JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 167 Detection of Surface Aggregates of Trace Amounts of Copper and Silver on Graphite Using Secondary Ion Mass Spectrometry at Elevated Temperatures" Invited Lecture Jason G. Jackson Rodney W. Fonesca and James A. Holcombet Department of Chemistry and Biochemistry University of Texas at Austin Austin TX 78712 USA Static secondary ion mass spectrometry is used to investigate the extent of dispersion of copper and silver on a pyrolytic graphite coated graphite surface. The appearance of secondary fragment ions consisting of more than one atom of the metal analyte is used as an indication that the sample is present on the surface as aggregates. This signal for the polymeric metal ion is compared with the signal for the monatomic fragment ion for various concentrations.Concentration studies were conducted for both metals and copper was found to exist as dispersed copper species at lower concentrations while at higher concentrations evidence suggests surface aggregates or microdroplets. Silver was found to exist as aggregates for the range of concentrations studied. Keywords Secondary ion mass spectrometry; copper; silver; graphite; surface aggregation The signal in electrothermal atomic absorption spectrometry (ETAAS) is dependent in part on the initial generation of atoms from the surface. If a kinetic model is considered for the release then the activation energy E surface coverage 0 and order of release n are the governing parameters. It has been suggested that the order of release may give some information on the morphology of the desorbing species on the surface during the desorption process.'-4 In vacuum desorp- tion studies Redhead' suggested that the order of release could be deduced from the general shift in the peaks when the surface coverage was altered.McNally and Holcombe2 applied this theory to vaporization processes with a graphite furnace in ETAAS applications. More recently Rojas and Olivares6 and Yan et al.7 presented a means of mathematically treating the absorbance data to permit quantitative determination of a value for n. However all these techniques provide information only at the temperature where desorption occurs and represent indirect evidence of the analyte morphology on the surface. Surface techniques have been used to study metal clusters directly on graphite although some require atomic level smoothness for their success (e.g.scanning tunnelling microsc~py*~~) and others require relatively high surface cover- ages for detection (e.g. Auger spectroscopy" and electron spectroscopy for chemical applications). Unfortunately in ETAAS extremely low analyte concentrations located on microscopically roughened surfaces are of interest It may be possible to overcome both of these limitations using secondary ion mass spectrometry to study trace metals deposited on pyrolytic graphite coated graphite surfaces used in ETAAS. In this study temperature programmed static secondary ion mass spectrometry (TPS-SIMS) was evaluated as a means of differentiating between clusters of analyte and dispersed adsorbed analyte on the surface.Specifically the extent of dispersion of Cu and Ag was investigated using TPS-SIMS both at room temperature and during a thermal ramp to determine if any morphological changes occurred preceding vaporization. Experimental Apparatus and Solutions The TPS-SIMS system has been described previously.12 The primary Ar' ion beam was operated at 1 nA and 2.3 keV * Presented at the XXVIII Colloquium Spectroscopicurn Inter- t To whom correspondence should be addressed. nationale (CSI) York UK June 29-July 4 1993. using a spot size of approximately 100p.m. The beam was rastered across a surface area of approximately 2 x 2 mm. This low primary ion flux and beam energy implies that only the outermost layers of the sample surface are probed and that there is minimal surface perturbation over the analysis time.This ensures operation under 'static' conditions. In most studies solution samples containing the dissolved analyte salt were deposited on a pyrolytic graphite coated graphite platform (10 x 5 x 1 mm Stackpole/UltraCarbon Bay City MI USA). The platform was radiatively heated by a tantalum-strip heater located below the platform. The heating rate was maintained by a proportionally controlled heating circuit with feedback provided from a thermocouple placed within a hole (0.6 mm diameter x 2 mm long) drilled into the graphite platform. All data collection was performed by software written in ASYST in this laboratory running on an 80486 based PC interfaced to a Keithly Series 500 data acquisition and control system.High-purity Ag wire was dissolved in HNO and diluted to prepare a stock solution of 1000 mg 1-1 Ag in 0.5% HNO,. A stock solution of copper nitrate was prepared by dissolving ACS reagent grade CU(NO,)~.~+ H 2 0 in distilled de-ionized water. The working solutions were prepared daily by dilutions of the stock solution. Procedure A 2 p.1 aliquot of either the Cu or Ag solution was placed on the graphite platform located on the translational rod12 and dried at approximately 60°C. The platform was then trans- ferred through two differentially pumped stages into the main vacuum chamber. The base pressure of the main chamber was less than 8 x lop9 Torr (1 Torr = 133.322 Pa) and was approxi- mately 3 x Torr with the SIMS Ar+ gun active.Once the platform was accurately positioned under the mass spec- trometer the temperature of the platform was ramped from ambient temperature to approximately 773 K at a rate of 1 K s-'. Results and Discussion The SIMS technique does not necessarily sputter 'stoichio- metric units' from the surface a fact which is often the source of the difficulty in spectral interpretation. However when clusters are removed they originate in the immediate vicinity of the primary-ion collision with the surface. As a consequence,168 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 t a) if condensed-phase microparticles microcrystals or micro- droplets of metal or metal containing compounds are located on the surface one would expect to see secondary ions containing more than a single metal atom.To verify this postulate a simple sheet of Cu was used in place of the graphite platform and served as the primary Ar+ beam target subject to static-SIMS analysis. Fig. l(a) shows the static SIMS spectrum for the higher masses of the pure Cu sheet at 298 K. Owing to the close proximity of Cu atoms on the surface peaks for Cu2+ and Cu3+ are evident. Oxide species (e.g. Cu20f) are present as a result of an expected oxide layer on the Cu. As the temperature increases the oxide peak decreases and Cu2+ and C U ~ ( O H ) ~ + dominate [Fig. l(b)]. At 751 K the Cu30+ increases in inten- sity relative to that observed at lower temperatures [Fig. l(c)]. The distinct appearance of polyatomic Cu ion species from these closely spaced Cu atoms of the Cu sheet suggests that such spectral features may be a characteristic of any micro- scopic clusters of a metal on a surface.In contrast dispersed metal species should exhibit a low probability of encountering two or more closely positioned metal atoms at the site of the collision of the primary-ion beam with the surface. As a result 1 cu,oi cuz+ CuJOH),' 100 120 140 160 180 200 mlz Fig. 1 and (c) 747 K S-SIMS spectra of a copper platform at (a) 298 (b) 429 only monomeric metal ions should be detected. Thus this information may be useful in investigating the surface mor- phology of metals on a graphite surface. It can also be seen in 'Fig. 1 that as temperature is increased there is a general decrease in the various oxygen containing species this could Ibe due to the changes in the degree of surface oxidation or some other change in surface species.Copper solution samples resulting in deposited Cu masses ranging from 2-2 000 ng were dried onto a pyrolytic graphite coated graphite plaiform. The ratio of the intensity of lCu,O+:Cu+ at room temperature was used as an indication of the extent of aggregation on the surface. As shown in Fig. 2 the ratio is relatively constant for masses above 40ng but dropped significantly for the 2 and 20ng sample loads. The Cu20f signal was used in place of another polymeric Cu ion because of its greater sensitivity. The error bars on Fig. 2 are due to both variations between different experiments and variations in the baseline. The former variation affects the higher masses 40-2000 ng and since for 2 and 20 ng there is no observable signal the variation in the baseline is dominant.If the Cu20+ signal were to retain the same ratio with respect to the Cu' signal it should be easily detected above the noise level even at 2 ng of deposited Cu. Thus its decrease suggests that the sample is present as dispersed Cu atoms or Cu containing molecules at the lower concentrations. Figs. 3 and 4 show the contrasting spectra for 40 and 20 ng sample loads respectively. The signal-to-noise ratio for the aggregates (e.g. Cu2O+ and Cu2+) is large with a 40ng sample but for the ;!Ong sample the polyatomic ion signals are barely detectable above the noise even though there is only a factor of two decrease in the amount of Cu that has been 0.25 0.20 .- 0 0.15 F 3 0.10 u .- v) C 0.05 - 0 -0.05 f 1 10 1 x 102 1 ~ 1 0 3 1 ~ 1 0 4 Mass/ n g Fig.2 Ratio of Cu,O+:Cu+ measured at room temperature for a 2 pl sample of a copper nitrate solution deposited on pyrolytic graphite coated graphite at increasing concentrations I4O I 120 - - I v) C v) 100 - e 80 - s m 6 0 - -. c > v) 5 40 20 .- - 4- C - - cu 0 50 60 70 80 90 100 110 120 130 140 150 m/z Fig.3 nitrate on a pyrolytic graphite coated graphite platform S-SIMS spectra of 40ng of Cu deposited as the aqueousJOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 169 - - - - - 3 0 ~ I 30 I 25 ," 20 4- 8 m 15 g > 4- .- l o 2 w - 5 - - 0 25 I m f 20 3 0 15 r -. Y > 'v 10 Q) C .+d - 5 50 60 70 80 90 100 110 120 130 140 150 m/z Fig.4 S-SIMS spectra of 20ng of Cu deposited as the aqueous nitrate on a pyrolytic graphite coated graphite platform deposited.While uncertainty in the analyte position after initial droplet drying makes absolute quantitative comparisons from one sample to the next very difficult the relative signal magnitudes within the same sample of different m/z values should be relatively accurate. These results indicate that Cu is located on the graphite surface as dispersed adsorbed species and not as microdroplets islands or any other aggregated form at solution concen- trations below 0.02 ppm Cu (40 ng). This is consistent with the first order of release and dispersed atoms as suggested by McNally and Holcombe2 and is supported by the work of Lynch et d3 The results also suggest that the low concen- trations of Cu are dispersed on the surface immediately after the sample is dried. For an Ag metal sheet the SIMS spectrum also shows the presence of poly atomic species.The dominant polymeric frag- ment observed for the Ag sheet was Ag,'. Silver samples with masses between 2 and 2000 ng were also deposited as 2 p1 sample volumes and dried onto a pyrolytic graphite coated graphite platform. Fig. 5 shows that unlike Cu the ratio of the intensities of Ag,+:Ag+ after drying is constant over the entire concentration range studied. This implies that the Ag on the surface is present as microdroplets or some other form of aggregates within the concentration range studied. The error bars represent primarily variations between different experi- ments and as in Fig.2 the error bars at lower sample masses are due to variations in the baseline. Using the alignment of the ETAAS peaks and the general variation of E Fonseca et concluded that Ag existed as aggregates above 0.1 ng and probably was present as dispersed atoms for less than 0.07 ng of Ag. Eloi et a1.,14 however found no evidence to suggest a surface bound species of Ag as they had seen for Cd 0'1° I Ma sstng Fig. 5 Ratio of Ag2+:Ag+ measured at room temperature for a 2 pl sample of a silver nitrate solution deposited on pyrolytic graphite coated graphite at increasing concentrations and Pb when using Rutherford backscattering spectroscopy as a probe. Instead they suggested that most of the sample was distributed in the bulk of the graphite platform. Various Cu and Ag ion signals were monitored continuously to provide data for TPS-SIMS.In this technique the tempera- ture is ramped to the desired final temperature (uiz. 773 K) then linearly cooled at the same rate. Multiple ion monitoring was used to record the temperature-dependent intensities for several masses. The resulting plots of signal intensity versus temperature can provide information on the reversibility of processes within the temperature range studied. Fig. 6 shows the TPS-SIMS plot for different Cu containing ions with a 400 ng load. It is interesting to note that the signal decreases rapidly at approximately 550 K. As the final tempera- ture which is below the atomization temperature in vacuum is reached and the sample begins to cool the signal remains below detectability.Data collection is stopped when heat loss from the platform is slower than the desired cooling rate. When the sample finally cools to room temperature the signal remains at baseline. This indicates that the Cu is irreversibly removed from the surface (e.g. sub-surface migration) or transformed into another chemical form which is less sensitive to positive ion SIMS detection. At the lower concentrations where the samples exist as dispersed atoms a different trend is seen in the TPS-SIMS scans. Fig. 7 shows that at lower concentrations there is a slow decrease in the signal intensity of the same masses shown in Fig. 6 but without a peak appearing near 550 K. Again little signal is detected in either case for these ions above 600 K. By monitoring evolved gases from a microgram sample load during a thermal ramp it is evident that the signal decrease at 550 K corresponds to the final decomposition of the basic 30 35 .- 25 v) 20 00 "0 15 ? -.4- r .- g lo a 4- - 5 0 300 400 500 600 700 Tern perat u re/K Fig.6 TPS-SIMS scan of 400ng of Cu deposited as the aqueous nitrate on a pyrolytic graphite coated graphite platform 160 140 7 120 m v) 100 13 5 80 > 2 60 40 20 .w .- a c - 300 400 500 600 700 Temperature/K Fig. 7 TPS-SIMS scan of 2 ng of Cu deposited as the aqueous nitrate on a pyrolytic graphite coated graphite platform170 JOURNAL OF ANALYTICAL ATOMIC SPECTROMETRY MARCH 1994 VOL. 9 dispersed molecules on the surface. Thus static-SIMS can be used to give indications of the extent of metal dispersion on a substrate.Data for Cu show that at sample amounts below 20 ng Cu is dispersed and probably exists as dispersed atoms or copper containing molecules. This adsorption to the surface occurs during desolvation (ie. the drying cycle). At masses greater than 40 ng polymetallic fragment ions are observed thereby indicating that Cu exists as microdroplets or some other form of surface aggregates at room temperature. Silver however shows evidence of aggregates throughout the concen- tration range studied. This work was supported in part by the National Science Foundation grant No. CHE9020591. - 5 - 300 400 500 600 700 Temperature/K Fig.8 Positive and negative TPS-SIMS scans of 2000ng of Cu deposited on a pyrolytic graphite coated graphite platform as the hydrated nitrate copper nitrate salt to CuO(s).It is possible that the CuO+ SIMS signal shown in Fig. 8 originates from fragmentation of the basic nitrate crystal on the surface by the primary-ion beam. This is supported by the negative SIMS ion spectra for CuO-. Fig. 8 shows the signal for CuO- increasing at tempera- tures where the CuO' signal decreases. This change near 550 K is located at the temperature expected for the conversion of the basic nitrate salt to C U O ( S ) . ~ ~ ~ ~ This is reasonable since surface CuO would be more likely to form a negative ion during sputtering and hence an increased CuO - signal. The peak for the CuO- signal centered around 330 K corresponds to the temperature of dehydration of the hydrated copper nitrate reported by Taylor et As noted in Fig.6 at the sample masses where there is evidence for surface aggregates the TPS-SIMS scan shows a sharp drop corresponding to the decomposition of basic nitrate salt this sharp decrease in signal is absent for masses where the sample appears to be dispersed. This indicates that either the dispersed sample exists as dispersed atoms possibly bound to the surface through a Cu-0-surface bond or that for the dispersed samples the bulk characteristics of the basic nitrate decomposition are lost. Conclusion Metals which exist as microdroplets give rise to sputtered ions containing multiple metal atoms. The probability of sputtering polymetallic secondary ions is low when the analyte exists as 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 References Arthur J. R. and Cho A. Y. Surface Sci. 1973 36 641. McNally J. and Holcombe J. A. Anal. Chem. 1987 59 1105. Lynch S. Sturgeon R. E. Luong V. T. and Littlejohn D. J. Anal. At. Spectrom. 1990 5 311. Fonseca R. W. McNally J. and Holcombe J. A. Spectrochim. Acta Part B 1993 48 79. Redhead P. A. Vucuum 1962 12 203. Rojas D. and Olivares W. Spectrochim. Acta Part B 1992 47 387. Yan X.-p. Ni Z.-m. Yang X.-t. and Hong G.-q. Spectrochim. Acta Part B 1992 48 605. Ganz J. Sattler K. and Clarke J. 1989 Surface Sci 219 33. Schleicher B. Jung Th. Hug H. and Burtscher H. 2. Phys. D. 1991 19 327. Mason M. G. Phys. Rev. 1983 B27 748. Eigelhoff W. F. Jr. and Tibbets G. G. Phys. Rev. 1979 B19 5028. Hassell D. C. Majidi V. and Holcombe J. A. J. Anal. At. Spectrom. 1991 6 105. Fonseca R. W. McNally J. and Holcombe J. A Spectrochim. Acta Part B 1993 48 79. Eloi C. Robertson J. D. and Majidi V. J. Anal. At. Spectrom. 1993 8 217. Ghose J. and Kanungo A. J. Thermal Anal. 1981 20 459. Taylor T. J. Dollimore D. and Gamlen G. A. Thermochim. Acta 1986 103 333. Paper 3105841 G Received September 28 1993 Accepted December 14 1993