Measurement of 18O tracer diffusion coefficients in thin yttria stabilized zirconia films.

In this paper we present a method to measure oxygen tracer diffusion coefficients in thin ion conducting films without being limited by slow oxygen incorporation kinetics. The method is based on a two step process. In the first step a substantial amount of 18O tracer is locally incorporated for example into an yttria stabilized zirconia (YSZ) layer at low temperatures with the aid of an electric current, thus overcoming slow thermal oxygen exchange while still limiting lateral diffusion to a minimum. In the second step controlled diffusion takes place at elevated temperatures in ultra high vacuum (UHV) to impede loss of tracer due to oxygen exchange at the film surface. In this second step the surface of the thin film may additionally be modified compared to the oxygen incorporation step. This allows to easily investigate effects of interfaces on ion transport. The achieved in-plane concentration profiles are then measured by secondary ion mass spectrometry (SIMS). Comparison with electrical measurements on YSZ thin films proves the applicability of the method.

Introduction

Ion conduction in thin films or heterolayers may strongly differ from the conduction known for bulk materials and highly interesting effects including strongly enhanced effective ionic conductivities are reported [1], [2], [3], [4], [5]. Among others, there have been reports of enhanced conductivity in thin yttria stabilized zirconia (YSZ) layers of up to four orders of magnitude [6] and even eight orders of magnitude in YSZ/SrTiO3 heterolayers [7]. However, especially in the latter case it has been discussed controversially [8], [9] if this increase was caused by electronic conduction. A method to rule out electronic contributions to the measured transport property is 18O tracer diffusion, but, to the best of our knowledge, tracer diffusion experiments in YSZ thin films are not available yet. Hence a drastically enhanced ionic conductivity along the interfaces between YSZ films and the substrate or the top layer (gas) cannot be excluded.

Oxygen tracer studies on YSZ thin films are expected to be far from trivial and raise several experimental problems: Among others this is the low surface exchange coefficient k* of YSZ for oxygen [10] which usually hinders preparation of depth profiles perpendicular to the thin film plane. This low k* value also constitutes a challenge for in-plane diffusion experiments since rather high temperatures are required for incorporation while the oxygen tracer has to be kept in the film during diffusion, for example by a protection layer. (For successful studies on high k* materials see Refs. [11] and [12]). Recently it was suggested that electrical pumping of 18O tracer, rather than relying on thermal oxygen exchange, can drastically enhance effective oxygen tracer exchange coefficients [13]. In such a manner, experiments with very shallow profiles and still acceptably high tracer concentrations can be conducted in fast oxide ion conductors. It shall be noted, that electric fields combined with elevated temperatures can lead to a correlation of microdomains in Ca-doped zirconia, hence altering the diffusion properties. No such effect was found for YSZ [14].

In this contribution we modified this field-assisted tracer incorporation approach such, that by a two step process it becomes possible to gain high tracer concentrations in YSZ thin films while keeping temperatures for diffusion relatively low. This method also allows a high flexibility in choosing and modifying interfaces during tracer diffusion.

Experimental

A YSZ (8-YSZ, Tosoh, Japan) layer was manufactured on a sapphire single crystal (0001) (Crystec, Germany) using pulsed laser deposition (PLD) and a substrate temperature of 600 °C. X-ray diffraction data showed, that the layer was fully crystalline and oriented in the cubic (111) direction. The thickness was determined to be 145 nm by sputtering down to the YSZ/sapphire interface via TOF-SIMS (TOF-SIMS 5, IONTOF, Germany) and measuring the resulting crater depth with digital holographic microscopy (DHM R1000, Lyncée Tec, Switzerland).

For current-driven oxygen tracer incorporation into the YSZ film a mixed conducting La0.6Co0.4CoO3 − x (LSC) layer of ca. 40 nm thickness was deposited on the YSZ film by PLD at a substrate temperature of 550 °C. A 5 μm broad gap was etched into the LSC layer using photolithography and hydrochloric acid thus yielding two electrodes for the electrical oxygen pumping experiments. LSC electrodes were chosen due to their low polarization resistance towards electrochemical oxygen exchange [15]. Electrical 18O incorporation into YSZ was performed at 305 °C in 221 mbar 18O2 atmosphere (98% 18O2, Cambridge Isotope Laboratories, GB). For the actual diffusion experiment a part of the YSZ film was covered by a 200 nm thick Pt film sputter-deposited after removal of the LSC electrodes. Tracer concentration analysis before and after diffusion was performed by TOF-SIMS. More experimental details including the results are given in Section 4.2.

For electrical characterization of the films by impedance spectroscopy (Alpha-A High Performance Frequency Analyzer, Novocontrol, Germany) 400 nm thick gold electrodes with a thin chromium adhesion layer were produced using lift-off photolithography. The electrodes were patterned as interlocking digits with a distance of 25 μm. Impedance data were measured between 150 and 550 °C.

Methodology: the two step diffusion experiment

The following two step method was employed to investigate oxygen tracer diffusion in the YSZ films: i) 18O tracer incorporation takes place at one electrode in 18O2 atmosphere under a continuous current flow between the two electrodes. Provided mixed conducting electrodes with low polarization resistance are used, this electrochemical pumping should lead to a narrow region of high oxygen tracer concentration in the gap area close to the cathode. The reason for the limitation of the tracer injection to the edge region of the mixed conducting cathode is the very high lateral ohmic resistance of the film compared to the low oxygen incorporation resistance of the cathode. After tracer incorporation, the mixed conducting electrodes (here LSC) are removed chemically and the resulting 18O profile is investigated by TOF-SIMS. ii) The main diffusion experiment is then carried out in UHV at a temperature being higher than that used during oxygen injection (step one). The final 18O profile is measured after quenching the sample. Since the second step is performed in UHV 18O stays in the film during diffusion and a protection layer on YSZ is not needed. This gives a large flexibility in choosing or modifying the surface of the YSZ layer for the actual diffusion experiment.

Results and discussion

Electrical measurements

The impedance measurements using small-spaced interdigit electrodes allowed us to not only analyze the total in-plane conductivity of the YSZ layers but also to separate grain bulk and grain boundary resistances. Details on this data evaluation will be given in Ref. [16]. In Fig. 1 the bulk conductivities and the total conductivities of the thin film are compared to those determined from a conventional macroscopic impedance measurement on a YSZ polycrystal of the same composition. The bulk conductivity of the macroscopic measurement and the investigated layer match quite well, while the total effective conductivity of the layer is about a factor of 2 to 5 lower. This can be attributed to impeded oxygen transport across the grain boundaries in the thin film. Via the Nernst–Einstein equation [10], [17]the conductivity diffusion coefficient D can be calculated from the conductivity σ. In Eq. (1) k denotes the Boltzmann constant and T is the absolute temperature; c refers to the volume concentration of the diffusing species, z to the charge number and e to the elementary charge. For 500 °C D has been calculated from the measured total conductivity of the YSZ film and the result is given in Table 1.

Fig. 1

Arrhenius plot of the bulk and total conductivities of the YSZ layer compared to macroscopic polycrystal bulk values.

Table 1

Diffusion coefficients and Haven ratios measured by tracer diffusion (D ) and impedance spectroscopy (D) at 500 °C.

	D⁎ [cm2s−1] at 500 °C	2 h [μm]	HR
Uncovered YSZ	9.0 · 10−10	15.8	1.4
Platinum covered YSZ	6.4 · 10−10	18.5	1.0
	Dσ [cm2s−1] at 500 °C
Conductivity measurement	6.6 · 10−10	–	–

Tracer injection and diffusion experiment

The sample was kept for 10 min at 305 °C in an atmosphere of 221 mbar 18O2 to allow the fast oxygen exchanging LSC electrodes to equilibrate. Then a direct current of 500 nA was applied to the electrodes for 20 min in order to locally pump 18O into YSZ (see sketch in Fig. 2a). Afterwards the sample was cooled to room temperature which completes the first step, which is also sketched in Fig. 2a. The ion images recorded on such a film by TOF-SIMS (Fig. 2b and c) clearly display that a confined area of the YSZ layer has been enriched with the oxygen tracer. The top view image in Fig. 2b shows that the area is not of perfect rectangular shape, most likely due to non-ideal electrode edges. However, this roughness of about 2.5 μm is small compared to the width of the final diffusion zone in step two and of no relevance in this proof of concept. The cross-sectional view of the profile (Fig. 2c) illustrates that in this 18O “rectangle” the entire depth of the film (y-direction) is filled by a virtually constant 18O concentration. The resulting lateral concentration profile integrated over the entire z-direction of Fig. 2b is plotted in Fig. 2d. The half width of the profile amounts to 13 μm, while the top plateau is 8 μm broad. The plateau concentration corresponds to about 28% 18O with respect to the total oxygen content. The width of the initial profile is somewhat surprising, since enrichment was only expected near the three phase boundary of the cathodically polarized LSC electrode (see also Section 3). It is also larger than the nominal gap between the LSC electrodes. This broadening might be caused by some local reduction of YSZ due to the high electric field but further investigations are necessary for final clarification. On the other hand, the almost rectangular shape of the resulting profile even simplified the analysis of the data obtained in the following second step of the experiment.

Fig. 2

a) Sketch of the set-up for current driven tracer injection. The question marks indicate that the exact implantation area is yet unknown; b) TOF-SIMS ion image in top view and c) in a cross-sectional view; d) 18O concentration profile in YSZ after the electrical tracer injection calculated from the intensities I of 16O and 18O isotopes.

After removing the LSC electrodes and depositing a dense platinum layer perpendicular to the tracer enriched area (see Fig. 3a), the main diffusion step was carried out in the UHV of the TOF-SIMS instrument. The sample was placed on a heating stage, heated to 500 °C, held at this temperature for 90 min and then cooled down to room temperature. Mass spectra were recorded both underneath the platinum layer and on the uncovered YSZ surface. The ion images in Fig. 3b and c illustrate, how the initial box-shaped profile has broadened during the heat treatment. Obviously, the two step tracer diffusion experiment was performed successfully.

Fig. 3

a) Sketch of the tracer diffusion experiment in UHV; b) TOF-SIMS ion image in top view and c) in the cross-sectional view for the measurement underneath Pt; d) 18O concentration profiles after diffusion at 500 °C for 90 min for Pt-covered and uncovered YSZ calculated from the intensities I of 16O and 18O isotopes.

For the special case of diffusion from a rectangular initial distribution an analytical solution of Fick's second law of diffusion exists [18] and the corresponding equation reads

C is the concentration as a function of the distance x and time t, h denotes half the width of the initial distribution and C represents the initial concentration at t = 0 for |x| ≤ h. C is introduced due to the background concentration of tracer ions.

Owing to the similarity of the measured initial profile and a rectangular initial profile, the concentration profiles in Fig. 3d were fitted to Eq. (2) with the diffusion time being set to 90 min and the initial concentration to 28% 18O. The background concentration was measured to be 0.3%, which is slightly higher than the natural abundance of 18O of 0.2%. This minor difference is most likely caused by the temperature driven 18O exchange along the entire LSC/YSZ interface before and during the electrochemical injection step. Parameters h and D were free for optimization. The results are summarized in Table 1 and compared to the results of the electric measurement.

The initial extensions of the tracer-enriched box 2 h resulting from the fit (15.8 and 18.5 μm) are somewhat larger than the half width of 13 μm measured before the diffusion experiment. However, since some broadening during heating to the actual diffusion temperature may play a role and a certain roughness or width variation of the initial profile can affect the results, this is a reasonable agreement. Comparing the profiles and the diffusion coefficients with (6.4 · 10−10 cm2/s) and without Pt top layer (9.0 · 10−10 cm2/s) it seems that the free surface slightly accelerates the effective oxygen diffusion. While this may be a true effect, the absolute difference is too small to make a final statement so far. Previously published tracer diffusion coefficients on bulk materials are in the same range. For example in Refs. [10] and [19] YSZ single crystals with 9.5 mol% yttria (9.5-YSZ) were investigated and both groups report a D(500 °C) of about 5 · 10−10 cm2/s, while in a similar study on 10-YSZ single crystals a D(500 °C) of about 9 · 10−10 cm2/s was measured [20]. Results for 8-YSZ can be found in Ref. [21], where a D(500 °C) of 2.7 · 10−10 cm2/s is reported for polycrystals with a grain size of about 50 nm. All these values are relatively close to each other and agree well with the results in our present study.

With the knowledge of both D and D the Haven ratio H [10], [17]can be calculated and is given in Table 1 for the cases of Pt-covered and uncovered YSZ. The values for H amount to 1.4 for uncovered YSZ and 1.0 for the Pt-covered layer, which is higher than the tracer correlation factor of a primitive cubic anion lattice (0.65 [17]). Previously published results in Ref. [10] on 9.5-YSZ single crystals vary between H = 0.33 for temperatures below 650 °C and 0.48 for higher temperatures; at 650 °C H = 1.1 is reported, which the authors attribute to a possible order–disorder transition. In Ref. [20] the temperature dependence of the Haven ratio is studied for 10-YSZ single crystals. At 500 °C they measure H = 1.1, which fits reasonably well to the results in the present study. However, an additional error from temperature differences between conductivity and diffusion experiments, which run on different heating stages cannot be excluded and might adulterate the Haven ratios measured in here. Further studies are in progress, but already these data show that our two step method suggested for measuring 18O tracer diffusion in thin oxide layers indeed works.

Conclusion

The method presented in this paper allows an easy and flexible determination of oxygen tracer diffusion coefficients in thin oxide layers. With current driven tracer injection it was possible to produce a well defined tracer profile before the actual diffusion step, which greatly simplifies the diffusion experiment and data analysis. The measured tracer diffusion coefficients correlate well with conductivity based values. This two step process also allows a high flexibility regarding the top interface of thin oxide films during the diffusion experiment. First indication of modified ion transport properties of the surface layer is given. With this approach it might thus be possible, to unambiguously clarify if high ionic conduction is indeed present in thin oxide films and how it is affected by interfaces such as SrTiO3/YSZ.