for 4 h followed by an OCV period for 40 h.
played in Fig. 3c. The values are in the lower mV/h range The potential values are IR corrected values, using the current
and therefore exhibit better potential stability than common interruption method for 50 ms, after the CP part. The first two
Li-foil REs (see Fig. S1). Here we would like to note that we potential plateaus shown in the insets of Fig. 4a and b belong
used the original E1/2(t) function for the derivation without to phase transformations of the L ­ ixAuy alloys. In the first step
smoothing the curve in advance. To be able to account even of alloying, Bach et al. found a hexagonal structure with com-
for these minimal changes in the potential of the AuLi-Ref, position ­Li5Au3 [21]. The second plateau corresponds to the
we always check its potential with the Fc/Fc+ redox couple thermodynamically stable L ­ i3Au phase with face centered
both before and after the actual measurements. cubic structure. The second plateau ends at ca. 4000 s and is
followed by the deposition of metallic lithium onto the formed
Lithiation of Au films — insights into potential alloy. In case of the thin Au film the nLi/nAu ratio as function of
stability lithiation time can be determined by the formula:
nLi i × r2 𝜋 × t MAu i×M Au
The potential stability of lithiated Au electrodes can be inves-
�
= × 2 = ×t =k ×t
nAu F r 𝜋 × h × 𝜌Au F × h ×
tigated using thin film (100 nm) Au deposited onto quartz. The 𝜌 Au

a) b)

c) d)

Fig. 4  Lithiation of 100-nm-thick Au EQCM electrodes for 4 h in a ured frequency change data (green curves). Apparent molar masses
1 mol·L−1 ­LiPF6 containing EC:EMC (3:7) and b 1 mol·L−1 LiTFSI calculated from the frequency change data during the galvanostatic
containing EC:EMC (3:7) with 200 µA·cm−2 current density with the lithiation experiments for c ­LiPF6 and d LiTFSI-based electrolytes
subsequent OCV periods (purple curves) and simultaneously meas-

13
Journal of Solid State Electrochemistry (2021) 25:2849–2859 2855

where h is the height of the gold film (h = 100 nm) and the resulting in Mpe >> 7 g·mol−1. At the end of the potential
side area of the cylinder has been neglected. In this case plateaus, the Mpe is close to zero, i.e., no mass change,
k = 2.1 × 10−3 s−1; i.e., after ca. 473 s, the nLi/nAu = 1 ratio neither related to SEI formation nor to Li intercalation
�

is achieved. Although the length of the plateaus is not well accompanies these phase transformations. In the last phase
correlated to this value (formation of ­Li3Au would need ca. of the 4-h long deposition starting at ca. 4000 s and for the
1418 s), it can be seen in Fig. 4 that for both electrolytes, rest of the CP period, the Mpe is very close to 7 g·mol−1
a stable potential of the lithiated Au electrode is achieved (the horizontal line at the figure is a constant function
under these conditions. The reason for the huge discrepancy at 7 g·mol−1) for the LiTFSI-based electrolyte. The total
between the calculated lithiation times and observed length frequency change (green line) is ca. half as much for the
of plateaus is most probably the invalidity of the assumption ­LiPF6 based electrolyte as for the LiTFSI-based electrolyte
that lithium deposition occurs with 100% FE. At this point, (ca. 95.9 µg for ­LiPF6 vs. ca. 216.2 µg for LiTFSI) at same
the analysis of the simultaneously measured EQCM data applied current density until the end of the lithiation. The
greatly helps in the explanation. reason for this is most probably the different SEI structure
Very interestingly, the frequency change curves (green for the two electrolytes. Therefore, we can say that while
lines in Fig. 4) also reflect the formation of the differ- both Au electrodes — lithiated in the two different elec-
ent alloy phases: the rate of frequency (i.e., mass) change trolytes — will exhibit stable potentials over a prolonged
decreases towards the end of the potential plateaus, i.e., period of time, it seems that Li deposition happens more
towards the completeness of the stable phase and remains readily through an SEI built up in LiTFSI. This difference
constant for the rest of the lithiation time (from ca. 4000 s in SEI build up has also a huge influence on the long-term
to 4 h). It is possible to calculate apparent molar masses potential stability of lithiated Au electrodes prepared in
from the measurements after calibration and assum- the two electrolytes with shorter lithiation times as will
ing the validity of the Sauerbrey-equation [22], which be demonstrated below.
states that the frequency change is directly proportional
to the mass change ( Δf ∝ −Δm ). It should be noted that
especially due to the formation of a possibly non-rigidly Short‑term lithiation — insights into the properties
attached SEI at the crystal’s surface, the applicability of of SEI
the Sauerbrey-equation is not given without restrictions.
The good agreement of theoretical and calculated mass In order to see how the potential stability of lithiated Au
change values, however, validates its use in this situa- electrodes is influenced by lithiation times, experiments
tion and also provides a clear basis for following calcula- with CPs for only 1000 s instead of 4 h have been con-
tions. We have carried out the calibration of the EQCM ducted (Fig. 5). Although after 1000 s lithiation, a theoreti-
cell by using 0.1 mol·L−1 ­AgClO4 dissolved in the battery cal nLi/nAu ratio of ca. 2 is achieved; in both cases, the E
electrolytes. ­AgClO4 dissolves very slowly, and to reach vs t curves do not stay constant and rapidly change most
a concentration of 0.1 mol·L−1, solutions were left over- probably after the spontaneous delithiation of gold under
night to dissolve. Surprisingly, the value of the integral OCV conditions. This is less surprising if considering that
sensitivity of the crystals was very close to the theoreti- completion of the lithium alloy formation is only achieved
cal one for LiTFSI-based electrolytes (17.3 ng·Hz−1·cm−2 after ca. 1400 s. Interestingly, however, the mass of the elec-
calculated vs. 17.7 ng·Hz−1·cm−2 theoretical), while the trode in ­LiPF6 decreases first right after the CP period and
difference for the L ­ iPF6-based electrolyte was somewhat stays more or less constant for LiTFSI. It increases for both
bigger (14.9 ng·Hz−1·cm−2 calculated). Although ­PF6− is electrolytes as the potential starts to rapidly increase. Fig-
supposed to be a weakly coordinating anion, it seems to ure 5 demonstrates an important difference between the two
interact strongly to the Au surface to hinder the deposition electrolytes: the potential stays stable for more than 30 h in
of Ag (we can rule out competing electrode reactions at LITFSI and for only ca. 10 h in ­LiPF6. We speculate that the
the potential of Ag deposition, which is ca. 3.8 V vs. Li). initial mass loss for ­LiPF6 is not related to lithium loss, since
Apparent molar masses or Mpe (mass per electron) values as we described above, the L ­ iPF6 type SEI does not facilitate
have been calculated for the lithiation experiments and are Li transfer as much as the LiTFSI type and the potential also
shown in Fig. 4c and d. The curves are the derivatives of stays in this phase rather constant. Maybe the inner inorganic
the corresponding m = m(Q/F) functions and are smoothed SEI layer [30] is being dissolved in lack of a thicker and
by adjacent averaging taking 50 adjacent points for the protective organic SEI in this region. However, this problem
sake of better visibility. needs further investigations to be able to convincingly prove
It is evident from the change of the Mpe values, which our speculation. The delithiation happens as the potential
decrease asymptotically after applying the constant lithia- increases. In this phase, Li is being oxidized and electrolyte
tion current that first an SEI is being built on the electrode reduced and deposited onto the electrode, causing the mass

13
2856 Journal of Solid State Electrochemistry (2021) 25:2849–2859

a) b)

c) d)

Fig. 5  Lithiation of 100-nm-thick Au EQCM electrodes for 1000 s in change data (green curves) demonstrating the instability of OCV
a 1 mol·L−1 ­LiPF6 containing EC:EMC (3:7) and b 1 mol·L−1 LiTFSI values. The initial 50 s after the application of the constant current
containing EC:EMC (3:7) with 200 µA·cm−2 current density with the together with the rate of frequency changes are shown in c for L­ iPF6
subsequent OCV periods and simultaneously measured frequency and d for LiTFSI based electrolytes

increase in both cases. These results demonstrate that lithi- simultaneously with lithiation of graphite electrode. The
ated gold electrodes will suffer potential instability on the authors found that “in the second stage, a stable, compact,
long term if not enough Li is deposited onto the electrode. and highly conductive SEI film is produced.” When look-
Before the first plateau, as shown in Fig. 5c and d, a ing at the frequency change curves between 300 and 320 s
potential dip can be observed between 310 and 330 s. This also two kinds of stages can be identified. However, whereas
is in accordance with previous observations [21]. The use this first phase can be characterized in L­ iPF6-based electro-
of EQCM provides a better insight into the initial reactions. lyte with a concave function, in LiTFSI-based electrolyte,
Right after the application of the constant current, the rate of a convex function is observed. After this ca. 20–30 s, the
frequency change is less than half of that registered for the lithiation rate becomes constant and will decrease again only
first plateau and showed by the fitted linear curves. There- towards the end of the first potential plateau (see Fig. 4). It is
fore, we speculate that this initial phase showing potential also apparent that in the initial two stages of lithiation (alloy
shoulders and a dip is rather related to the reductive decom- formation), the differences between the deposited mass for
position of the electrolyte, i.e., the formation of the SEI and ­LiPF6 and LiTFSI are marginal. Only during the deposition
maybe to the initial nucleation of Li on Au. Zhang et al. of lithium, i.e., after the alloy formation is finished, differ
[31] showed that the SEI formation takes place through two the frequency change rates substantially.
major stages in case of graphite anodes. The first stage takes We carried out lithiation with wire type Au electrodes,
place before lithiation of graphite, during which a loose and too, that also showed the instability of the potential after an
highly resistive film is formed. The second stage occurs extended OCV period. In the delithiation rate, the surface
at a narrow voltage range of 0.25–0.04 V, which proceeds area over volume ratio might play an important role, if the

13
Journal of Solid State Electrochemistry (2021) 25:2849–2859 2857

charge transfer is the rate determining step. In case of the electrolyte. Interestingly, the deposited layer retards the rate
thin films, this is proportional to the inverse of the layer of deposition at lower potentials that suggests that an electri-
thickness, whereas for wires to the inverse of the wire’s cally resistive layer is forming. As we have already seen in
diameter. In our case, there are at least three orders of mag- the course of the galvanostatic measurements, this SEI layer
nitude differences between these two ratios for the thin films also progressively decelerates the rate of Li deposition onto
and wires that reflect in the slower delithiation rate of the Au compared to LiTFSI (see the frequency change curves
wires. It can be concluded that the SEI facilitates the deposi- in Fig. 4a and b). The second cathodic peak corresponds
tion/insertion of Li in general. to the reduction of trace amounts of water in the electro-
During cyclic voltammetric experiments (CVs), the afore- lyte. After this peak, the rate of frequency change becomes
mentioned differences in the SEI forming properties of the smaller. The potential has been reversed at 0.4 V vs. Li to
two electrolytes can be investigated in more detail. In both avoid the underpotential deposition of Li onto Au [23]. Upon
cases, a new gold crystal is taken for the experiments and the potential reversal, the mass increase continues in the anodic
electrodes are polarized with 5 mV·s−1 scan rate from their scan until the OCV potential is reached again and the cur-
OCV to cathodic directions. In the ­LiPF6-based electrolyte, rent remains in the cathodic region at around − 50 nA·cm−2.
first, the frequency increases slightly (Fig. 6a), maybe due to The frequency change is irreversible and finally accounts
the repulsion of the adsorbed ­PF6− anions from the surface to ca. 180 Hz mass increase. Since the measured currents
as the potential decreases. However, the frequency starts to are in the nA range, the charge passed through the electrode
decrease already at around 2.1 V and a sharp cathodic peak is also minimal compared to the large frequency changes.
appears at this potential. This cathodic peak corresponds After converting the frequency changes to mass changes
to the reduction of dissolved oxygen in the electrolyte and using the coefficient determined from calibration, apparent
disappears in the second cycle [32]. The frequency change molar masses are in the range of > 1000 g·mol−1, referring to
is in good agreement with the work of Yamamoto et al. mainly chemical reactions causing the deposit on the surface
[24], who investigated the deposition processes in ­LiClO4 rather than electrochemical.
containing tetraethylammonium perchlorate dissolved in Contrary to ­LiPF6, no oxygen reduction peak can be
propylene carbonate. The authors did not observe any sub- observed in the LiTFSI-based electrolyte (Fig. 6b). The
stantial mass change without the addition of L ­ iClO4 to the absence of this peak could either originate from varying oxy-
solution; therefore, they concluded that Li containing spe- gen levels of the electrolytes caused by the manufacturer or
cies are deposited onto the surface. The rate of frequency differences (i.e., inhibition) in the oxygen reduction mecha-
decrease, i.e., mass increase, is the highest in the follow- nism in the presence of ­TFSI− anions. The overall frequency
ing potential region that refers to the decomposition of the change for this LiTFSI electrolyte is very moderate down

Fig. 6  CV measurements (red a)

curves) with 100-nm-thick
Au EQCM electrodes using
a 1 mol·L−1 ­LiPF6 contain-
ing EC:EMC (3:7) and b
1 mol·L−1 LiTFSI containing
EC:EMC (3:7) with a scan rate
of 5 mV·s−1 and simultaneously
measured frequency change
data (green curves). The lower
potential limit was 0.4 V for
­LiPF6 and 0.15 V for LiTFSI-
based electrolytes b)

13
2858 Journal of Solid State Electrochemistry (2021) 25:2849–2859

to 0.5 V compared to L ­ iPF6. Another difference is that the change until the completion of formation of the stoichiomet-
rate of frequency change never decreases until the potential ric Li-Au alloy is roughly the same for the two electrolytes,
reversal of 0.15 V, which refers to an SEI with good electri- compared to ­LiPF6, the amount of deposited lithium after
cal conductivity. This potential limit was chosen because of 4 h is more than double in LiTFSI-based electrolyte. The
the very steep increase in current, which starts at around ca. calculation of apparent molar mass values reflects a rela-
0.17 V vs Li and is related to the underpotential deposition tively fast buildup of the SEI that is followed by a period in
(UPD) of Li [32] (simultaneously to ongoing SEI forma- which nearly exclusively Li is deposited. The differences of
tion as the frequency change curve shows). Most probably the two types of SEI layers that are already revealed in the
the deposited Li initiates a faster buildup of the SEI. Upon frequency change curves and also reflected during first CV
potential reversal, two anodic peaks appear at around 0.4 V cycles are the most probable reason for this huge difference
and 0.5 V, that are related to the oxidation of Li. Parallel to in the lithiation behavior of the different electrolytes.
these processes, the frequency increases and stabilizes after
the peaks. We speculate that the two peaks are the result Supplementary information The online version contains supplemen-
tary material available at https://d​ oi.o​ rg/1​ 0.1​ 007/s​ 10008-0​ 21-0​ 5060-3.
of oxidation of energetically two different kinds of Li: Li
deposited onto Au and already alloyed with Au. From the Acknowledgements The authors acknowledge I. Szendrő from Micro-
frequency change curve, it seems that the irreversible change Vacuum Ltd. (Budapest, Hungary) for providing us with their new
corresponds to the formed SEI (around 40 Hz) and the rest eQCM-I Mini System for testing for the here described study.
of the reversible change to the electrochemical oxidation of
Li. Compared to L ­ iPF6, it seems that LITFSI forms less SEI Funding Open Access funding enabled and organized by Projekt
DEAL
in the same potential range; however, the SEI formed from
LiTFSI-based electrolyte has beneficial properties towards Availability of data and material All experimental data is available
the deposition of Li onto Au (see also Fig. 4). This might upon request from the authors.
have some implications also for graphite electrodes and on
the intercalation/deintercalation kinetics of Li, which needs, Declarations
however, further studies.
Conflict of interest The authors declare no competing interests.

Conclusions Open Access This article is licensed under a Creative Commons Attri-
bution 4.0 International License, which permits use, sharing, adapta-
tion, distribution and reproduction in any medium or format, as long
We have examined the electrochemical lithiation of gold as you give appropriate credit to the original author(s) and the source,
in two different kind of electrolytes in form of wires and provide a link to the Creative Commons licence, and indicate if changes
thin films with the aim to build stable reference electrodes were made. The images or other third party material in this article are
included in the article's Creative Commons licence, unless indicated
with well-defined mechanical properties. Gold wires after otherwise in a credit line to the material. If material is not included in
72 h of lithiation with a current density of 200 µA·cm−2 the article's Creative Commons licence and your intended use is not
showed excellent electrochemical stability over a prolonged permitted by statutory regulation or exceeds the permitted use, you will
test period and are highly suitable to build PEEK embodied need to obtain permission directly from the copyright holder. To view a
copy of this licence, visit http://​creat​iveco​mmons.​org/​licen​ses/​by/4.​0/.
reference electrodes with variable mechanical properties.
These electrodes also showed outstanding potential stabil-
ity not only over the course of several days but also over
an entire year. Experiments with thin film Au electrodes in
References
form of coated EQCM crystals provided insights into the
effects of varying lithiation times and electrolytes. While 1. Inzelt G, Lewenstam A, Scholz F (2013) Handbook of reference
too short lithiation times will result in unstable potentials, electrodes. Springer, Berlin Heidelberg, Berlin, Heidelberg
also, the use of ­LiPF6 will result in lower potential stability 2. La Mantia F, Wessells CD, Deshazer HD, Cui Y (2013) Reliable
reference electrodes for lithium-ion batteries. Electrochem Com-
compared to LiTFSI-based electrolytes. The EQCM meas-
mun 31:141–144
urements also revealed differences in the kinetics of lithia- 3. Costard J, Ender M, Weiss M, Ivers-Tiffée E (2016) Three-
tion of thin film Au compared to wired type Au electrodes. electrode setups for lithium-ion batteries. J Electrochem Soc
Well-pronounced potential plateaus appeared on the chro- 164:A80–A87
4. Waldmann T, Hogg B-I, Kasper M, Grolleau S, Couceiro CG,
nopotentiometric curves, and the simultaneously measured
Trad K, Matadi BP, Wohlfahrt-Mehrens M (2016) Interplay of
frequency change curves also show the changing lithiation operational parameters on lithium deposition in lithium-ion cells:
kinetics towards the ends of the potential plateaus. These systematic measurements with reconstructed 3-electrode pouch
frequency change curves could also explain the large differ- full cells. J Electrochem Soc 163:A1232–A1238
ences between the two electrolytes. Whereas the total mass

13
Journal of Solid State Electrochemistry (2021) 25:2849–2859 2859

5. Barsoukov E (1999) Kinetics of lithium intercalation into carbon model anodes for lithium ion batteries: electrochemical charac-
anodes: in situ impedance investigation of thickness and potential terization. Electrochim Acta 164:81–89
dependence. Solid State Ionics 116:249–261 21. Bach P, Valencia-Jaime I, Rütt U, Gutowski O, Romero AH, Renner
6. Ender M, Weber A, Ellen I-T (2011) Analysis of three-electrode FU (2016) Electrochemical lithiation cycles of gold anodes observed
setups for AC-impedance measurements on lithium-ion cells by by in situ high-energy X-ray diffraction. Chem Mater 28:2941–2948
FEM simulations. J Electrochem Soc 159:A128–A136 22. Inzelt G (2010) Electrochemical Quartz Crystal Nanobalance,
7. Dees DW, Jansen AN, Abraham DP (2007) Theoretical examina- Chapter II.10, pp. 257-269 In F. Scholz (ed) Electroanalyti-
tion of reference electrodes for lithium-ion cells. J Power Sources cal methods: guide to experiments and applications, 2nd edn.
174:1001–1006 Springer, Berlin Heidelberg, Berlin, Heidelberg
8. Wandt J, Lee J, Arrigan DWM, Silvester DS (2018) A lithium 23. Aurbach D, Zaban A (1995) The use of EQCM for the study
iron phosphate reference electrode for ionic liquid electrolytes. of nonactive metal electrodes in propylene carbonate - LiAsF6
Electrochem Commun 93:148–151 solutions: significance of the data obtained. J Electrochem Soc
9. Gómez-Cámer JL, Novák P (2013) Electrochemical impedance 142:L108–L111
spectroscopy: Understanding the role of the reference electrode. 24. Yamamoto N, Hirasawa H, Ishida H, Tatsuma T, Oyama N (1994)
Electrochem Commun 34:208–210 Electrochemical deposition and dissolution processes of lithium
10. Verbrugge MW, Baker DR, Koch BJ (2002) Mathematical mod- compound on gold from propylene carbonate. Bull Chem Soc Jpn
eling of high-power-density insertion electrodes for lithium ion 67:1296–1300
batteries. J Power Sources 110:295–309 25. Koike S, Fujieda T, Wakabayashi N, Higuchi S (1997) Electro-
11. Zhou J, Notten PHL (2004) Development of reliable lithium chemical and quartz microbalance technique studies of anode
microreference electrodes for long-term in situ studies of lithium- material for secondary lithium batteries. J Power Sources
based battery systems. J Electrochem Soc 151:A2173–A2179 68:480–482
12. Abraham DP, Poppen SD, Jansen AN, Liu J, Dees DW (2004) 26. Shpigel N, Levi MD, Sigalov S, Daikhin L, Aurbach D (2018)
Application of a lithium–tin reference electrode to determine elec- In situ real-time mechanical and morphological characterization
trode contributions to impedance rise in high-power lithium-ion of electrodes for electrochemical energy storage and conversion
cells. Electrochim Acta 49:4763–4775 by electrochemical quartz crystal microbalance with dissipation
13. Solchenbach S, Pritzl D, Kong EJY, Landesfeind J, Gasteiger monitoring. Acc Chem Res 51:69–79
HA (2016) A gold micro-reference electrode for impedance and 27. Zhu D, Fan H, Wang H (2021) PF6–intercalation into graphite
potential measurements in lithium ion batteries. J Electrochem electrode from propylene carbonate. ACS Appl Energy Mater
Soc 163:A2265–A2272 4:2181–2189
14. Simon FJ, Blume L, Hanauer M, Sauter U, Janek J (2018) Devel- 28. Gritzner G, Kuta J (1984) Recommendations on reporting elec-
opment of a wire reference electrode for lithium all-solid-state trode potentials in nonaqueous solvents (Recommendations 1983).
batteries with polymer electrolyte: FEM simulation and experi- Pure Appl Chem 56:461–466
ment. J Electrochem Soc 165:A1363–A1371 29. Laoire CO, Plichta E, Hendrickson M, Mukerjee S, Abraham
15. Aurbach D, Moshkovich M, Cohen Y, Schechter A (1999) The KM (2009) Electrochemical studies of ferrocene in a lithium
study of surface film formation on noble-metal electrodes in alkyl ion conducting organic carbonate electrolyte. Electrochim Acta
carbonates/li salt solutions, using simultaneous in situ AFM, 54:6560–6564
EQCM, FTIR, and EIS. Langmuir 15:2947–2960 30. Aurbach D (2000) Review of selected electrode–solution interac-
16. Dey AN (1971) Electrochemical Alloying of Lithium in Organic tions which determine the performance of Li and Li ion batteries.
Electrolytes. J Electrochem Soc 118:1547 J Power Sources 89:206–218
17. Ranninger J, Wachs SJ, Möller J, Mayrhofer KJJ, Berkes BB 31. Zhang S, Ding MS, Xu K, Allen J, Jow TR (2001) Understanding
(2020) On-line monitoring of dissolution processes in nonaqueous solid electrolyte interface film formation on graphite electrodes.
electrolytes — a case study with platinum. Electrochem Commun Electrochem Solid-State Lett 4:A206–A208
114:106702 32. Aurbach D, Daroux M, Faguy P, Yeager E (1991) The electro-
18. Ranninger J, Mayrhofer KJJ, Berkes BB (2021) The crucial role of chemistry of noble metal electrodes in aprotic organic solvents
water in the stability and electrocatalytic activity of Pt electrodes. containing lithium salts. J Electroanal Chem Interf Electrochem
J Phys Chem C 125:13254–13263 297:225–244
19. Wachs SJ, Behling C, Ranninger J, Moeller J, Mayrhofer KJJ,
Berkes BB (2021) Online monitoring of the transition metal dis- Publisher's Note Springer Nature remains neutral with regard to
solution from a high Ni content cathode material. Mater Interfaces jurisdictional claims in published maps and institutional affiliations.
in press, accepted, ACS Appl
20. Bach P, Stratmann M, Valencia-Jaime I, Romero AH, Renner FU
(2015) Lithiation and delithiation mechanisms of gold thin film

13