402
Anal. Chem. 1985, 57,402-406
Fourier Transform Ion Mobility Spectrometry
F. J. Knorr, R. L. Eatherton, W. F. Siems, and H. H. Hill, Jr.*
Department of Chemistry, Washington State University, Pullman, Washington 99164
A mobility Interferogram Is generated when the entrance and
exit gates of a two-gate ion moblllty spectrometer are simultaneously opened and closed by a frequency sweeping
square wave generator. Fourier transformatlon of the Interferogram recovers the normal moblllty spectrum. A theory
for the method is developed, and transformed spectra are
obtained for the posltlve background reactant Ions and for
l-methylnaphthalene In N, drlft gas. Potential benefHs of the
method for moblllty detection In gas chromatography include
rapid collection of complete mobility spectra and Increased
detector sensltlvlty. The new method should be generally
applicable to other tlme dlsperslve technlques such as tlme
of flight mass spectrometry.
Ion mobility spectrometry (IMS), also known as plasma
chromatography, separates atmospheric pressure gas phase
ionic species on the basis of their mobilities ( 1 ) . The importance of IMS for analytical chemistry derives from its
potential as a versatile and ultrasensitive detector for trace
organics (2). Since the first use of IMS for gas chromatographic detection (IMD) (3), emphasis has shifted from
qualitative analysis via complete mobility scans of peaks to
quantitative determinations using various selective and nonselective continuous mobility monitoring modes (4). Now that
design changes in the IMD have made it compatible with
high-resolution capillary columns (5), even the fastest IMS
data collection technique is too slow to obtain a complete
mobility spectrum of a single chromatographic peak.
Currently there are three methods of recording ion mobility
spectra (2)--single scan, signal averaging, and moving second
gate. In all three methods an entrance gate is pulsed open
for a short time admitting ions to the drift region. In the single
scan method the mobility spectrum, typically a 20-ms wave
form, is monitored directly with an oscilloscope. Noise levels
are severe and the method is unusable with high-resolution
chromatographic separations. In the signal averaging method
many single scans are summed with a computer. Generally
500-1000 repetitions are required for acceptable S I N levels,
so 10-20 s are needed to generate an ion mobility spectrum.
This is an order of magnitude too slow for high-resolution
chromatography. In the moving second gate method an exit
gate, located a t the end of the drift region, is repeatedly pulsed
open with a particular phase delay relative to the entrance
gate. An electrometer with its time constant set too long to
track high-frequency noise or individual ion pulses, measures
the average ion current. An ion mobility spectrum is generated
by slowly sweeping the phase delay of the exit gate. In general
it is difficult to obtain moving second gate spectra with acceptable S I N in less than 1 or 2 min.
None of these three methods make efficient use of the
available ions. With typical entrance pulse durations of -0.2
ms, only about 1%of the available ions can contribute to the
mobility spectrum. Furthermore, in the moving second gate
method with exit gate durations also -0.2 ms, on the average
only about 1% of the ions which pass the entrance gate
contribute to the mobility spectrum. Opening the gates for
longer periods of time increases the signal but lowers the
0003-2700/85/0357-0402$01.50/0
resolution of the mobility spectrum.
In this study an alternative Fourier transform (FT) operating mode for a two-gate ion mobility spectrometer is presented. The potential of this method, in which 25% of the
available ions contribute to the measured signal, was evaluated
with respect to sensitivity, scan time, and resolution. The
minimum scan time so far achieved with the F T mode is 10
-
S.
THEORY
A two-gate ion mobility spectrometer (5, 6 ) modified to
operate in the FT mode is shown schematically in Figure 1.
The drift tube assembly is unchanged from the normal instrument. The F T instrument differs from the normal
spectrometer in four ways: (1)the gating signal generator
produces a binary (on, off) square wave rather than a train
of narrow pulses; (2) the entrance and exit gates are always
driven simultaneously by the same square wave, with zero
phase delay; (3) the scanning parameter is the square wave
frequency rather than phase delay; and (4)the F T IMS interferogram is recorded with a microcomputer which also
performs Fourier transformation of the data to recover the
normal ion mobility spectrum.
Gating Correlation Function. In both normal and FT
mode operations of a two-gate spectrometer, a particular gate
timing sequence is repeated many times before the scanning
parameter is changed. The time constant of the damped
picoammeter is too long to follow individual ion pulses or
high-frequency noise. Its output represents the time averaged
dc ion current at the present value of the scanning parameter.
The gates themselves act to filter the ions streaming through
the drift tube. Depending on the characteristic transit times
of the different ions present and the actual gate timing sequence, some ions will reach the detector with maximum
intensity, some with intermediate intensity, and some will not
reach the detector a t all. The filtering and time averaging
may be represented by a gating correlation function
where e ( t 3 is the entrance gate function, f(t? is the exit gate
function, and T is the time constant of the detection electronics. The gate functions e(t? and f ( t ?represent the on-off
action of the gates in real time, whereas the domain of t(t)
is ion transit time. t ( t ) is a correlation function in the
mathematical sense of the term (7).
A value of t(t) represents the fraction of ions having transit
time t that reach the detector. If a sample having the ion
mobility spectrum m(t)is presented a t the entrance gate, the
detected signal is
S = l m ( t )t(t) dt
That is, the intensity of ions with transit time t , rn(t),is
multiplied by the fraction of these ions which reach the detector and summed over all transit times to yield the detector
signal.
In both the normal and FT modes the gate functions have
a characteristic periodicity (frequency u , period T = u-*) and
0 1985 American
Chemical Society
-
Dr f t
ANALYTICAL CHEMISTRY, VOL. 57, NO. 2, FEBRUARY 1985
T m t
0;
53-wi
Control
=
1
_ _ _ - _
Electronics
and
403
F C’I
Ertrance Gat?
Gate
Driver
Ex t Gate
~
Time
E ectrorneler
F a s t ADC S t o r a g e Scope
Transit Time
Figure 3. (a)Typical entrance and exit gate functions, FT mode. (b)
Typical gating correlation function, FT mode.
Figure 1. Schematic diagram of the FT IMS apparatus.
7
-
height. On the other hand, if m ( t ) is an actual diffusion
broadened spectrum, in the limit as w o 0
a
t(t, At)
0
-
wo
r [6(t - A t )
+ 6 ( t - (At +
Time
Transit Time
Figure 2. (a) Typical entrance and exit gate functions, normal mode.
(b) Typical gating correlation function, normal mode.
a characteristic phase delay ( A t ) . In a normal scan frequency
is held constant and phase delay is swept, while in an FT scan
phase delay is held at zero while frequency is swept. In both
cases the scanning parameter generates a family of gating
correlation functions. In what follows t(t) will be labeled by
the parameter of which it is a function.
Normal Mode. Figure 2a shows typical gate functions for
the normal mode and shows how they are characterized by
a period T , phase delay A t , and gate open time wg. Figure 2b
shows the corresponding correlation function t(t, At). Although only the positive part of the function is shown, strictly
it extends into the negative transit time region with the same
periodicity. The peaks in &, A t ) are triangular as is expected
for the correlation of rectangular pulses. As an example of
how to interpret ~ ( tA,t ) , the cutoff point on the low transit
time side of the fiist peak is associated with fast ions that pass
through the entrance gate just as it closes and pass through
,
the exit gate just as it opens. The maximum values of ~ ( tAt)
are equal to w o / T , the fraction of time that the entrance gate
is open.
The signal obtained by scanning At may be written
S(At)= J m ( t )
t(t, At) d t
(3)
Peaks in beyond the first do not contribute to the measured
signal unless some ion in the sample has a transit time longer
than the gate period 7,a situation which is generally avoided.
In the absence of broadening by diffusion and other nongate
effects, m(t)is a series of 6 functions and the signal predicted
by (3) is a series of triangular peaks, all of width woat half
+
6(t - ( A t -t 27)) +
s(At)
I
T))
WO
+
-m(At)
...I
(4)
(5)
T
FT Mode. Figure 3 illustrates the simultaneous square
wave entrance and exit gate functions. Figure 3b shows the
positive half of the corresponding gating correlation function.
Since the gate functions are equal, t(t, u ) is an autocorrelation
function, an even function with a negative transit time half
mirroring the positive half shown. The maximum value of
t(t, v ) is 0.5, the fraction of time the entrance gate is open.
Again the correlation function represents the filtering action
of the gates. With a gate modulation frequency u , the only
ions that reach the detector with full intensity have transit
times 0, l / v , 2 / v , 3/u, ... . The only ions which do not reach
the detector a t all have transit times 1 / 2 u , 3 / 2 u , 5 / 2 v , ... .
The signal obtained by scanning the square wave frequency
is
S(u) = J m ( t )
4,
u ) dt
(6)
where the integral is over all drift times. To imagine the
general form to expect for S(u),suppose the mobility spectrum
consisted of a single 6 function of intensity 1, at transit time
t. As v is increased, t(t, v ) remains a triangle wave, but collapses accordion fashion toward the origin. Then S(v)is itself
a triangle wave with maxima and minima
Y = 0, l / t , 2 / t , 3 / t , ...
s(u),,, = 0.51,
S(V),~,= 0
u = 1/2t, 3/2t, 5/2t,
...
(7)
If m(t)is a sum of 6 functions, S(u) will be a sum of interfering
triangle waves. It should be anticipated at this point that in
real situations these triangle waves will not continue to oscillate between extremes of 0 and 0.51, as u increases. Diffusional broadening of the ion pulses will subtract intensity
from the maxima and add it to the minima until at some
frequency S(v) will be indistinguishable from a steady dc level
Of
0.251,.
F o u r i e r Transformation. The form of eq 6 suggests that
S(v) and m ( t ) are “+transforms” of one another. However,
since fast FT programs are readily available, we shall consider
Fourier transformation of S(u).
Since t(t, u ) is the autocorrelation of a periodic function,
it is itself an even periodic function with period l/v. Therefore
~ ( tu ), may be expanded in a Fourier series to which only the
404
ANALYTICAL CHEMISTRY, VOL.
57, NO. 2, FEBRUARY 1985
cosine terms make nonzero contributions. The expansion
coefficients are found t o be
A , = 1/4; An = 0 ( n = 2, 4,6, ...)
2
A , = - (TI = 1, 3, 5 , ...)
(8)
x2n2
and the FT IMS signal may be written
2
S(v) = L j m m ( t ) d +
t 4
L J m m ( t ) cos 2xnvt d t
a2n=1,3.5.
...n2
0
(9)
’t
Identifying Io = J m ( t )d t and changing variables in the higher
cosine terms give
1
4
S(v) = -I,
2
+c L j m m ( t / n )cos 2avt d t
a2n=1,3,5... n3 0
=
1
4
-I,
1 1
+ -E--F,m(t/n)
n3
x2
1
-10
4
2
6(t)+ -
c
1
-m&/n)
(13)
-
=
e-idvl+uz)t
[F8(v)sinc ((v2 - vl)t)]
(14)
Thus the transform is shifted inphase and broadened by
convolution with the sinc function. The experimental trace
we finally extract is the positive part of the magnitude of the
transform
1
IFS(v),,,I+ = -m(t)* b
x*
1
+ ---m(t/3)*
27x2
b
5
6
must be recovered from the interferogram.
As in previous work the detector signal was amplified with a
Keithley 417 high-speed picoammeter (Keithley Instruments,
Cleveland, OH). In both normal and F T mode experiments the
damping control of the instrument was set to give a bandwidth
of -7 Hz. Data from both the normal and FT modes were
collected with an Apple 2e microcomputer equipped with an
Applescope data collection system (RC Electronics, Santa Barbara,
CA). Essentially, the Applescope is a microcomputer emulation
of a digital signal analyzer incorporating a modified tracking type
8 bit ADC with a 3.5-MHz maximum sampling rate. The Fourier
transforms were obtained by using the SDL-001 Spectrum
Analysis Software option (RC Electronics, Santa Barbara, CA).
Interferograms were collected as 1024 points, the maximum
number of points possible to use with the spectrum analysis
software.
Ion Mobility Studies. In order to compare the performance
of the two data collection techniques, ion mobility spectra were
obtained for the background reactant ions and l-methylnaphthalene using both the normal time dispersive and the FTtechniques. A stainless steel sample holder containing the 1methylnaphthalene was attached to the branch of a
in. “tee”
fitting inside the oven of a gas chromatograph. The tee fitting
was connected to an unused injection port which supplied a flow
of nitrogen gas. The nitrogen passed through the fitting, sweeping
1-methylnaphthalenevapors into a short length of noncoated fused
silica transfer line used to connect the sampling apparatus to the
ion mobility spectrometer. The amount of compound entering
the IMS was controlled by setting the nitrogen head pressure and
temperature of the GC oven. A concentration on the order of 1
ppm 1-methylnaphthalene was used.
Operating parameters of the IMS common to all experiments
were as follows: ion drift length, 7.60 cm; electric field gradient,
239 V/cm; gate voltage, A30 V; temperature, 150 “C; pressure
691.2 torr (702.3 torr for 1-methylnaphthalene study). Drift and
makeup gases were both prepurified nitrogen (Liquid Air, Inc.,
San Francisco, CA). Gas flow rates were drift gas 600 mL/min
and makeup gas 20 mL/min. 1-Methylnaphthalene (Chem
Service, Inc., West Chester, PA) was used “neat”for the continuous
bleed experiment.
where II is a rectangle function of value unity in the scan range
v1
v2 and zero outside it. Before transforn_lation the dc level
~ ~ .the
is subtracted, leaving the ac component S ( U ) ~Since
tranform of 11 is a sinc function (7) the FT of eq 13 is taken
&(v)erp
L
IkHzl
(11)
where mE(t) is the even part of m(t).
Experimentally, all of S(v) is not measured. In practice
scans begin at a few tens of a hertz and extend into the kilohertz range. The actual measured signal is
s(v),,p
= S(v) Wl, v2)
3
Frequency
Figure 4.
(12)
x 2n=1,3,5 ...n3
2
(10)
where F, stands for the cosine transform. The FT IMS signal
is nearly equal to the 0.251, dc level plus the cosine transform
of the mobility spectrum. However, since ~ ( tv ), is in fact a
triangle wave, the signal includes minor contributions from
odd overtones of the ion mobility spectrum. The n = 3 term
has 1/27ththe intensity of the n = 1 term. Fourier transformation of eq 11 yields
FSb) =
1
+ ... (15)
where b = Jsinc ( v 2 - u l ) t J .
EXPERIMENTAL SECTION
Modification of the Ion Mobility Spectrometer. Design
and construction of the ion mobility spectrometer with a 63Ni
ionization source have been previously reported (5,6). This design
was used to obtain results in the normal mode. For the Fourier
transform mode, the AIM 65 microcomputer (Rockwell International, Anaheim, CA) was replaced by a H P 3325A synthesizer/function generator (Hewlett-Packard, Palo Alto, CA). It was
operated in its square wave mode a t an output level of 5 V peak
to peak with a 2.5-V offset to create the proper logic levels required
to trigger the gate driver. The output was connected to both
inputs of the gate driver in order to drive both gates at the same
frequency. The frequency scans were accomplished by using the
function generator in its single linear sweep mode. The initial
frequency of 10 Hz was chosen to lie outside the bandwidth of
the electrometer. The frequency range was selected to produce
less than 0.2-ms peak broadening. Square wave scan rate R (Hz
s?) were chosen to be less than f ~ / t - ,where f~ is the electrometer
bandwidth in Hz and t , is the longest transit time (0.020 s) which
Mobility interferogram, positive background reactant ions:
square wave scan range, 10-6000 Hz; total scan time, 36 s.
RESULTS AND DISCUSSION
Interferograms and Transforms. Figure 4 shows a
typical mobility interferogram, in this case for the positive
background reactant ions formed in the N, drift gas. The
triangular wave form conforms to the predictions of eq 6 and
7 . As anticipated the wave form is damped a t high frequencies
because of diffusional broadening of the ion packets. If the
IMS spectrum m ( t ) is taken to be a 6 function spectrum
broadened by convolution with some function d ( t ) that describes the diffusion, then in the transform domain S(u) is
modulated by multiplication with D ( v ) ,the transform of d ( t ) .
In particular, if m ( t )is convoluted with a Gaussian function
of width u = wd, then S ( V )will be damped by a Gaussian
envelope of width u = 1 / t u d . For the interferogram of Figure
ANALYTICAL CHEMISTRY, VOL. 57,NO. 2, FEBRUARY 1985
405
I
5
5
15
10
D r i f t Time
Figure 7. FT mode spectrum,
L
._..
--.-
5
15
10
Drift Time
20
(rnsl
Figure 6. FT mode spectrum, positive background reactant ions:
obtained by transformation of interferogram in Figure 4;reduced mobilities, 2.93,2.80,2.54 cm2 V-l s-'.
4 the difference between the maxima and minima has decreased by a factor of 2 at about 2.0 kHz, so we can estimate
0.25 ms.
the diffusional broadening as t d
Figure 5 is a normal mode mobility spectrum of the positive
background reactant ions. Noise is evident in the normal
mode scan because the signal levels are lower than those in
Figure 4. The scan in Figure 5 is about as rapid as possible
with the normal mode. The bandwidth of the electrometer
is such that faster scans would begin to distort and broaden
the peaks. Increasing the bandwidth would increase the noise
levels and swamp the minor peaks.
Figure 6 is the result of transforming the interferogram of
Figure 4. Aside from the broad band at 0-1 ms, it is equivalent
to the normal mode spectrum. Transit times of the peaks
correspond and the minor peaks are well-resolved. The 0-1
ms band corresponds to a slow decrease in the dc level of the
interferogram. We attribute this to a "gate depletion
effect"-a lower concentration of ions just outside the entrance
gate due to the high electric field of the gate. At low square
wave frequencies, sampling extends deep into the reaction
region, but a t high frequencies the depleted region supplies
a larger proportion of the sample. This interpretation was
supported by experiments in which the gate voltage was
varied. When the differential voltage on the gate wires is
increased to f45 V, the drop in dc level at high frequencies
becomes much more noticeable and the low drift time band
in the interferogram increases in intensity.
In normal mode mobility spectra peaks are broadened by
ion diffusion and by the gate width w,,.To increase resolution
to the diffusion controlled limit, womust be decreased with
a proportional loss in signal strength (see eq 5 ) . Thus in the
normal mode there is a trade-off between resolution and
sensitivity. The FT mode spectrum is affected by diffusion
broadening to the same extent as the normal mode. However
the instrumental contribution to the peak broadening arises
not from the gate width but from the frequency range scanned
(see eq 15). The diffusion controlled limit is approached by
-
15
20
D r i f t Time lrnsl
Ims)
Figure 5. Normal mode spectrum, positive background reactant ions:
total scan time, 54 s; T = 20 ms; w o = 0.15 ms; reduced mobilities
2.95,2.79,2.55 cm2 V-' s-I.
\--
10
20
-
1 ppm 1-methylnaphthalene,in N,
drift gas: square wave scan range, 10-6000 Hz; total scan time, 36
s; reduced mobility of major product ion, 1.99 cm2 V-' s-'.
scanning a wider range of frequencies, which may be done
without loss of signal. Thus an advantage of the F T method
is that there is no trade-off between resolution and sensitivity.
Comparing the half height widths of the major peak in Figures
5 and 6 shows the two to be indistinguishable. This is as
expected since wo= 0.15 ms and ( u 2 - ul)-l = 0.17 ms.
Equation 15 predicts a satellite spectrum of 1/27threlative
intensity with transit times corresponding to the major
spectrum multiplied by three. Thus we expect to see a small
peak at about 22 ms. The full transform of Figure 4,which
has information about drift times out to -42 ms, does have
a small peak at 22 ms with intensity about that predicted by
the theory.
Figure 7 is the FT mode mobility spectrum obtained by
continuously bleeding 1ppm 1-methylnaphthalene into the
IMS apparatus. It is equivalent in transit times, peak widths,
and relative intensities to the normal mode spectrum. Thus
it appears that the FT mode is capable of recovering complex
mobility spectra.
Rapid Scanning. An 18 s scan, with all other parameters
including the -7 Hz electrometer bandwidth equal to those
for Figure 4,yielded a mobility spectrum virtually identical
with that of Figure 6. This is about as rapid a scan as possible
with the 7-Hz bandwidth. With faster scans the electrometer
is too slow to track the interferogram component of an ion
of 0.020 s transit time. If f E is the electrometer bandwith in
Hz and R is the square wave scan rate in Hz s-l, then
-
where t,,, is the longest transit time that can be accurately
recovered by Fourier transformation of the inteferogram.
Using bandwiths wider than 7 Hz, mobility spectra of positive
background reactant ions have been obtained with scan times
under 10 s. The transformed spectra have lower intensity than
those in Figure 6, but the transit times and peak widths are
unchanged. T o obtain mobility spectra of high-resolution
chromatographic peaks, scan rates perhaps as high as 10 kHz
s-l and an electrometer bandwidth of 200 Hz would be needed.
The potential for rapid scanning of high-resolution chromatographic peaks via the the F T method arises from the
signal levels obtained and the multiplex advantage. The
A signal levels are high enough to reduce concern about
noise problems and amplifier speed. Also, since the interferogram is a multiplex signal, changing sample concentration
in the IMD should result in a low-frequency modulation of
the interferogram but not distortion of relative intensities of
peaks in the spectrum. On the other hand, the multiplexing
may result in the loss of low-intensity peaks in the transform,
a problem common to other FT methods. The recovery of
minor peaks in Figures 6 and 7 is encouraging in this regard.
With scan rates as high as 10 kHz s-l an assumption implicit
in our interpretation of eq 1 breaks down. The assumption
is that during the time it takes ions to traverse the drift region
-
406
Anal. Chem. 1985, 57. 406-411
the square wave frequency does not change appreciably. This
condition is well met in the work reported here, but it would
not be correct a t all with very fast scans. Preliminary experimental and theoretical work indicates that in very fast
scans individual interferogram components lose intensity and
are shifted in phase by amounts determined by transit time
and scan rate, but that the FT method is still valid. However,
further theoretical and experimental investigations are needed
before very rapid scanning can be employed.
Other Applications. Although this work centers on FT
IMS, the same general approach is applicable to any time
dispersive technique. A time dispersive technique separates
the components of a mixture on the basis of propagation times
of the individual components through some medium. In addition to IMS, time dispersive techniques include chromatography, electrophoresis, time of flight mass spectrometry,
TOF spectrophotometry, RADAR, and LIDAR. TOF mass
spectrometry in particular could benefit from the FT method.
In any appication the essential elements of the FT method
are (1)modulation of the source (or injector or transmitter)
by a continuous frequency rather than a narrow pulse, (2)
multiplication of the detected signal by the source modulation
function and long time constant averaging of the product
function, (3) obtaining a signal S(v) by sweeping the modulation frequency, and (4) recovering the time dispersed signal
as normally measured by calculating the magnitude of the FT
of S(v). In a particular application the potential benefits of
the FT method would arise from one or more of the following:
(1) increased duty cycle of the sample source, which could be
used to increase signal levels, increase sensitivity, or speed
data collection, (2) reduction of measurement broadening
without loss of sensitivity, (3) multiplexing, since signals from
all components of a sample are detected simultaneously, or
(4) reduction of need for fast detection and recording electronics, since the recording electronics need be at most only
as fast as the sweep of the modulation signal.
ACKNOWLEDGMENT
We wish to express our appreciation to Steven D. Brown
and John M. Frame for helpful discussions.
LITERATURE CITED
Cohen. M. J.; Karasek, F. W. J . Chromatogr. Sci. 1970, 8 , 330.
Hill, H. H.. Jr.; Baim, M. A. I n "Plasma Chromatography"; Carr, T. W.,
Ed.; Plenum: New York, 1984; p 143.
Karasek, F. W.; Keller. R. A. J . Chromatogr. Sci. 1972, 10, 626.
Karasek, F. W.; Hill. H. H., Jr.; Kim, S. H.; Rokushika, S. J. J . Chromatogr. 1977, 135. 329.
Baim, M. A.; Hill, H. H., Jr. Anal. Chem. 1982, 5 4 , 38.
Baim, M. A,; Eatherton, R. L.; Hill, H. H., Jr. Anal. Chem. 1983, 55,
1761.
Bracewell, R. N. "The Fourier Transform and Its Applications", 2nd
ed.; McGraw-Hill: New York, 1978; Chapters 3 and 4.
RECEIVED
for review July 23,1984. Accepted October 19,1984.
This work was supported in part by a grant from the Public
Health Service. The work was presented in part at the
Northwest Regional Meeting, American Chemical Society,
June 8. 1984.
Tree Ring Wood Analysis after Hydrogen Peroxide Pressure
Decomposition with Inductively Coupled Plasma Atomic
Emission Spectrometry and Electrothermal Vaporization
Henryk Matusiewicz' and Ramon M. Barnes*
Department of Chemistry, University of Massachusetts, Amherst, Massachusetts 01003-0035
A method utlllzlng pressure decompodtlon to mlnhnlze sample
pretreatment Is described for the lnductlvely coupled plasma
atomic emlsslon spectrometric analysis of red spruce and
sugar maple. Cores collected from trees growlng on Camels
Hump Mountain, Vermont, were dlvided Into decade lncrements In order to monltor the temporal changes In concentrations of 21 elements. Dried wood samples were decomposed In a bomb made of Teflon wiih 50% hydrogen peroxlde
heated In an oven at 125 OC for 4 h. The dlgestlon permitted
use of aqueous standards and mlnlmlzed any potentlal matrlx
effects. The element concentrations were obtained sequentlally by electrothermal vaporlratlon ICP-AES uslng 5 pL
sample allquots. The method preclslon varied between 3 and
12%. Elements formlng oxyanlons (AI, As, Fe, Ge, Mn, SI,
V) were found at elevated concentratlons durlng the most
recent three decades, whlle other metal (e.g., Mg, Zn) concentrations were unchanged or decreased.
Many studies have been carried out to investigate the uptake of various elements by plants, especially in forest ecoOn leave from Technical University of Poznafi, Department of
Analytical Chemistry, 60-965 Poznaii, Poland.
0003-2700/85/0357-0406$01 S O / O
systems (1). Tree rings, which represent a chronological record
of elemental changes that are not contaminated naturally by
outside sources like lake-core sediments or glacial cores, can
be considered indicators that record environmental disturbances. The occurrence of elevated trace and especially toxic
metals concentrations in accurately dated tree-ring sequences
from trees in certain regions is closely linked to environmental
effects (e.g., acid rain), and these rings represent records of
environmental influence during the past several years ( 2 ) .
Thus, analysis of tree rings for metals is an important indicator
of atmospheric pollution. Some effort has been devoted recently to detecting the presence of heavy metals in wood
(3-16).
A new method of sample preparation for cellulose materials
was sought in this research that would be applicable to wood
and yield a solution from which elements could be determined
by inductively coupled plasma atomic emission spectrometry
(ICP-AES). Bomb digestion with hydrogen peroxide was
evaluated because it provides a relatively rapid means of
decomposition that assures complete recovery of elements in
tree ring wood.
Analysis for trace and major elements in solid samples
generally requires decomposition of the organic matter followed by dissolution leading to a solution for subsequent
analytical determination. This ashing may be achieved by
0 1985 American Chemical Society