Characterisation of matrix-based polyatomic interference formation in laser ablation-inductively coupled plasma-mass spectrometry using dried micro-droplet ablation and its relevance for bioimaging

Dried micro-droplets were used to characterise the formation of polyatomic interferences in a commercial laser ablation-inductively coupled plasma-mass spectrometer (LA-ICP-MS). Droplets containing 4 ng each of Al, As, Ca, Cd, Co, Cu, Fe, Mg, Mn, P, Se and Zn in the presence of potentially interfering isotopes were deposited on silicate microscope slides. Comparisons of the total signal recorded for each dried droplet showed no detectable influence of polyatomic matrix-based interferences. Visualisation of acquired signal was achieved by the generation of two-dimensional maps. The natural abundance pattern of elements with two measurable isotopes was confirmed for each droplet. The lack of interferences was due to the absence of major gas molecules (e.g. N2, O2, CO2) from the laser cell and minimal total matrix load on the plasma when compared to standard solution nebulisation (approx. ng s 1 versus 10– 20 mL s ) ICP-MS conditions.


Introduction
Laser ablation (LA) has greatly expanded the capability of conventional ICP-MS instruments, allowing for rapid analysis of trace elements via direct solid sampling. 1In the biological sciences, LA-ICP-MS has also revolutionised spatial measurement and mapping of total metal distributions, providing an accessible alternative to synchrotron-based microprobes with more than adequate sensitivity for most biologically-relevant metals. 2 It is oen assumed that LA-ICP-MS suffers from signicant isobaric interferences from polyatomic species 3 in line with traditional solution nebulisation (SN) ICP-MS.While it has been suggested that interferences, particularly those arising from water in the sample matrix, are considerably lower in LA-ICP-MS, 4,5 reduced interference formation in the context of bioimaging has not been conrmed.ICP-MS instruments with quadrupole mass lters typically have mass resolution of 0.1 atomic mass units.Therefore, polyatomic species with the same nominal mass as the ion of interest cause signicant interfering signal that precludes accurate measurement.For SN-ICP-MS either a higher resolution mass lter (such as a sector eld design 6 ) or collision/reaction gas 7,8 is required to negate the effect of, or physically remove interferences.Sector eld instruments command a considerably higher price than quadrupole instruments with minimal or even no increase in sensitivity and a more limited range of sequentially scanned isotopes. 9Reaction/collision cell equipped quadrupole ICP-MS exhibit a signicant drop in signal abundance (oen by a factor of 10 or more), which is unfavourable when introducing comparatively smaller amounts of material to the plasma via LA.
Ablated analytes transported to the ICP-MS reach the plasma in a very different state when compared to SN.Most signicantly, solvent-free particles of a much smaller diameter 10 are produced by commercially available UV lasers.The ablation process involves desorption, vapourisation and ionisation of the analyte and subsequent condensation and aggregation from the laser plasma to form laser aerosols. 11,12Therefore, it is reasonable to assume that applying operational parameters similar to solution nebulisation is unnecessary in LA-ICP-MS to achieve adequate determination.
MO + , MN + and MH + based interferences primarily arise from air (N 2 , O 2 etc.) and the solvent in use (mainly H 2 O based matrices, along with Cl + and N + species in certain acidic matrices).These highly abundant ions are generated within the plasma, 13 and are somewhat controlled in solution nebulisation ICP-MS by manipulation of parameters that inuence plasma temperature such as carrier gas ow rate, radio frequency (RF) power; or residence time such as sampling depth, and to a lesser extent sampling cone orice dimesions. 14The degree to which these parameters can be controlled is limited, as adequate desolvation and ionisation is dependent on the duration the particle resides within the plasma and the applied power.The 'dry' nature of the plasma in LA-ICP-MS is advantageous, as the confounding effects of H 2 O and N on polyatomic interference from the typical acidic matrices used in solution nebulisation ICP-MS are signicantly reduced.Reducing the potential for interference formation can sufficiently negate the need for post-plasma treatment, such as collision gases or high resolution mass lters.
Dried droplet ablation has been suggested as an alternative method to electrothermal vapourisation (ETV) ICP-MS for the discrete sampling of small volume aqueous samples. 15Günther et al. 16 showed that direct ablation of dried solutions yielded no signicant elemental fractionation.Dried droplet analysis was also shown to be suitable for the simultaneous determination of 21 isotopes with absolute detection limits in the sub ng g À1 range. 5Volumes as low as 20 mL have been used to quantify mg g À1 concentrations of selenomethionine in yeast via droplet analysis of HPLC fractions. 17It has also been shown that dried droplet analysis yielded a 100-fold decrease in hydride formation of radionucliotides when compared to both solution nebulisation and ETV-ICP-MS. 18Ablation of dried blood droplets on lter paper has been suggested as a suitable technique for the measurement of Pb toxicity. 19Recently, 5-20 mL droplets of acidied standard solutions placed on lter paper supports has been successfully used for multi-point external calibration in imaging experiments. 20he object of this study was to characterise interference formation in a commercial LA-ICP-MS instrument using dried sub-mL volume droplets.The experimental design was based on the theoretical formation of polyatomic isobaric interferences to determine the inuence of matrix components on true analyte signal.

Instrumentation
A New Wave UP-213 laser ablation unit (Kenelec Scientic, Mitcham, Victoria, Australia) with a Nd:YAG laser emitting a 213 nm laser beam was used.The laser was hyphenated to an Agilent Technologies 7500ce ICP-MS (Mulgrave, Victoria, Australia) system tted with 'cs' lenses for enhanced sensitivity.High purity liquid argon (Ace Cryogenic, Castle Hill, NSW, Australia) was used for all analyses.The system was tuned for sensitivity prior to all analyses using NIST 612 glass, with 7 Li, 59 Co, 89 Y, 238 U and responses monitored.Oxide formation was monitored as the 232 Th + / 232 Th 16 O + ratio and was <0.3% for all experiments.High laser energy uence can introduce contaminates from the underlying glass matrix, depending on the glass composition. 21n average laser uence of 0.15 J cm 2 and repetition rate of 20 Hz was used, which is well below the reported threshold for ablation of soda-lime glass (the major matrix of the support slides used; see below) for UV lasers of approximately 0.80-0.86J cm À2 . 22For other support matrices, uence should be selected based on the appropriate threshold of ablation for the material in question.Comparisons of the gas blank and ablated areas of the support slide only showed no discernible increase in Si and Na (the major elemental components of glass) signal (Fig. 1).
Reagents and solutions 10 AE 0.2 mg L À1 solutions of each element shown in Table 1 were used (Choice Analytical, Thornleigh, NSW, Australia).Solutions were made to mass using 1% Seastar Baseline grade HNO 3 (Choice Analytical) with approximately 10 mg mL À1 amido black added to aide in visualisation of dried droplets.
Deposition of droplets 0.4 mL of each solution (amounting to 4 ng of each element) was dropped directly onto high-purity KNITTEL StarFrost soda-lime glass microscope slides (ProSciTech, Townsville, Queensland, Australia) using a Gilson air displacement pipette.The solution was expelled to a point where a droplet formed on the pipette tip, and was then placed onto the slide surface.The slide was then allowed to dry in an enclosed desiccator prior to analysis.
Five solutions were prepared (see Table 1).These consisted of a blank solution containing only amido black and solutions containing potential interfering elements (a list of theoretical polyatomic interferences are listed in Table 2), analyte elements and a mixture of all elements used (Fig. 1a).The amido black was added to locate the dried spots using the CCD camera on the laser unit.The amido black was also a source of carbon to evaluate potential 12 C and 13 C-based interferences.Nitrate from the 1% HNO 3 matrix remaining on the slide surface aer drying provided a potential source of 14 N-based interferences.Approximate total amounts of N, O and C from the nitrate anion and 10 mg mL À1 amido black dye are 124 ng, 429 ng and 1.72 pg, respectively.

Data processing
Each line of ablation in the x-axis produced a single data le in comma separated value (.csv) format.Using a Visual Basic macro in Microso Excel, 23 data was combined into a single ASCII le for importation into ENVI (Exelis Visual Information Solutions, Boulder, Colorado, USA).Droplet images were standardised to 13 C from the amido black according to methods previously described. 24Raw count data was then extracted from ENVI by selecting the area of the droplet and exporting as a rich text format le.Mean background signal for each m/z was subtracted and the sum of raw counts for each droplet was tabulated and compared.Five replicate measurements of experiment were carried out.Composite images shown in Fig. 1b and c were produced in ImageJ, with Pearson correlation coefficients calculated using the 'Intensity Correlation Analysis' plugin. 25

Results and discussion
The optimised plasma operating conditions are summarised in Table 3.In typical solution ICP-MS measurement using the 7500ce system an RF power of 1500 W is applied to the plasma.Additionally, solution ICP-MS requires a sample depth of 6-8 mm to allow adequate ionisation of material.Lower plasma power (1200 W) results in less ionising potential of the plasma, with fewer energetic electrons available for ionising analyte.Additionally, the short residence time in the plasma due to decreased sample depth gives less opportunity for polyatomic species to form. 27 5 Â 30 mm area encompassing all deposited droplets was ablated to ensure all material was transported to the ICP.Analysis time was approximately four hours per sample set.Images for m/z 56 and 57 are shown in Fig. 1b.Correlation analysis of m/z 56 and 57 images showed a Pearson correlation coefficient of 0.979, indicating no additional signal above the background was recorded from one of the most prevalent polyatomic interference normally encountered, 40 Ar 16 O + on 56 Fe + .A background signal was present for m/z 56 when the laser was not operational, produced from trace O 2 impurities present in the argon gas.Furthermore, this background was   25 Mg + .Droplet 'Int' contained 10 mg kg À1 of both 31 P and 25 Mg, and droplet A1 contained 10 mg kg À1 40 Ca.No signal was recorded for m/z 63 or 65 in these ablated droplets.This indicated that Ar-or O-based polyatomic interferences were below the instrument's limit of detection, and the Pearson's correlation coefficient was 0.954 for the resulting images.
Further investigation of the inuence of polyatomic interferences was carried out by examining the abundance ratios of elements with two measurable isotopes (Table 4).Close correlation between each element's natural isotopic abundance and measured uncorrected abundance ratios were conrmed.While the sequential mass analyser in a quadrupole ICP-MS is more prone to intrinsic uncertainty due to the mode of detection when compared to multicollector designs as heavier isotopes are more efficiently transmitted and detected, 28 our data was either within or close to the measured uncertainty expected for a quadrupole analyser.This comparison clearly demonstrated polyatomic interferences on the abundance ratio were negligible.
These data support the use of continued use of LA-ICP-MS for interference-free imaging applications in the biological sciences, showing that the most commonly expected matrixbased species arising from highly abundant elements such as Na, K, P and O are absent using typical LA-ICP-MS operating procedures, in contrast to that which was previously claimed but not experimentally demonstrated. 29We have previously discussed the utility of reaction gases, such as hydrogen (in an octopole-equipped ICP-MS) 30 and oxygen (for mass-shiing using triple quadrupole-design ICP-MS systems), 31 though in both cases gases were used for either improving signal to noise ratios for extremely small sample volumes or low-abundance elements such as Se, respectively.

Conclusions
Ablation of micro-volume dried droplets did not produce a detectable amount of interfering polyatomic species.Dry plasma conditions, lower plasma power and shorter sample depth was adequate for limiting the formation of matrix-based and spectral interferences.Natural isotopic abundance patterns were conrmed.The suitability of LA-ICP-MS for detection of sub-nanogram amounts of material was also demonstrated.

Fig. 1
Fig. 1 (a) Schematic representation of imaging dried droplets on a standard microscope slide by LA-ICP-MS.(b and c) Relative intensity maps of iron (m/z 56, 57) and copper (m/z ¼ 63, 65) in deposited droplets with corresponding composite images, showing negligible interferences on any measured mass due to theoretical polyatomic species (see Table 2).The area left of the dashed white line indicates the gas blank signal, showing no discernible increase in signal arising from potential 40 Ar 16 O + or 23 Na 40 Ar + interferences on 56 Fe or 63 Cu, respectively.
Fig. 1 (a) Schematic representation of imaging dried droplets on a standard microscope slide by LA-ICP-MS.(b and c) Relative intensity maps of iron (m/z 56, 57) and copper (m/z ¼ 63, 65) in deposited droplets with corresponding composite images, showing negligible interferences on any measured mass due to theoretical polyatomic species (see Table 2).The area left of the dashed white line indicates the gas blank signal, showing no discernible increase in signal arising from potential 40 Ar 16 O + or 23 Na 40 Ar + interferences on 56 Fe or 63 Cu, respectively.

Table 1
Composition of interference (Int) and analyte-containing (A1, A2 and mix) solutions a All solutions contained 10 mg mL À1 amido black.

Table 2
Theoretical major polyatomic interferences.Adapted from May et al. 26 Ar 35 Cl + , 59 Co 16 O + , 36 Ar 39 K + , 43 Ca 16 O 2 + , 23 Na 12 C 40 Ar + stable throughout the entire experiment, indicating that the formation of matrix-based 40 Ar 16 O + was negligible.Images for m/z 63 and 65, corresponding to the two natural isotopes of Cu are shown in Fig. 1c. 63Cu + was potentially interfered by 40 Ar 23 Na + , 40 Ca 23 Na + and 31 P 16 O 2 + , and 65 Cu + by 40 Ar

Table 4
Isotopic abundance patterns for analysed elements with multiple isotopes

Table 3
LA and ICP-MS operating conditions