1362
Anal. Chem. 1984, 56, 1362-1370
Embedded Ion Exchange Beads as Standards for Laser Microprobe Mass Analysis of Biological Specimens Armand H. Verbueken and Ren6 E. Van Grieken* Department of Chemistry, University of Antwerp (U.I.A.), B-2610 Wilrijk, Belgium
Guido J. Paulus Department of Pathology, University of Antwerp (U.I.A.), B-2610 Wilrijk, Belgium
Wim C. de Bruijn Centre for Analytical Electron Microscopy, Laboratory for Electron Microscopy, 2333 AA Leiden, The Netherlands
A new Internal standard system based upon Chelex-100 ion Chelating resin beads loaded with varlous elements is proposed for laser microprobe mass analysis (LAMMA) of tlssue-bound elements In thln sections of blologlcai materlai. The beads can be coembedded with the blologlcal specimen and cut to the desired thlckness. I n thls way, the standard materlai is avallable together wlth and close to the tissue In the same sectlon, thus avolding section thickness measurements and speclmen exchange. Chelex-100 beads can be loaded wlth various elements In a controlled way and the spatial elemental dlstrlbution is sufficlentiy homogeneousfor use In mlcroanalysls. The slgnal reproduclbillty, the proportlonallty between the LAMMA response and the elemental concentration, and the influence of the specimen composltlon on the LAMMA mass spectra were studied. Cheiex-100 was also compared to other types of ion exchange beads. Thls versatlie standardlzatlon technique opens new vistas for quantltatlve laser microprobe appllcatlons In bloiogical and blomedlcai research.
Considerable interest is currently being focused on laser microprobe mass analysis (LAMMA), primarily due to the commercial advent of the LAMMA-500 instrument. A description of this new microanalytical instrument and discussions of its advantages, drawbacks, and applications can readily be found in the literature (1-3). The technique was originally developed for biomedical purposes, but hitherto its impact remained mainly limited to qualitative biological aspects. Absolute quantification of subcellular elemental distributions in biological specimens is, of course, not trivial at all. Structural localization in the uncolored and unstained histological sections is often problematic. Moreover, the LAMMA spectra of biological preparations can be highly complex, variable, and difficult to interpret: mass interferences arising from organic molecules or fragment ions, both from the specimen itself and from the embedding or supporting media, are often important, while the laser interaction is intricate and complete understanding about laser-induced microplasma ion production is lacking. Finally, suitable calibration standards are necessary, not only to be applied directly for biomedical samples but also to be used as simple model systems to study the general analytical features of the laser microprobe. Only a few calibration experiments for real biological and biomedical samples have been carried out up to now (4-7). In principle, many of the standards proposed and used in conventional X-ray microanalysis (8-1 0) can be considered for LAMMA. Ideal elemental calibration standards should fulfill the following requirements: (i) well-defined chemical
composition and structure, (ii) similar matrix composition as the specimen, (iii) homogeneous at the level of spatial resolution, (iv) controllable elemental doping possible, (v) a wide variety of elemental dopants possible, (vi) elemental concentrations adjustable within a range typical for biological specimens, (vii) elemental concentrations assessable by several analytical techniques, (viii) inert with respect to surrounding material, (ix) resin (co-)embedding and thin sectioning possible. In LAMMA microanalysis, additional specific features should be similar for standards and specimens: (i) section thickness, (ii) matrix mass density, (iii) optical characteristics and surface texture, (iv) instrumental conditions. Standards made by dissolving organic compounds in the resin used for sample embedding (7-9,11-13) and thin films of vacuum deposited metals or dielectric materials ( 7 , 8 )are most widespread. Another interesting possibility for LAMMA calibration could be the use of anionic surfactant films loaded with metal cations, as proposed by Gijbels et al. (14). Proteinaceous standards consisting of an organic matrix like gelatin (15)or albumin doped with a variety of elements have been investigated, mostly as thin cryosections, for electron probe X-ray microanalysis (16-ZO), secondary ion mass spectrometry (21, 22), and LAMMA (23, 24). Following the original idea that introduced the use of ion exchange beads as chemical microstandards (25),de Bruijn (10,26) recently investigated the characteristics and possibilities of this material for application in X-ray microanalysis. Ion exchange resin beads have also been proposed recently as bulk solid standards for electrothermal atomic absorption spectrometry (27) and neutron activation analysis (28, 29). The sorption properties of chelating ion exchange resins with iminodiacetate functional groups have thoroughly been studied for a number of elements (30-35). In this paper we propose a novel standard system for LAMMA analysis of biological thin sections, based on Chelex-100 ion chelating resin beads, which can be coembedded with the biomedical specimen. The use of an internal standard for LAMMA analysis of biological sections avoids the important problems that may arise when samples and standards potentially differ in section thickness and in other respects, as is the case for the commonly used and commercially available external standard foils, like organometal-doped epoxy resins. Schroder (7) proposed doping of the embedding media with small amounts of organometal compounds to serve as internal standards. However several problems emerged: at the micrometer level the metal distribution in the epoxy resin was not always sufficiently homogeneous, and preferential binding of the metal occurred at some sites within the biological material together with large local variations in the mass fraction ratio of specimen and embedding material. On the contrary, in the proposed LAMMA calibration method with embedded ion chelating resin beads, the added metals are well immobilized within the
0003-2700/84/0356-1362$01.50/00 1984 American Chemical Society
ANALYTICAL CHEMISTRY, VOL. 56, NO. 8, JULY 1984
resin particles, and this represents an additional advantage. Without disregard of the inherent multielement character, interest was mainly focused on platinum and aluminum. Other elements also present in the ion exchange standards were sodium, potassium, and calcium, being representative for important physiological cations. The platinum model system was preferred in view of current quantitative LAMMA studies of tissue sections after treatment with the cisplatin antitumor drug (36). Aluminum standardization is important in relation to microanalytical studies of the accumulation of this element in tissues of chronic hemodialysis patients (37). EXPERIMENTAL SECTION Reagents. Chelex-100 analytical grade chelating resin was purchased from Bio-Rad Laboratories. The preferred bead size was 35-75 pm (dry sodium form). No preceding purification of the resin was carried out. The platinum stock solutions were made so as to obtain two different chemical states, HzPtC16and Pt(NHJ4C1+ The Pt(1V) standard solution was prepared by dissolving an appropriate amount of pure metal foil (Johnson Matthey & Pauwels) in aqua regia and expelling the excess nitric acid by repeated evaporations with concentrated hydrochloric acid. Dilution was made with bidistilled water. The Pt(I1) solution was obtained by dissolving platinum(I1) chloride (Aldrich Chemical Co.) in concentrated ammonia under stirring and heating for a few hours. The solution volume was finally reduced by evaporation to remove excess of ammonia and diluted with bidistilled water. The platinum content was assessed by X-ray fluorescence analysis against the Pt(1V) standard: the platinum content amounted to 94% of the theoretical stoichiometric value. In the case of pH restricted equilibration experiments the buffer solutions for pH 5.0-5.5 were either sodium citrate or acetate. Stock solutions for the multielement or the other one-element loading experiments were prepared by dissolving appropriate water-soluble salts of the elements of interest. In some cases commercial AAS-certified standard solutions have been used instead. All acids and other produds used were of analytical grade quality. Ion exchange beads of the type Dowex 50W (X8) in the hydrogen form and the AG1-X8 in the chloride form were obtained from Fluka AG and Bio-Rad Laboratories, respectively. Both were of the 200-400 mesh size range. Cellulose powder was modified with diethylenetriamine (DEN) in this laboratory (38)according to an optimized procedure. The resulting 2,2'-diaminoethylamine-(deoxy)cellulose powder (cellulose-DEN powder) was repeatedly washed with water and NHIOH. Apparatus. For the LAMMA analyses, the LAMMA-500 instrument of Leybold-Heraeus (Koln, F. R. Germany) was used; schematic diagrams and detailed construction data for this machine are readily available (1). A high-power pulsed UV laser of the Nd:YAG type (A = 266 nm; 7 = 15 ns) is focused onto the specimen area of interest indicated by a red collinear He-Ne search laser. After perforation of the section the created microplasma is extracted into a time-of-flight mass spectrometer. The ions ("+" or *-' option) are collimated by an ion optical lens system and separated according to their mass, producing a complete mass spectrum for each laser pulse action. Throughout this work the lower magnification mode of the LAMMA light microscope (32X objective) was used to focus the laser beam onto the samples. The energy-dispersive X-ray fluorescence (XRF) apparatus included a high-voltage generator, an X-ray tube, a secondary fluorescer, a Kevex-0810 subsystem, and a Si(Li) detector. Both Mo and Ti excitation were used if possible to enhance the sensitivity and to disclose possible particle size effects for the lighter elements. The data reduction routine (39) directly delivered corrected elemental concentration values. Dried ion chelating resin beads were suspended in bidistilled water and filtered off on a Nuclepore membrane to obtain a homogeneous 2-3 mgcrn" load and a reproducible target geometry for XRF analysis. The neutron activation analysis (NAA) involved irradiations at IO1*n.s.-1.cm-2 in the Thetis reactor of the State University
1363
of Ghent. An amount of dried beads was weighed in a small bag made from "Fresh-Pak" (Union Chimique Belge) and pressed into a pellet. After irradiation and suitable decay periods, the y radiation was measured by Ge(Li) spectroscopy. Transmission electron microscopy (TEM) was performed on a Zeiss EM 109 unit and electron probe X-ray microanalyses were carried out on a Jeol Superprobe 733 or a Philips EM 400 unit combined with a Tracor-Northern software system. Procedure. Loading of the Zon Exchange Beads. (a) Pt(1V) Experiments. Undried resin in Na form was equilibrated under gentle stirring with Pt(1V) solution in aqueous acetone medium (1:l)in the presence of glycine with a molar ratio Pkglycine of either 1:20 or 1:lO (40). The solution was acidified with hydrochloric acid to ensure a pH below 1. An equilibration time of at least 48 h was preferred. (b) Pt(I1) Experiments. Undried resin in Na form was equilibrated with Pt(I1) solution in pure water at pH 5 (sodium citrate buffer). (c) Pt/Na Experiments. Freshly prepared resin in H form was equilibrated with a sodium acetate buffer of varying molarity. Each time the same amount of Pt(I1) was added. (d) Al/Fe Experiments. Undried resin in Na form was equilibrated with a solution containing AI and Fe at acid pH (pH C3). (e) One-Element Loading Experiments for Ca, Pt, Na, and H. Undried resin in the suitable ionic form was equilibrated with a solution containing a high concentration (e.g., 1 M) of the particular element to be loaded. ( f ) Multielement Experiments. Stock solutions containing either Al, Ca, Cr, Fe, Ni, Sr, Rb, Cs, La, and Pb (solution A) or Ca, Cr, Mn, Co, Rb, Cs, and Ce (solution B) were equilibrated with undried resin in the Na form. The former was equilibrated at very acid pH and also at a pH about 5 (addition of ammonia did not induce formation of any precipitates). The latter was equilibrated at only very slightly acidic pH (no extra acid was added to a solution of chlorides in bidistilled water). For most of the loading experiments described, 2 g of undried resin and a total volume of about 80-90 mL of equilibrating solution were taken. Preparation for L A M M A Analysis. After equilibration, aliquots of the Chelex-100 beads were taken from the solution and filtered off immediately on a Nuclepore filter under suction. The beads were rinsed thoroughly with the buffer solution and/or bidistilled water and then oven dried at 60 "C. Before being embedded, bead clusters were separated by gently pressing with a glass rod. While part of the loaded beads were kept apart for additional bulk analysis, some of the powder was poured into the top of BEEM capsules and freshly prepared final Epon 812 mixture was added. After dispersion of the beads in the Epon solution by stirring, polymerization was achieved as is usual in the electron microscopy laboratory (41). Thin sections of the desired thickness (usually 0.2-0.35 pm) were cut on a Reichert Ultracut ultramicrotome with glass knives and collected on unfilmed 300 mesh copper grids. It should be noted that beads loaded at acid pH provided the best sections after embedding. Figure 1is an illustrative electron micrograph of such a thin section with Pt-loaded Chelex-100 beads coembedded with kidney tissue. Other resin bead types were prepared similarly and were also embedded directly in Epon. The AG1 beads floated on the Epon mixture and had to be cut at the backside of the BEEM capsules. The dried cellulose-DEN powder was very hard to cut and resulted in small irregular fragments in the sections. The other resin types gave rise to clean circular bead sections, but often many beads left the embedding medium during the cutting step leaving holes in the sections. RESULTS AND DISCUSSION Metal Uptake of Chelex-100 Beads. The metal uptake characteristics of Chelex-100 resin have been studied by various authors and compiled by the manufacturer (35). Very few data (40, 42) are, however, available as to its behavior toward Pt, the element of main interest in this work. Therefore the loading of Chelex-100 with Pt(1V) and Pt(II), from solutions containing alkali and alkaline-earth ions, was studied in some detail. Both NAA and energy-dispersive XRF
1364
ANALYTICAL
CHEMISTRY. VOL. 56. NO. 8. JULY
1984
Flgure 2. Influence of sodium molarity on Pt(l1) uptake by the Chelex-100 resin beads. An initial 3.5 X IO-' mol of Ptlg of wet Chelex (H form) was used.
Electron micrograph (magnification:360x1 of a thin seclion containing Rime& Chelex-100 beads coernbedded wilh kaney t i i c e . supported by a conventional capper electron microscopic grid. (The tissue is seen in the top iefi square. a complete Cheiex-100 bead section in the lower right square. broken bead sections in t h e lower iefi square.) W r e 1.
with Mo excitation were invoked to determine Pt in the loaded beads. The XRF results which showed about 11% coefficient of variation in the range l(t50 fig of Pt/g dry heads, agreed within 10-15%, on the average, with those obtained by NAA (N= 7). In the higher concentration range from 50 to 7000 ppm the XRF results showed a Coefficient of variation of about 790,ranging from 5 t n 8%. The Pt content of loaded Chelex-100 beads was measured as a function of the initial Pt(1V) concentration of the equilibrating solution. Two different glycine-to-metal concentration ratios were compared (g1ycine:metal ratio equal to 10 and 20; N = 6 and N = 1, respectively). The Pt(IV) uptake from solution remained controllable and related proportionally to the original Pt concentration in the range 0.4-60 p g mL-'. The initial Pt:glycine ratio in the solution did not seem to influence the metal retention. In these static investigations the retention of metal ions was expressed as a percentage in relation to the initial concentration of cations in the solution. The data correspond to 8090 uptake yield, in full agreement with literature data (40). If one defines the distribution coefficient D as
D=
a m t of metal taken u p by 1 g dry resin a m t of metal remaining in 1 m L solution
*
the data point to a log D value of 2.8 0.2 for the Pt(1V) loading experiments under the used conditions of low pH. The affinity of Chelex-100 for Pt(I1) ions, under the given experimental conditions, appeared considerably lower, with log D in the range 1.3-1.8. In this case the data reflect a 12-29% Pt retention yield. Since most transition metals exhibit higher log D values, their uptake will be more efficient. The loading of Chelex-100 (H form) with alkali ions showed very low resin affinity, as is well known. It was, e.g., found that the uptake of 3.5 X lod mol of Pt(II)/g of wet Chelex-100 resin in the H form was suppressed only in buffers (pH about 5.5) with a Na molarity above 5 X IO-* (see Figure 2). Still, as NAA measurements showed, Chelex-100 can be loaded in
a controlled way with Na+ also, provided a high concentration of the alkali metal is present in the absence of considerable amounts of transition-metal elements, which show far greater resin affinity. Loading saturation is appearing a t the higher initial Na molarities (about 0.3 M) corresponding to Na concentrations above 290 (w/w) in the heads. In the AI/Fe loading experiments the AI content of the Cbelex-100 resin heads was determined with NAA. Also for Fe some NAA measurements were made as a cross-check with the XRF results. Again, as for Pt, very good agreement within about 10% was ohtained ( N = 2). In our experimental loading conditions, using a pH below the optimal level of 4 (42),we obtained metal ion retention yields ranging from 4 to 26% for AKIII) and 30 to 90% for Fe(II1). Alkaline-earth elements like Ca also allow controllable loading. A linear relation was found between the difference in concentration of Ca before and after equilibration (assumed to he taken up by the beads) and the initial concentration of the solution, as measured by atomic absorption spectrometry (10). Also, simultaneous loading with cations showing similar resin affinity (Ca*+,Mg2+)is possible when the total ionic strength of the solution is kept constant (10). However, for the simultaneous loading of Pt(1I) and Ca onto the Chelex-100 heads, the weight ratios between the two elements on the beads, as measured by XRF, did not closely follow the initial elemental concentration ratios. For multielement loading purposes one must simultaneously take into account the various elemental selectivity coefficients for the particular resin type, the relative elemental concentrations used, and the solution pH during equilibration. One can easily imagine the interest of loading the chelating resin simultaneously with heavy metals and physiological cations under biologically equivalent proportions. LAMMA Perforation Characteristics and Dimensions. The laser-induced perforations in the specimens under investigation have to he studied carefully because knowledge of the type and amount of evaporated and analyzed material is essential to quantitative work. It is important that the perforation dimensions and geometry can be evaluated accurately and that the laser-drilled hole is sharp-edged. while edge mass desorption effects should he minimized A t the same laser power the head sections were more easily perforated than the epoxy resin embedding medium, and usually larger laser holes were ohtained. In this way, they are more comparable to biological specimens, probably because of a more similar local mass density, and this represents an advantage in the study of soft tissues or individual cells. The
ANALYTICAL CHEMISTRY, VOL. 56, NO. 8, JULY 1984
resulting laser hole diameter can be controlled to a certain extent by varying the laser power. Electron microscopy can be used to determine the mean laser perforation diameter and to monitor the section quality. For a given sample the minimum perforation dimensions that can be attained with the element(s) under investigation still being detected depend on the sensitivity, concentration, and spectral characteristics of these element(s). In principle, hole diameters of about 1-2 pm should easily be obtained. In this study the laser-induced perforations varied from about 4 to 8 pm in dimension for the Pt measurements, because of the higher laser power needed to detect relevantly low Pt concentrations (36). LAMMA Analysis of Several Types of Embedded Resin Beads. The typical organic mass fragmentation pattern for the matrix of the various ion chelating and ion exchange resin beads under consideration can result in mass interferences on analytical mass lines, especially at trace levels. Therefore the mass spectral background pattern of the different resin beads was studied. Particularly for lower mass elements it might be interesting and sometimes necessary to change from either one to the other standard system depending on which exhibits a more suitable organic mass fragmentation pattern for the trace element under research. In any case due to various possible influences on the obtained background pattern, to be illustrated later, the choice of standard matrix material is never trivial. We selected the following ion chelating and ion exchange resin beads for LAMMA mass spectral microanalysis: (i) Chelex-100 in the H form, (ii) Dowex-50 W in the H form, (iii) the anion exchange resin Bio-Rad AGl-X8 in the C1 form, (iv) 2,2'-diaminodiethylamine-(deoxy)-cellulose powder (cellulose-DEN powder). The H form was preferred for mass spectrometric analysis in order to limit the formation of inorganic molecular cluster ions arising from loaded metal ions. Positive Ion LAMMA Spectra. In Figure 3 the positive ion LAMMA mass spectra of the selected ion exchange beads are shown (spectra a-d) together with a mass spectrum originating from the Epon embedding medium (spectrum e). All mass spectra presented were recorded under identical instrumental conditions. The well-defied chemical structure of Chelex-100 and also of the other bead types invites one to examine the obtained mass spectra more thoroughly. Embedding resins, on the contrary, consist of mixtures of several organic compounds under variable proportions and polymerization conditions. The mass spectra (a), (b), and (c) of Figure 3 originate from beads with a polystyrene-divinylbenzene matrix. The corresponding functional groups are iminodiacetate, sulfonic acid, and quaternary ammonium, respectively. Similar mass spectra are observed for the Chelex (a) and Dowex (b) samples in contrast to the AG1-anion exchanger (c). In the former two cases the presence of Na, K, and Ca is evident. The mass peaks at m / z 57 and m / z 63-66 are probably inorganic molecular cluster ions related to these elements, as will be discussed further below (see also Figure 5). These mass peaks interfere, e.g., with the elemental mass lines of Cu and Zn. Also Fe is interfered at m/z 57 in both cases, but the main Fe mass line at m/z 56 is essentially free from any interference (CaO+), particularly in the Chelex-100 mass spectra. At low mass we should note C2Hn+-, C3H,+- carbon cluster ions and the CH,=N+H2 (mlz 30) mass fragment from the Chelex sample. The intense line at m / z 24, which was found to be typical for Chelex-100 in the H form (see also Figure 5), should probably be identified as an organic mass fragment, but Mg+ cannot be excluded. Anyway, mass interference for the detection of Mg is clearly present. For the Dowex samples, the fragment with m / z 27 introduces mass interference with the monoisotopic Al+ mass line. A t high mass, the LAMMA
Na'
I
Na'
I
'K
1365
La
II
K+
1 :"'
CSH+ 58
GH'
44
I
K'
55?
Figure 3. LAMMA mass spectra ("+") from embedded and thin sectioned (0.35 pm) ion chelating and ion exchange resin beads (a) Chelex-100 (H form); (b) Dowex-50W (H form); (c) Bio-Rad AG 1-X8 (Clform); (d) 2,2'-diaminodiethylamine-(deoxy)-cellulose powder (celIulose-DEN powder) and (e) the Epon embedding medium. Instrumental conditions (laser energy per pulse in pJ, ion lens potential in V, input range in V) are for all spectra: 11, -1 150, 0.5, respectively.
spectrum (a) of Figure 3 shows some ions typical for Chelex-100 (H form) at m / z 91, 115, 117, and 160, which are probably of organic origin. The mass peaks at m l z 91 and m / z 115 could be assigned to fragment ions from the polystyrene-divinylbenzene matrix, namely, C8H5CHz+ and C6H6C=C=CH2+, respectively. However, it is surprising that these features do not appear in the other two polystyrenedivinylbenzene matrices (spectra b and c in Figure 3). It was therefore also tempting to ascribe them to the iminodiacetate functional group of Chelex-100. However, even by comparison with the LAMMA mass spectra obtained from powdered pure iminodiacetic acid, it was not possible to identify these high mass ions unambiguously. Mass spectrum c of Figure 3 obtained for the AG1 anion exchange resin using the same laser power density as before, shows much more organic carbon fragmentation. The resulting C,H,+ ions follow a certain periodicity in ion intensity: for uneven n (n = 3, 5, 7,9) and even n ( n = 4, 6, 8, lo), C,H+ and C,H,+ are the most predominant ions, respectively. In absolute intensity, the carbon cluster ions (C,+) containing an uneven number of carbon atoms are more intense than the corresponding even carbon number ions. The specific organic fragment ions corresponding to the quaternary ammonium functional group of the AG1 resin were detected at m / z 42, 44, and 58 (CH,= N+=CH2,CH3N+H=CH2, (CH3),N+=CH2,respectively). At lower laser power densities these particular mass peaks (together with Na+ and K+) progressively increased with respect to the organic background C,H,+ ions, findy resulting in mass spectra showing almost exclusively m/z 58 ions. By variation of the laser power density during analysis, it is thus possible to differentiate between common pyrolysis products and specific organic mass fragments related to the functional
1366
ANALYTICAL CHEMISTRY, VOL. 56, NO. 8, JULY 1984
I
c;
c;
c;
80 I81
183
79
I
i
97
91 115
'a
E--
L
50
100
150
TI2
Figure 4. LAMMA mass spectra ("-") from the same samples as
illustrated in Figure 3. Instrumental conditions (laser energy per pulse in pJ, ion lens potential in V) are for all spectra: 11, 4-1 100. The input sensitivity was set on 0.5 V, except for (c), where 0.2 V was used. groups of the resin type. The "+" LAMMA mass spectra of the cellulose-DEN powder (spectrum d of Figure 3) predominantly showed specific organic fragment ions at mlz 30,44, 56, and 58, corresponding to the diethylenetriamine functional group. The relative abundance of the mlz 30,44 mass peaks is indicative of the presence of a primary -CHzCHzNH2chain of the DEN substituent. Comparison can also be made with LAMMA mass spectra from the Epon embedding medium (spectrum e of Figure 3) to evaluate possible organic background interferences encountered when using organometal-doped epoxy resin sections as standards for LAMMA analysis. Besides contamination by Na and K, the epoxy resin mass spectra consist of C,H,+ and C,H,O,+ ions. Typical masses are the couple at mlz 27 and 29 (C2H3+ and CzHs+)and mlz 41,43,55, and 58. Negative Ion LAMMA Spectra. Figure 4 presents the negative LAMMA mass spectra of the same ion exchange bead samples as described above. The spectra a and b originating from Chelex and Dowex, respectively, show a very similar C,H,- organic background pattern. For Chelex no characteristic fragment ions of significant intensity were observed, except for some traces at mlz 42,59,80, and 117. On the other hand, the sulfonic acid functional group of the Dowex sample could clearly be recognized: the most important ions are detected a t mJ z 64,80,81, and 183 and represent SO2-, SO;, HS03- and CHz==cHC6H4S03-, respectively. The matrix ion peaks C,H,- again follow certain systematics in their ion intensity distribution (see spectra a-c of Figure 4): for uneven n (n = 3,5,7,9) we obtained almost exclusively C; ions, with at most a very low intense C,H-, while for even n (n = 2, 4, 6, 8) the ions Cn-, C,H-, and CnH2- are all of comparable intensity (almost no C,H 0.99). power. In this case. one single section was measured again after repositioning in the sample chamber. Table 11. Results of Calibration Curves
a
concn range, ppm
no. of measmts per std
mean RSD,
element
no. of stds
%
corr coeff
intercept (std rlev), arbitrary units
Pt(1V) Pt( IV)" Pt(I1)" Al(II1)
5 3 4 6
50-920 50-920 50-4900 10-300
5-13 22-26 10-25 14-38
31 28 42 23
0.994 0.965 0.996 0.998
51 (8) 37 (20) -27 (44) -75 (56)
slope (std dev), arbitrary units/ppm 0.21 0.39 0.22 9.54
(0.013) (0.10) (0.02) (0.32)
Both series were measured in the same run. and the "'Pt+ peak height was alwavs used.
LAMMA Signal Reproducibility. The experimental signal variability of the nonnormalized absolute LAMMA ion intensities obtained for random shots on sections of embedded Chelex-beads loaded with Pt(1V) was optimally around 20-25% (coefficient of variation) and somewhat worse for Pt(I1). This signal variability includes contributions from variations in laser power density, focalization of the laser beam, section thickness, and evaporated volume. For the other elements loaded, the variability of the absolute LAMMA signals typically ranged from 20% to 45%. Easily ionizable elements like A1 usually had better signal reproducibility than elements with higher first ionization potential (like Fe). Specific metal-ligand bonding characteristics and element loading properties for Chelex-100 probably also influence the final signal variabilities obtained by LAMMA. Normalizing the absolute elemental signals to an organic background mass fragment of the Chelex-100 matrix (e.g., m / z 30,91,or 117) did not invariably improve the signal coefficient of variation. No significant LAMMA signal difference was observed when scanning from the edge of the embedded beads to their core. Also in most cases, for carefully loaded resin batches, the variability between LAMMA signals from shots on different beads was not significantly larger than between those obtained from a single bead. A one-way analysis of variance was applied onto a data set obtaihed by LAMMA microanalysis of a section containing numerous embedded Chelex-100 beads loaded with A1 and Fe simultaneously. For this particular sample, the F ratio, defied as the ratio of the between-groups variance estimate to the within-groups variance estimate, was equal to 1.09 for A1 and 15.7 for Fe, respectively, and after comparison with the critical F-value at P = 0.95 (i.e., 2.84 for the used degrees of freedom), it was found to be nonsignificant for Al, while for Fe there existed a real significant difference in this case. Apparently, for some elements the intrinsic loading properties or the quality of the loading might contribute t o the overall signal variability of the LAMMA measurements leading to a larger coefficient of variation of the ion signals. The variation in ion signal intensity between shots from different sections at a particular element concentration was
measured for some samples containing either A1 and Fe or Pt. The results obtained are presented in Table I. The relative discrepancies experimentally obtained for the absolute ion intensities range from a few percent up to considerable values but remain comparable to the overall precision of measurement. Having the two sections (of possibly slightly different thickness) on the same support grid, or on two different grids (implying additional uncertainties due to positioning and laser focusing), or repositioning the same section, did not lead to sufficiently systematic differences to allow conclusions about the relative importance of the sources of errors. The used instrumental conditions (particularly laser power density) revealed to be an important factor in relation to the observed ion signal variabilities and differences. The relative standard deviations per measurement appeared to be optimally 10-20%, but worse a t lower laser power, and for Fe that may be less homogeneously distributed throughout the sections. Results of Calibration Curves. The influence of elemental concentration on the absolute ion signal intensities was measured by LAMMA microanalysis of sections of Chelex-100 beads loaded with Pt or Al, which are known to be of significant biomedical importance (36, 37). For Pt, the abundant isotopes lg4Pt,lg5Pt,and 196Ptyielded the best results at the lower concentration end, because of spectral interferences on the minor isotope lg8Pt,but the reverse was true at the higher concentration end due to saturation effects in the detection and data storage system. The calibration curve using the peak heights of these isotopes showed a satisfactory linear relation up to a Pt concentration of a t least 1000 ppm by weight. The appropriate statistics applied for the data points in the linear range is illustrated in Table I1 (first row). A satisfactory correlation coefficient and a precision of 30% are observed. Relative to most other elements, the LAMMA sensitivity for Pt is low, which is not surprising regarding its high ionization potential. Still, it was felt that measurements down to about 20-30 ppmw are possible under optimal instrumental conditions and with high laser power. This value is very advantageous compared to results of wavelength-dispersive electron probe X-ray microanalyses on 0.35-pm sections of Pt-containing ion exchange beads and of
1370
ANALYTICAL CHEMISTRY, VOL. 56, NO. 8, JULY 1984
Pt-doped Epon resin, where a detection limit of about 1% was typically obtained. The experimental calibration curves obtained for Pt(I1)and Pt(1V)-loaded Chelex-100 samples were especially compared, as can be seen in the second and third lines of Table 11. The Pt(1V) standard type showed somewhat greater sensitivity, which might lead to faster signal saturation. A log-log plot indeed indicated a slope value of 1.04 f 0.09 (correlation coefficient = 0.993) for Pt(II), but only of 0.70 f 0.12 (correlation coefficient = 0.986) for the Pt(IV) samples. Such phenomena, either related to instrumental or physical properties, have to be examined further. For Al, using the integrated 27Alcsignal intensities, a linear signal-concentration relationship with a high correlation coefficient was obtained, also showing an overall precision around 25% (Table 11). A1 was found to be a very sensitive element toward LAMMA analysis, as can be seen from the higher slope of the calibration curve (see last line of Table 11). This is probably due to its fairly low first ionization potential. The favorable detection limit, which is estimated to be around 5 ppm, presents great promises for the microanalysis of tissue Al in biomedical investigations, particularly in nephrology (37).
ACKNOWLEDGMENT The assistance of the electron microscopy laboratory personnel at the Academic Hospital in Edegem is greatly appreciated. The help of L. Van't dack in carrying out the NAA measurements is gratefully acknowledged, as are the secretarial skills of M. Stalmans. This work was partially supported by the Belgian Ministry of Science Policy under Grant 80-85/10. Registry No. Na, 7440-23-5;Al, 7429-90-5;Fe, 7439-89-6;Ca, 7440-70-2; Pt, 7440-06-4; Cr, 7440-47-3; Ni, 7440-02-0; Sr, 744024-6; Rb, 7440-17-7;Cs, 7440-46-2;La, 7439-91-0;Pb, 7439-92-1; Mn, 7439-96-5; Co, 7440-48-4; Ce, 7440-45-1. LITERATURE CITED Proceedings of the LAMMA Symposlum held in Dusseldorf, Federal Republic of Germany, Oct 6-10, 1980 (Organizers: F. Hlllenkamp and R. Kaufmann), Fresenius' 2.Anal. Chem. 1981, 308, 193-320. Denoyer, E.; Van Grieken, R.; Adams, F.; Natusch. D. F. S. Anal. Chem. 1982, 5 4 , 28A-41A. Hercules, D. M.; Day, R. J.; Balasanmugam, K.1 Dang, T. A.; Li, C. P. Anal. Chem. 1982, 5 4 , 280A-305A. Schmldt, P. F.; Fromme, H. G.; Pfefferkorn, G. Scannlng Electron Mic ~ O S C .1980,
II * 623-634.
Seydel, U.; Llndner, B. In!. J. Quantum Chem. 1981, 2 0 , 505-512. Sprey, B.; Bochem, H.-P. Fresenius' Z . Anal. Chem. 1981, 308,
de Bruljn, W. C. Scanning Electron Microsc. 1981, I I , 357-367. Chandler, J. A. J. Microsc. 1976, 106, 291-302. Nakamura, K.; Orii, H. Anal. Chem. 1980, 52, 532-536. Wieser, P.; Wurster, R.; Seller, H. Scanning Electron Microsc. 1982, I V , 1435-1441. Gijbels, R.; Veriodt, P.; Tavernler, S. Paper presented at the 16th Annual Meeting of the Microbeam Analysis Society, Washington DC, 9-13 A u ~1982. Anderson, D. H.; Murphy, J. J.; White, W. W. Anal. Chem. 1972, 4 4 ,
2099-2 100.
Ingram, F. D.; Ingram, M. J.; Hogben, C. A. M. I n "Microprobe Analysis as Applied to Cells and Tlssues"; Hall, T., Echlin, P., Kaufmann, R., Eds.; Academic Press: London, 1974; pp 119-146. Dorge, A.; Gehring, K.; Nagei, W.; Thurau, K. I n "Microprobe Analysis as Applied to Cells and Tissves"; Hall, T., Echlin, P., Kaufmann, R., Eds.; Academic Press: London, 1974; pp 337-349. Roomans, 0. M.; SevBus, L. A. J. Submlcrosc. Cytol. 1977, 9 ,
31-35.
Trump, B. F.; Berezesky, I. K.; Pendergrass, R. E.; Chang, S. H.; Bulger, R. E.; Mergner, W. J. Scannlng Electron Microsc. 1978, II,
1027-1039.
Hagler, H. K.; Lopez, L. E.; Flores, J. S.; Lundswick, R . J.; Buja. L. M. J. Microsc. 1983, 131, 221-234. Burns-Bellhorn, M. S.; Flle, D. M. Anal. Biochem. 1979, 92,213-221. Zhu, D.; Harrls, W. C., Jr.; Morrison, G. H. Anal. Chem. 1982, 5 4 ,
419-422.
Edelmann, L. Fresenlus' Z . Anal. Chem. 1981, 308, 218-220. Verbueken, A. H.; Jacob, W. A.; Frederlk, P. M.; Buslng, W. M.; Hertsens, R. C.; Van Grieken, R. E. J. Phys. (Orsay, Fr.), in press. Freeman. D. H.; Paplson. R. A'. Nature (London) 1989, 216, 563-564. de Bruiin, W. C. Beitr. Elektronenmlkroskor,.Direktabb. Oberfl. 1981, 14, 369-372. Alder, J. F.; Batoreu, M. C. C. Anal. Chim. Acta 1982, 135, 229-234. Kayasth, S. R.; Iyer, R. K.; Sankar Das, M. J. Radioanal. Chem. 1
.
1960,59,373-379.
Kayasth, S. R.; Dhai, y. B. Radiochem. Radioanal. Lett. 1980, 4 4 ,
403-412. Leyden, D. E.; Underwood, A. L. J. Phys. Chem. 1984, 6 8 ,
2093-2097. Loewenschuss, H.; Schmuckler, G. Talanta 1964, 1 1 , 1399-1408. Riley, J. P.; Taylor, 0.Anal. Chlm. Acta 1988, 4 0 , 479-485. Luttrell, G. H., Jy.; More, C.; Kenner, C. T. Anal. Chem. 1971, 4 3 ,
1370-1375.
Van Grieken, R. E.; Bresse!eers, C. M.; Vanderborght, B. M. Anal. Chem. 1977, 4 9 , 1326-1331. Blo-Rad Laboratories Chemical Division, Technlcal Bulletln 2020, March 1981. Verbueken, A. H.; Van Grleken, R. E.; Paulus, G. J.; Verpooten, G. A,; De Broe, M. E. Blomed. Mass Spectrom., In press. Verbueken, A. H.; Boelaert, J.; Pauius, G. J.; Van de Vyver, F. L. J.; Roels, F.; Van Grieken, R.; De Broe, M. E. N6phrologie 1983, 4 , 95. Reggers, 0.; Van Grieken, R. Fresenlus' Z . Anal. Chem., 1984, 317,
520-526.
Van Dyck, P. M.; Van Grieken, R. E. Anal. Chem. 1980, 52,
1859-1864.
Brajter, K.; Miazek. I. Talanta 1981, 26, 759-764. Luft, J. H. J. Biophys. Biochem. Cytol. 1961, 9 ,409-414. Brajter, K.; Grabarek, J. Analyst (London) 1978, 103, 632-642. de Bruljn, W. C. J. Phys. (Orsay, Fr.), in press. de Bruijn, W. C.; Zeelen, J. Ph. Beitr. Elektronenmikroskop. Direktabb. Oberfl. 1983, 16, 385-388. Danzer, K.; Marx, G. Anal. Chim. Acta 1979, 110, 145-151.
239-245. Schrijder, W. H. Fresenius' 2. Anal. Chem. 1981, 308, 212-217. Spurr, A. R. J. Mlcrosc. Blol. Cell. 1975, 22, 287-302. Roomans, G. M. Scannlng Electron Microsc. 1979, II, 649-657.
RECEIVED for review September 9, 1982. Resubmitted December 5, 1983. Accepted February 22, 1984.