Anal. Chem.
1982, 54, 499-503
the Cd net concentration in the lake samples. If one assumes
that contamination affected individual samples and blanks
equally on average, then most of the irreproducibility in the
MSID determinations is attributable to variability in the
extent of contamination.
CONCLUSIONS
The ED-MSID combination proved to be a suitable method
for determining Cd at very low concentrations in freshwater.
Once the proper precautions are taken, the procedure yields
a concentrate that is free of significant interferences and which
gives steady ion currents from samples and blanks containing
25-100 ng total Cd. In practice the lower working limit is
determined by the blank value, rather than inherent sensi-
tivity. And it is the variability introduced by contamination
that appears to controll the overall precision.
In our experience, tlhe most persistent and troublesome
source of this contamination is Cd-laden particulate matter
in the laboratory and shipboard atmospheres. Airborne Cd
often reaches several micrograms per cubic meter around
urban-industrial centers (14) and can be higher in dusty en-
vironments or where tobacco is being smoked. These particles
readily contaminate not only open liquid samples but also the
surfaces of containers and reagents exposed to unfiltered air.
Much of the associated Cd is found near the surface of these
particles (15) and is readily available for solubilization.
Our use of closed conlahers and clean air work stations was
effective in dealing with these problems. In some control
samples, Cd was undetectable even by ED/MSID, indicating
that further experience with the technique will lead to even
lower mean blank values. However, other reports ( 1 , Z ) in-
dicate that clean room conditions may still be required if metal
concentrations below 10 ng L-l are to be studied routinely.
Unrecognized difficulties in controlling contamination
and/or lack of sensitivity appear to have plagued earlier efforts
Direct Chemical Ionization Mass Spectrometry with
Polyimide-Coated Wires
V. N. Reinhold” and !Steven A. Carr
Department of Biological Chemistry and Laboratory of Human Reproductlon and Reproductive B/ology, Harvard Medical School,
Boston, Massachusetts O.? 115
A procedure is descrlbed for improving the results obtalned
in the analysts of thermally labile organic compounds by direct
chemlcal Ionization (DCI)mass spectrometry. The nichrome
wire sample support is coated wlth a polyimide material
(Pyre-ML) In a two-step procedure uslng an eiectrlcal current
to induce polymeriratlon. Comparison of the DCI mass
spectra of numerous bioioglcaliy Important compounds (ln-
cluding oilgosaccharides and peptldes) obtained with poly-
imide-coated wire vs. bare metal Indicates a dlminutlon in
pyrolysis and a correspondlng Increase In the intensity of ions
related to molecular welght for samples desorbed from the
coated wlres. Avolding metal-sample interactton by coatlng
the wire with a polylmide surface extends the DCI approach
to compounds of greater polarity and higher molecular weight.
A major goal of mass spectrometric research is to develop
ionization methods which will enable the determination of
0003-2700/82/0354-0499$0 1. W O
499
to determine Cd coincentrationsin the Great Lakes. After a
critical review of the historical data, Rossman (16) concluded
that none was reliable. For Lake Michigan, the Cd data
reported here are the first to be gathered with sufficient at-
tention to contamination control, using independent methods
of sufficient sensitivity, to provide a glimpse of the actual
present concentrations. In a separate report (17), we have
shown how these rrneasurements are crucial to arriving at a
budget for Cd in Lake Michigan and to establishing the rate
of its accumulation in the future.
LITERATURE CITED
(1) Settle, D. M.; Patterson, C. C. Science 1980, 207, 1167-1176.
(2) Rosman, K. J. R.; DeLaeter, J. R. J. R . SOC.West. Aust. 1977, 59,
91-96,
(3) Batley, 0. E.; Malousek, J. P, Anal. Chem. 1977, 49, 2031-2034.
(4) Torsi, 0. Ann. ChJm. (Rome) 1977, 67, 557-566.
(5) Brandenberger, H.; Bader, H. At. Absorpt. Newsl. 1987, 8 , 101-104.
(6) Falrless, C.; Bard, A. J. Anal. Left. 1972, 5433-437.
(7) Fairless, C.; Bard, A. J. Anal. Chem. 1973, 45, 2289-2291.
(8) Lund, W.; Larsen, B. V. Anal. Chim. Acta 1974, 72, 57-64.
(9) Catanzaro, E. J. ,/.-Water Pollut. Control Fed. 1975, 47, 203-204.
(IO) Elzerman, A. W.; Armstrong, D. E; Andren, A. W. Environ. Sci. Tech-
no/. 1979, 13, 720-725.
(11) Allen, H. E; NOH, K E.; Jamljun, 0.; Boonlayangoor, C. Radlological and
Environmental Research Dlvision Annual Report, Jan-Dec 1977, Ar-
gonne National Laboratory, ANL-77-65. Part 111, pp 70-73.
(12) Cameron, A. E.; Smith, D. H.; Walker, R. L. Anal. Chem. 1989, 4 7 ,
525-529.
(13) Bickford, M. E.; Sllka, L R.; Shuster, R. D.; Anglno, E.;Ragsdale, C. R.
Anal. Chem. 1978, 50, 469-472.
(14) Thompson, R. Adv. Chem. 1979, 772, 54.
(15) Natusch, D. F. S. Environ. Health Perspect. 1978, 22, 79.
(16) Rossman, R. Absiracts of the 24th Conferance of Great Lakes Re-
search, 28-30 April 1981, Columbus, OH, p 34.
(17) Muhlbaler, J.; Tisue, G. T. Water, Air, Soil Pollut. 1981, 15. 45-59.
RECEIVED for review October 7,1981. Accepted December 8,
1981. A preliminary account of this work was presented at
the 2nd Midwest Water Chemistry Workshop, Minneapolis,
MN, 8-9 Oct 1979.
highly polar and thermally labile compounds with a minimum
of decomposition. One approach has been to rapidly heat
nonvolatile samples coated on an extended direct insertiton
probe within a chemical ionization reagent gas plasma. This
technique, known as direct chemical ionization (DCI), has
evolved through the contributions of a number of research
groups into a useful alternative approach for obtaining
structural information on thermally labile samples. For a
recent review of this topic see ref 1.
In contrast to conventional solids probe electron impact or
chemical ionization mass spectrometry, the spectra of polar
compounds obtained by direct sample insertion into the
reagent gas plasma contain (and are often dominated by) ions
indicative of the molecular weight as well as structurallly
significant fragment ions. The DCI technique is more sensitive
than field desorption (FD) mass spectrometry, and invariably
more fragmentation is observed, if only, in some cases, a
spectrum of the products of pyrolysis. However, a number
of involatile compounds fail to exhibit ions diagnostic of
0 1982 American Chemical Society
500 ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982

8000
I 200
Total Ion
,:'"',.. . /Current 175
I .

6000 - 150

5000 - I25
I=(342+360) R =
4000 - IO0

3000 - 75

2000 c 50

, I , , I 25

@,r,*W--fl

0 0
50 55 60 65
Scan
Figure 1. Total and selected ion current profiles obtained by DCI of kojibiose from uncoated nichrome wire. Profile I = C [ m / r342 m / z
3601 ([M - H20) + +
NH,]' and (M NH,)', respectively) which are ions related to the intact or nearly intact molecule. Curve R is the ratio of
pyrolysis ions m / z 180 and m l z 198 to I as a function of heatlng current and mass spectral scan number. The wire was heated linearly at 2
mA/s.
molecular weight by DCI but do by FD.
Sample analysis by DCI is most frequently initiated by a
programmed heating current through a resistive wire (such
as rhenium or nichrome) on which the sample has been coated.
Since heating and sample-surface interactions are involved,
pyrolysis is a constant problem and the DCI technique must
contend with this fact. As would be anticipated, pyrolytic
products are enhanced over molecular weight related ion
products as the sample increases in molecular weight, de-
creases in amount, and possesses greater ionic character. Thus,
obtaining molecular weight information on nanogram amounta
of large thermally labile compounds by DCI can be exceedingly
difficult.
We report here the use of polyimide-coated (PIC) wires,
a technique which extends the limits of the DCI approach.
EXPERIMENTAL SECTION
Polyimide Wire Coating. A 5 cm length of nichrome wire
(0.0125 in. diameter) bent into a "U" shape is attached to a
variable ac power supply and the voltage is adjusted to obtain
between 0.7 to 0.9 A flowing in the circuit (- 1 V ac). The wire
is then immersed briefly in polyimide primer (primer and topcoat
are available from American Durafilm Co., Inc., Newton Lower
Falls, MA, under the trade name Pyre-ML) diluted 1:l with
N-methylpprolidone. Upon removal of the wire from the primer
the current initiates polymerization (5-10 8); droplets which may
form on the wire are removed by touching with a glass rod. This
step is repeated. the polymerized primer covered wire is then
dipped into a topcoat solution and treated identically. Fully
coated wires are attached to the DCI probe, inserted into the CI
chamber, and programmed slowly through a heating cycle to
remove interfering background. Although a relatively large
amount of material desorbs during this initial "burn-in", we have
experienced no loss in sensitivity or increase in the rate of ion
source contamination. To initiate desorption a linear programmed
heating current (mA/s) was started coincident with repetitive
scanning. Current applied to the PIC wire during an analysis
should not exceed that which is sufficient for sample desorption,
since excess heating shortens the usable life of the polyimide
coating. The coated wires are reusable (up to 10 different samples)
with no problem of cross-contamination.
Instrumentation. The mass spectrometer utilized in this
study was a Finnigan-MAT 312 instrument with reverse geometry
70 75 80 85
Number
+
(magnetic sector preceding electric sector) and fitted with a
combined CI/EI ion source. The programmable DCI probe power
supply unit for ramping the wire current was obtained from
Finnigan-MAT, Sunnvale, CA. Modifications were made to read
applied DCI wire current through an ammeter on the ion source
control unit (2).
RESULTS AND DISCUSSION
The ideal mass spectrum for an unknown material would
be one that yields ions indicative of the molecular weight as
well as structurally informative fragment ions. Unfortunately
for many classes of compounds, including carbohydrates,
glycoconjugates, amino acids, and peptides there is often a
disproportionate distribution of ions when using electron im-
pact (EI) or field desorption (FD). Conventional chemical
ionization (CI) and electron impact ion sources are designed
to analyze samples in the gas phase with sample material
vaporized external to the ionization chamber. Under these
conditions, molecular weight related ions can only be observed
for polar and thermally labile materials which have been
derivatized prior to their analysis. The mass spectra of com-
pounds ionized by FD are usually dominated by the molecular
weight related ions with very few (if any) fragment ions.
In contrast, vaporization of involatile samples by DCI gives
rise to intact or nearly intact neutral molecules which are
subsequently ionized to produce both molecular and fragment
ions ( I ) . Ions indicative of the molecular weight are observed
very early in the desorption profile and usually last for only
a few seconds ( I , 3 ) . Desorption and subsequent ionization
of neutral pyrolysis products formed on the support results
in an abundance of low mass ions which become dominant
in the latter part of the total-ion profile. These factors can
be demonstrated with the disaccharide kojibiose and a plot
of the ratio of ions associated with glycosidic bond rupture
( m / z 180 and 198) to those ions related to the molecular
weight ( m / z 342 = [(M - HzO) + NH4] and m / z 360 = (M
+ NH4)+)against the repetitive scan number (Figure 1). This
ratio (R) is observed to increase dramatically on the de-
scending side of the selected ion profile for the intact di-
saccharide (I). The abundance of ions desorbing after the
initial molecular weight related ion plume (i.e., total ion plot
 ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982 501

For compounds more polar than simple saccharides (such as

Flgure 2. Total ion Intensity and selected ion plot of m l r 537 obtained
from equal amounts of the pentapeptide (Phe-Asp-Ala-Ser-Val) de-
sorbed from (a) bare, wire and (b) PIC wire as a function of increasing
scan number and linear heating current. Mass 537 corresponds to [(M
+
- H20) NH,]' rather than M', as demonstrated by using "NH3 as
the reagent gas ( I ? ) .
and R, Figure 1)indicatee that sample remaining on the wire
is pyrolyzing as a result of the increased heating current and
the closer proximity to the metal surface. Mass spectra ob-
served from this trailing edge of the total-ion profile consist
of a series of high intensity low mass ions formed by pyrolysis.
amino acids and peptides) the total-ion current profile usually
maximizes well after the molecular weight related ion current
profile and recovers only slowly to base line. Furthermore,
for these materials the percentage of the total-ion currenit
carried by molecular weight related ions is much smaller and
is observed for a shorter duration.
Other workers have indicated that the relative ratio of
pyrolysis products to intact neutrals depends largely on two
factors: the nature of the surface on which the sample is
coated and the rate of sample heating (3-7). To reduce
sample-surface interactions and, therefore, increase the de-
sorption of intact neutrals, experimenters have used glass (51,
quartz (7), Teflon (:I), silicone gum (IO), and Vespel ( 4 ) .
However, these materials are not good conductors and are
difficult to heat rapidlly in a controlled fashion. In an effort
to combine the conductivity of wire with the inert surface
properties of the above materials, we explored various ther-
mally stable polymeric coatings. The polyimide material used
in a two-step coating of primer and topcoat proved most
satisfactory.
The influence of wire heating and the resulting pyrolysis
can be decreased by desorbing samples from this PIC wire.
The strength of sample-surface binding is probably much
lower for these polymer-coating wires than for bare metal andi,
therefore, the heat of vaporization is now more related tio
overcoming sample-s,ample interaction. This can be dem-
onstrated by desorbing equal amounts of the pentapeptide
Phe-Asp-Ala-Ser-Val(mol w t = 537) from bare wire and PIC
wire and assessing the results by plotting the selected ion
profiles of m / z 537 ([(M - H20) ",It) +
and the total ion
profiles, (Figure 2) (11). The peptide desorbs intact from both
substrates at essentially the same heating current but is
sustained as such for a longer period of time with greater
molecular weight related ion intensity in the PIC wire ex-
periment. Not only is there a quantitative increase in the ion

105

13'5
280
I

'i I I I 188
M W = 537

537

135

(M-17,- 18)'
(M-17)'

373 393 427 449

55 le@ 150 200 3 0 16 358 400 458 500 550 608

/z
Flgure 3. Mass spectrum obtained at the maximum of m l z 537 for the pentapeptide (Phe-Asp-Ala-Ser-Val) using (a) bare wire (scan no. 36,
Figure 2a) and (b) PIC wire (scan no 38, Flgure 2b).
502 ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982

' C0.e

AAI :N + YH if

*' IS88

ZAl

M W = I490

Reduced Perrnethylated Heptasaccharide

+
ElgUre 4. Mass spectra obtained at the polnt of maximum Intensity of (M NH4)+ for the reduced and permethyiated glucose polymer isolated
from corn syrup and purified by HPLC. The A series ions correspond to cleavage of the glycosidic llnkage with charge retention on the reducing
end whereas the 2 series ions correspond to charge retention on the nonreducing portion of the molecule.

MW = 481

34 1

B
Flgure 5. Mass spectrum of 1-0-stearoyi-2-hydroxy-sn lyceryl-3-phosphorylcholine(lyso platelet activating factor) taken at the point of maximum
molecular weight related ion intensity ( m / z 482 = (MH) ).

522 /

N-Palmi toy1 DI h y d r o l a c t o c e r e b r o s i d e

Flgure 6. Mass spectrum of N-palmitoyl dlhydrolactocerebroside taken at the polnt of maxlmum molecular weight related ion intensity ( m / z 864
= (MH)' and m / z 881 = (M 4- NH4)+).

current carried by the molecular weight related ion but there We have compared DCI using PIC and bare wire for a
is also a qualitative difference in the spectra with the ap- number of compound classes and noted in every case two
pearance of more ions related to sequence structure when major advantages with the PIC wires: (1)A larger percentage
using the PIC wires (Figure 3). of the ion current within each scan is carried by molecular

1La-
Flaure 7. Mass sDectrum of 12.13-dihexanoate ahorbol ester taken at the polnt of maximum molecular weight related intensity ( m / z 578 = (MI
+NH,)+).
weight related ions [(M t- H)+ and (M + NHd)+], and (2) a
greater number of scans within the total-ion profile contain
this molecular weight information. A comparison of the limits
of sample detectability for coated and uncoated wires has not
been pursued extensively in out laboratory. However, pre-
liminary results by single-ion monitoring of the disaccharide
melibose and the phorbol diester 12-myristate-13-acetatehave
indicated sensitivities down to the 2-5 ng level using the PIC
wires. Similar experiments on these compounds using bare
wires were unsuccessful at the 50-ng level.
Earlier studies with FDMS had indicated that derivatization
aids in sample analysis by yielding a more lasting and stable
ion current (8). For DCI analysis of polar compounds de-
rivatization is also helpful and becomes mandatory at higher
molecular weights. As an example, 100 ng of the underivatized
trisaccharide raffinose yielded a protonated or cationized
molecular weight related ion which was only 5% of the base
ion. The molecular ion was not detectable using a bare wire.
Using the combined approach of a PIC wire and derivatization,
a permethylated heptasaccharide of a(1-4) linked glucose
residues yielded a base ion at m / z 1508 corresponding to the
ammonia cationized molecular ion (Figure 4). It is important
to note that the fragment ions produced in this spectrum allow
the sequence to be “read” from either end of the molecule
whereas both the field desorption and fast atom bombardment
spectra of this same compound consist almost entirely of ions
indicative of the molecular weight.
Further comparison of PIC and bare wires for two classes
of lipids, and the very labile phorbol diesters indicated again
that desorption from PIC wires provides greater intensity of
high mass ions with a minimum of pyrolysis. One class of
lipids, the phosphatidylcholines, were evaluated by using as
an example l-O-stemoyl-2-hydroxy-sn-glyceryl-3-
phosphorylcholine (lyso platelet activating factor). This
zwitterionic material gave an intense protonated molecular
ion with numerous structural fragments which allowed ex-
cellent characterization of‘ the lipid and phosphorylcholine
moieties (Figure 5). In a contrasting lipid class, the neutral
but very polar cerebrosides were analyzed. The DCI mass
spectrum from PIC wire of one example in this lipid class
(lacto-N-palmitoyl cerebroside) is presented in Figure 6.
Again, the protonated molecular ion (MH+ = 864), the am-
monium attachment ion ( m / z 881), and the dehydration
product [ m / z 846 = (MH - HzO)’] aid in molecular weight
characterization. Additional fragments characterizing loss of
the galactose (m/z 702) and glucose (m/z 522) moieties as well
as ions related to the fatty acid and sphingosine residues are
also observed. The phorbol esters, potent tumor promoters
isolated from croton oil, are a class of compounds which show
ANALYTICAL CHEMISTRY, VOL. 54, NO. 3, MARCH 1982 5031
578
extreme lability of the ester side chains and field desorption
or chemical ionization (9) is required to determine their
molecular weights. Using PIC wires these tumor promoters
showed a threefold enhancement of the molecular weight
related ion intensity over bare wires and a detectability to the
low nanogram range. ‘Theresults obtained for 12,13-dihexa-
noate phorbol ester were typical (Figure 7 ) .
From an instrumental standpoint, DCI is simple to set up,
easy to perform, and inexpensive to operate. The use of
polyimide-coated wires as the emitter substrate instead of the
commercially supplied bare metal wires extends mass spectral
analysis to higher masses with greater sensitivity and at a
minimum of cost. We would expect this technique to be most
helpful for materials of high polarity which are often available
only in trace amounts.
ACKNOWLEDGMENT
The authors thank C.E. Costello for providing the phorbol
ester samples and for excellent assistance in gathering and
interpreting the data. We also thank Bruce Stratton, Pres-
ident, American Durafdm, for the samples of Pyre-ML topcoat
and primer, Peter Blmberg for the samples of phorbol diester,
and Edward Goetzl for the sample of lyso platelet activating
factor.
LITERATURE CITED
(I) Cotter, R. J. Anal. Chem. 1980, 52, 1589 A.
(2) Kaufman, H., Flnnigan-MAT, Bremen, Germany.
(3) Hansen, G.; Munson, 13. Anal. Chem. 1880, 52, 245.
(4) Cotter, R. J.; Fenselau, C. C. Blomed. Mass Spectrom. 1979, 6 , 287.
(5) Baldwin, M. A,; McLafferty, F. W. Org. Mass Spectrom. 1973, 7 ,
1353.
(6) Beuhler, R. J. Flanagan, E.; Greene, L. J.; Friedman, L. J. Am. Chem.
Soc . 1874, 96, 3990
(7) Ohashi, M.; Nakayamti, N. Org. Mass Spectrom. 1978, 73,642.
(8) Costello, C. E.; Wilson, B. W.; Biemann, K.; Reinhold, V. N. I n “Cell
Surface Glycolipids”; dweeley, C. C., Ed.; American Chemical Society:
Washington, DC, 1980; p 35; ACS Symp. Ser., No. 128.
(9) Solomon, J. J.; Van Duuren, B. L; Tseng, S A . Biomed. Spectrom.
1978, 5 (2), 164.
(IO) Howlln, J. G.; Carroll, D. I.; Dzldlc I.; Hornlng, M. G.; Stillwell, R. H.;
Hornlng, E. C. Anal. Lett. 1979, 72, 573.
(11) Carr, S. A.; Reinhold, V. N. I n “Proceedings of the I V International
Conference on Methods in Protein Sequence Analysis”; Elzlnga, M.,
Ed.; The Humana Press: Clifton, NJ, 1981, in press.
RECE~VED for review September 24,1981. Accepted November
24, 1981. A preliminary account of this work was presented
at the 29th Annual American Society for Mass Spectrometry
meeting held in Minneapolis, MN, May 1981, papers num-
bered RPA 12, RPA lB, and RPA 14. This work was sup-
ported by the National Institute of General Medical Sciences,
National Institutes of Health (Department of Health and
Human Services, USA), Grant No. 5 RO1 GM2625.