Chem. Mater. 1992,4,129-I31
129
Synthetic Strategies in the Formation of Iron-Modified Polyimide Films Joseph J. Bergmeister and Larry T. Taylor* Virginia Polytechnic Institute and State University, Department of Chemistry, Blacksburg, Virginia 24060 Received March 16, 1992
Composite polyimide films were prepared by casting and thermally curing a viscous homogeneous solution of either a poly(amide acid) or a poly(amide ester) doped with either tris(acetylacetonato)iron(III) or
trihydrogen hexachloroiron(II1). The various polyimide precursors (with or without coordination sites) and iron dopants (of different thermal stability) were chosen to examine their effect on the formation of iron-rich surface layers and dopant bulk particle size after curing. Films prepared with both polyimide precursors and Fe(a~ac)~ produced materials with very small bulk particles and no iron-rich surface layers. Films produced with both polyimide precursors and HsFeCI, produced large bulk particles and surface layers of iron oxide. Films prepared, however, with H3FeCI, poly(amide ester) solutions had larger surface layers than films prepared from H3FeC&/poly(amideaci ) solutions.
d
Introduction Metal/metal oxide modified polyimides comprise a unique class of materials, in that they can exhibit properties of both the polyimide matrix (e.g., thermal stability and chemical inertness) and the metal modifier (e.g., magnetic susceptibility and electrical conductivity).l A great deal of research has been devoted to the synthesis and characterization of these materials because of their potential in aerospace and electrical applications.2 Typically, there are three synthetic routes used to obtain metal/metal oxide modified polyimides: deposition, infusion, and in situ reaction. A deposition process consists of adding a metal or metal oxide layer directly onto the surface of the polyimide film via sputter ~ o a t i n gelec,~ trolyais,4 and/or laminati~n.~ An infusion process consists of diffusion of a metal complex into a partially or fully cured film followed by additional chemical or thermal treatment to produce the metal/metal oxide.6 An in situ process consists of casting a film of a homogeneous solution of a metal complex and a polyimide precursor, where conversion of the metal complex to its zerovalent state or oxide occurs during the curing cycle of the pre-polyimide. In our laboratory we have pursued the in situ synthesis of polyimide metal/metal oxide composites prepared with a wide range of metal^.^ Several authors have reported the in situ synthesis of iron oxides in polymeric materials, with the focus of the work being on the transformation of encapsulated iron These complexes to iron oxides via base hydroly~is.~?~ reactions emulate known solution reactions for the synthesis of iron oxides. The decomposition of iron complexes (1) (a) Boggess, R. K.; Taylor, L. T., J. Polym. Sci., Polym. Chem. Ed. 1987,25,685. (b) Achar, B. N.; Fohlen, G . M., Parker, J. A. J. Polym. Sci., Polym. Chem. Ed. 1983, 21, 3063. (c) Varma, I. K.; Saxina, S., Tripathi, A., Goel, T. C., Varma, D. S. J. Appl. Polym. Sci. 1986,32,3987. (2) (a) Auerbach,A. J. Electrochem. SOC.1984,131,937. (b) Angelo, R. J. E.I.DuPont de Nemours Co.; US Patent, 3,073,785, 1959. (c) Porta, G. M.; Rancourt, J. D., Taylor, L. T. Chem. Mater. 1990, 1, 269. (d) Bergmeister, J. J.; Rancourt, J. D., Taylor, L. T. Chem. Mater. 1990,2, 640. (3) Hanazono, M.; Narishige, S., Kawakami,K., Saito, N., Takagi, M. J. Appl. Phys. 1982,53, 2608. (4) Crapentin,J.; Mahlkow, H., Skupsch,J. US Patent 4,517,254,1985. (5) Nakaguchi, T.; Furuya, K. Japan Patent 60,50,590, 1985. (6) Mazure, S.; Reich, S. J. Am. Chem. SOC.1986, 90,1365. (7) Taylor, L. T.; Rancourt, J. D. In Inorganic and Metal Containing Polymeric Materials; Sheata, J., Ed.; Plenum Press: New York, 1990; pp 109-126. (8) Okada, H.; Sakata, K., Kunitake, T. Chem. Mater. 1990, 2, 89. (9) Sobon, C. A.; Brown, H. K., Board, A., Calvert, P. D. J. Mater. Sci. Lett. 1987, 6, 901.
to magnetic species in polymeric materials has been investigated with emphasis on the decomposition of iron pentacarbonyl to iron powder,1° iron oxide,ll or some combination of the two.12 We find in our laboratories that when polyimide precursors were doped with Fe(C0I5 and thermally cured, nearly all of the Fe(CO)5evaporated out of the film, even at very high metal loadings (1&15 wt ’3). The decomposition of Fe3(C0)12in a polyimide matrix has been reported by Sen and ceworkers.13 In their study, they cast and cured poly(amide acid) solutions containing up to 2 w t ’% Fe3(C0Il2to form iron-modified films. On the basis of X-ray photoelectron spectroscopy analysis of the iron-modified films,they concluded that Fe3(C0)12was converted to FeOOH during the curing of poly(amide acid) to its polyimide (i.e., the water of imidization reacts with the Fe3(C0)12to produce submicron particles of FeOOH). They suggested that the acid sites on the pre-polyimide may inhibit the formation of iron-rich surface layers. In other words, as the film is cured, the dopant may coordinate to the polyimide precursor and decompose to nonsoluble products before imidization and metal migration can occur. In this study we also suggest similar polymer-metal interactions; however, we find that the thermal stability of the dopant may have a greater effect than polymer/metal interactions on the formation of iron-rich surface layers as well as the formation of large bulk particles. For example, during the curing cycle a thermally stable dopant may melt or dissolve in the polyimide precursor or low molecular weight polyimide and therefore either aggregate to form large particles in the bulk of the film or migrate unchanged to the surface of the film where transformation or other metallic species could occur on the surface. To investigate the metal coordination and additive stability hypotheses, a variety of pre-polyimide precursors and iron dopanta were chosen. The iron dopants used were (tris(acetylacetonato)iron(III) and trihydrogen hexachloroiron(III). Fe(acacI3melts and decomposes at 193 OC, while H3FeC1, decomposes to FeC1, which melts at 306 (10) Smith, T. W.; Wychick, D. J. Phys. Chem. 1980,84, 1621. (11) Reich, S.;Goldberg, E. P. J . Polym. Sci., Polym. Phys. Ed. 1983, 21, 869. (12) (a) Griffiths, C. H.; O’Horo, M. P., Smith, T. W. J. Appl. Phys. 1979,50, 7108. (b) Tannenbaum, R.; Goldberg, E.P., Flenniken, C. L.
In Metal-Containing Polymeric Systema; Sheata, J. E.,Carraher, C. E. Jr., Pittman, C. U. Jr., Eds.; Plenum Press: New York, 1986; pp 303-339. (13) Nandi, M.; Conklin, J. A,; Salvati, L. Jr., Sen, A. Chem. Mater. 1990, 2, 772.
0891-4156/92/280~-0129$03.00/0 0 1992 American Chemical Society
Bergmeister and Taylor
730 Chem. Mater., Vol. 4, No. 3, 1992
PMDA 0
II
I1
ODA
0
0
DMTC
P
o
0
0
Fe(acac)!
Figure. 1. Chemical structures of the monomers and dopants used in this study.
OC.14 The polyimide precursors were either a poly(amide acid) or a poly(amide ester). It is proposed that the coordination site on the polyimide precursor is the carboxylic acid functionality. Therefore, by replacing the acid group with a less coordinating ester group, the tendency for metal to bond to the polyimide precursor is reduced.
Experimental Section The monomers used in this study were 3,3',4,4'-benzophenonetetracarboxylic acid dianhydride (BTDA), 1,2,4,5benzenetetracarboxylicacid dianhydride (PMDA), 2,5-dicarbomethoxyterephthaloyl chloride (DMTC), and 4,4'-oxydianiline (ODA). The chemical structures of the monomers and polymers are shown in Figures 1 and 2, respectively. BTDA and PMDA (Allco) were dried at 90 "C under vacuum prior to use. Zonerefined ODA (Aldrich) was dried at 70 "C under vacuum prior to use. DMTC was synthesized by the method described by Bell16 and was used without further purification. Tris(acety1acetonato)iron(IID, F e ( a ~ a c(Harshaw), )~ was dried under vacuum at 50 "C. HPU: grade NJ-dimethylacetamide @MAC),(Aldrich) was stored over molecular sieve and under a nitrogen atmosphere. Anhydrous ferric chloride, FeC13 (Aldrich), was used as received and stored under dry nitrogen. Poly(amide acid)s were prepared by reacting 4.0 mmol of the appropriate dianhydride (BTDA or PMDA) with 4.0 mmol of ODA in DMAc under anhydrous conditions. All solutions were 13-15% solids. The resulting poly(amide acid) was stirred for at least 4 h. At this point either 1.0 or 2.0 mmol of Fe(aca& was added to the poly(amide acid) solution. The resulting solution was stirred for an additional 2 h. Iron chloride solutions were prepared by reacting the Fe(acac),/poly(amide acid) solution with HCl gas. The conversion was termed complete when the red color of F e ( a ~ a cwas ) ~ replaced by the yellow color of [FeC1,]3-X( x = 3-6). This latter solution had to be cast quickly and cured since the HCl gas catalyzed the decomposition of the poly(amide acid). Poly(amide ester) solutions were prepared by reacting 4.0 mmol of DMTC and 4.0 mmol of ODA in DMAc under anhydrous conditions at -10 "C. After stirring for 2 h, the solution was warmed to room temperature and stirred for an additional 2 h. The poly(amideester) was then precipitated in and washed with copious amounts of water followed by vacuum drying at room temperature overnight. The poly(amide ester) was then redis(14) The Handbook of Chemistry and Physics, 65th ed.; Weast, R. C., Ed.; CRC Press: Boca Raton, F1, 1985; p B103. (15) Bell, V. C.; Jewell, R. A. J.Polym. Sci. A 1967, 5 , 3043.
solved in DMAc to give a viscous yellow solution. Thii procedure was employed to remove dissolved HCl (by products) from the previous step. At this time either 1.0 or 2.0 mmol of F e ( a ~ a c ) ~ was added to the solution and stirred for 2 h. Solutions modified with iron chloride were prepared as follows: The poly(amide ester) was prepared as described above with the omission of the precipitation step. At that point either 1.0 or 2.0 mmol of F e ( a ~ a c ) ~ was added to the solution. The dissolved HC1 converted the iron dopant to iron chloride. This was always observed when 1.0 mmol of F e ( a c a ~was ) ~ added to the polymer solution; however, in some instances the conversion to iron chloride did not occur with 2.0-mmol batches of F e ( a ~ a c ) ~The . conversion was completed by the addition of a small amount of HCl gas. These solutions were quickly cast and cured for the same m n s as the poly(amide acid) cases. Films were prepared by casting the iron-modified polyimide precursor solutions at 20 mil onto a clean dust-free soda lime glass plate. The films were cured at 80 "C for 20 min and 100, 200, and 300 "C each for 1h under a dynamic atmosphere of dry air. After cooling to room temperature the films were removed from the glass plate by peeling a corner of the film back with a razor blade and placing a drop of distilled water under the corner of the fii. Within 30 min most of the f i b s delaminated from the glass plate. The films were then left on a paper towel to dry overnight. The side of the f i expoaed to the curing atmosphere is referred to as the air-side surface. Likewise, the side of the f i iin contact with the glass plate is referred to as the glass-side surface. Transmission electron micrographs (TEM) were taken with a Philips Model 420 scanning transmission electron microscope. Samples for TEM analysis were embedded in Polyscience ultra low viscosity resin and cured for 8 h at 70 "C. Samples were then sectioned to between 500 and 800 A with a Reichert-Jung ultramicrotome using a Microstar diamond knife. Thin sections were placed on 200-mesh copper grids prior to analysis. X-ray photoelectron spectroscopy (XPS) data were obtained with a Perkin-Elmer Phi Model 530 ESCA system using a magnesium anode (Ka= 1253.6 eV) at 400 W. Samples were attached into aluminum mounts with double stick transparent tape. The binding energies were referenced to the aromatic C(1s) photopeak at 284.6 eV. Thermogravimetric analyses were performed with a PerkinElmer Model TGS-2 system at 10 "C/min heating rate with a dynamic atmosphere of air or nitrogen (50 mL/min). Elemental anal(iron and chlorine) were obtained by Glabraith Analytical Laboratories, Knoxville, TN. Magnetic properties of the modified films were obtained using a LDJ 7500A BH meter. Films were wrapped around a glass rod (5" 0.d.) and placed in the detector at 23 "C. A field strength of 2000 Oe was needed to saturate the samples.
Results and Discussion The polyimide precursors and imidized products used in this study are shown in Figures 1 and 2. The major difference in the polyimide precursors is that two are poly(amide acid)s and one is a poly(amide ester). The poly(amide ester) of BTDA/ODA could not be synthesized in high enough molecular weight; therefore, it was not included in this study. The PMDA/ODA poly(amide acid) and the DMTC/ODA poly(amide ester) both imidize to give the same polyimide; however, they differ by containing either an acid or ester functionality and expel either H20 or methanol respectively during the imidization process. A series of solutions was prepared in order to determine with which sites on the polyimide precursor the metal dopant might coordinate. We used Fe(aca& and anhydrous FeC13 as probes and a poly(amide acid) and a poly(amide ester) as polyimide precursors. These two dopants were chosen because of their reactivity. It has been shown that F e ( a ~ a cw )il ~l undergo a ligand exchange under mild conditions (acid catalyzed at 50 OC),16while FeC13is readily (16) Aly, M. M.; El-Awed, A. M., Inorg. Chem. Acta 1980, 38, 3.
Iron-Modified Polyimide Films
0
Chem. Mater., Vol. 4, No. 3, 1992 731
0
-2 HzO
0
0
0
/-2
f
MeOH
0
Figure 2. Chemical structures of the poly(amide acids) and poly(amide ester) used in this study.
solvated by various nucleophilic functionalities (e.g., DMSO, H20, and alcohols)17and may even undergo exchange of one, two, or all three18J9chloride ligands very rapidly. For these reasons we should observe greater metal to polymer interactions upon the addition of FeC1, to the polyimide precursor solutions than with Fe(aca& Solution Observations. Nondoped solutions, typically 15-18 w t ?% solids,of the polyimide precursors were yellow and viscous. The dopant Fe(acac), dissolved in a DMAc solution of poly(amide acid) or poly(amide ester) gave a ruby red colored solution. A slight visual decrease in the viscosity of the poly(amide acid) solution was observed after the addition of Fe(acac)3. A decrease in the viscosity a t this point suggested the Fe(acac), coordinates to individual polyimide precursor chains primarily through the acid functionality by the loss of the one acac ligand (eq 1). If the iron were coordinating to two or more sites on II II \
x + Y = ~ xs3
+
0 0 H -N-C
could be cast. Gelation, in comparison, was observed when FeCl, was added to a DMAc solution of a poly(amide acid). Gelation of the poly(amide acid) occurred upon the addition of FeC13and not Fe(acad3,we feel because all three C1- ligands can more easily be replaced by amide or ester donor groups on the polyimide precursor (eqs 2 and 3). As
H C-N-A(I
i
FeCI3
0 'Fe(acac),
-C'
+ Hacac (1)
' 0 '
the polymer backbone then it would be acting as a crosslinking agent between chains and an increase in viscosity would be observed. On the other hand, no visual change in the viscosity was observed when F e ( ~ c a c was ) ~ added to a DMAc solution of poly(amide ester). Thus,the dopant probably does not interact with the poly(amide ester). When FeC1, was added to a DMAc solution of a poly(amide ester) a very viscous brown solution resulted; however, the solution was still free flowing such that a film (17) Drago, R. S.; Carlson, R. L.; Purcel, K. F. Znorg. Chem. 1965,4, 15.
(18) Paul, R. C.; Naruca, R. C.; Kaur, S.; Vaisht, S. K. Trans. M e t . Chem. 1976,1, 284. (19) LaMar,G. N.;Van Hecke, G. R. J. Am. Chem. SOC.1969,91,3442.
the iron binds to the carboxylic acid groups of different poly(amide acid) chains, a network is formed, the viscosity increases,and if the interactions are strong enough gelation occurs (as in the case of the poly(amide acid)). In the case of the poly(amide ester), no acid coordination sites exist, yet an increase in the viscosity was observed. Thus it is believed that the iron is bonding to the poly(amide ester) chains through the amide carbonyl groups (eq 2). The cross-linked FeC13/poly(amide acid) golden gel matrix can be redissolved into DMAc by the addition of Hacac (eq 4) to yield a red solution. Infrared and UV spectroscopies indicated the re-formation of Fe(aca& in solution. This observation suggested that the iron chloride is the actual cross-linking agent and not a catalyst for an intermolecular cross-linking reaction between the polyimide precursor chains.
732 Chem. Mater., Vol. 4, No. 3, 1992
f - eb-/ ]
+ 3H(acac)
-
Fe(acac)3 + 3HOOC-
Bergmeister and Taylor
1
(4)
J
Since iron chloride could not be added directly to a solution of poly(amide acid), it was prepared in situ by transformation of Fe(acac), to [FeC1,..sx (x = 3-6) with HC1 gas or the dissolved HC1 which arose from the reaction of the acid chloride and the diamine to form the poly(amide ester). The iron chloride synthesized by the reaction of Fe(acac), and HC1 is believed to be H,FeC& for the following reasons. The addition of either FeC1, or KFeC1, to the poly(amide acid) solution resulted in gelation: In previous studies high HC1 concentrations converted FeC1, to H3FeC&.20 Furthermore the addition of K,FeC& to a poly(amide acid) solution did not cause gelation. We therefore chose to use H,FeC& instead of K,FeC& since HC1 can readily be released from the polyimide during curing, while KC1 will remain in the polymer matrix. Characterization of the Modified Films. The films produced by the various combinations of dopants and pre-polyimides fell into two categories: those with and without iron-rich surface layers. Modified films prepared with H,FeC& contained iron-rich surface layers and large bulk particles of FeC13, while f h prepared with F e ( a ~ a c ) ~ contained submicron particles of iron oxide. Physical data pertaining to the Fe(acac),-modified polyimide films are presented in Table I. Fully imidized nonmodified polyimide films were yellow, flexible, and creasible. All of the modified films were dark brown to black and some were flexible; however, all the films fractured when creased. Films containing PMDA adhered strongly to the glass plate during the cure such that removal of these films could be accomplished only by scrapping them off the plate with a razor blade. This mode of removal yielded a black powder. The PMDA/ODA/ Fe(acac), films (Table I, films 3 and 4) were then better prepared by casting the Fe(acac),-modified poly(amide acid) solution onto a partially cured PMDA/ODA film instead of glass, followed by curing to 300 "C as stated in the Experimental Section. This method also yielded brittle films; however, they could be sufficiently handled to perform further limited analysis (magnetic and XPS). The thermal properties of the polyimide f h s were determined by monitoring the polymer decomposition temperature (PDT). The PDT is the temperature at which 10 w t % of the sample is lost. It was observed that doping with Fe(acac), decreased the PDT by 60 to 180 "C. Both the modified and nonmodified films had a higher PDT when performed under an inert atmosphere of N2 then under 02.The iron-modified films were believed to decompose through an oxidative pathway since the presences of oxygen decreased the PDT of the modified films so drastically (as compared to the nonmodified films). The oxidative pathway is thought to occur via a metal catalyzed pathway. As seen in Table I, the iron-modified polyimide films retain 90-96% of total iron after cure. In fact, we find that only 5% of iron (or "initial sample mass") is lost when Fe(acac), by itself is heated in a thermal analyzer under similar conditions used for polyimide curing (i.e., temperature, heating rate, and atmosphere). Figure 3 shows the transmission electron micrograph (TEM) of film 6. Similar TEMs were obtained for all six Fe(acac),-modified f h . The most striking feature in each (20) Hatfield, W. E.; Fay, R. C.; Pflyger, C. E.; Piper, T.S.J. Am. Chem. SOC.1963,85,265.
Figure 3. Transmission electron micrograph of film 2.
TEM is the formation of highly dispersed particles in the polymer matrix. Furthermore, these particles have a narrow size distribution ranging between 20 and 80 nm. Also apparent in the micrograph is the presence of a depletion zone (of approximately 4.5 pm) near the air-side surface. The loss of 5% of the dopant material during curing cannot account for this zone. Therefore, this region is believed to contain either atomically dispersed particulate matter that cannot be observed by TEM or the dopant has migrated from this area of the film. The other five films contained similar depletion zones. We have found that with other metal-modified polyimides that metal rich surface layers are ~ b s e r v e d .However, ~ with these iron-modified films, even under high magnification, it is difficult to distinguish a phase contrast from possible surface layers. Evidence for a surface layer of iron oxide, vide infra, is however suggested by XPS, where up to 10 at. % iron is observed in the first 50 A of the air-side surface of these films. Thus the hypothesis of an atomically dispersed metal oxide layer on or near the surface may be valid. The most powerful technique for determining the composition of the surface region is XPS. With XPS the relative concentrations and chemical states of most elements can be determined. Figure 4 shows a typical oxygen 1s photopeak obtained from the analysis of the air-side surface of film 2. A curve-fitted spectrum comprised of four bands is also shown in Figure 4. Bands at 533.2, 532.3, and 531.6 eV are assigned to the ether, ketone, and imide oxygen functionalities of the BTDA/ODA polyimide. These oxygen functionalities have a measured relative atomic concentration of 1:1:4 which is what would be expected for a BTDA/ODA polyimide. Both the position and relative concentrations are consistent with literature
Chem. Mater., Vol. 4, No. 3,1992 733
Iron-Modified Polyimide Films
Table I. Appearance of the Fe(acac)8-ModifiedFilms w t % iron film no.
monomers BTDA/ODA BTDA/ODA PMDA/ODA PMDA/ODA DMTC/ODA DMTC/ODA
1 2 3 4 5 6
theor mol % Fea 0.20 0.33 0.22 0.33 0.22 0.33
actual 2.64 5.13
theory 2.76 5.31
C
C
C
C
3.26 5.06
3.42 6.43
PDT" flexibleb
color brown black brown black black black
YES NO NO NO YES YES
BTDA/ODA PMDA/ODA DMTC/ODA
air 421 413
N2 526 505
C
C
C
C
438 428 593 586 523
557 522 628 631 589
Polymer decomposition temperature (i.e., 10% weight loss at heating rate in air of 10 OC/min). Determined on a polyimide and Fe203 basis, assuming complete imidization and no residual DMAC or acac ligand. 'Could not be determined, these were double-cast films.
2-
n n
a
z*
Y
z
I 1.0
536
532
528
BINDING ENERGY, eV Figure 4. Oxygen (1s) photopeak of film 2 (-), best curve fit (- -).
-
1 along with the
valuesz1 and values obtained in-house on nonmodified films. A fourth oxygen band at 530.1 eV is assigned to an iron oxide. A similar binding energy has been reported for oxygen in various neat iron oxides.22 The oxygen 1s photopeaks obtained with the PMDA-derived polyimides were similar with the exception that the band corresponding to the ketone functionality, as expected, was absent. The relative atomic concentration between oxide oxygen and iron, on both the air and glass-side surfaces, as determined by XPS in the six films ranged from 1.34to 1.70, which corresponds to the expected stoichiometry for Fez03 (FeOl.d or Fe304(FeOl.%). The oxidation state of the iron in the films could not be unequivocally determined by the iron 2p photopeak band width and/or position since various iron oxide model complexes give similar XPS spectra. It should be noted that XPS is sensitive only to those atoms on the surface, and therefore such data should not be used conclude the composition of bulk particles in the film. Therefore the oxidation state as well as the chemical composition of the iron in the modified films had to be ascertained indirectly by magnetic susceptibility measurements and model reactions. The magnetic properties of the iron-modified polyimide films were measured with a BH meter. Figure 5 shows a typical hysteresis loop obtained from the modified films. The most striking feature in the hysteresis loop is its skewed shape. This is a result of having the magnetic axis of the particles randomly oriented throughout the bulk of the film.23 A square loop would be observed if the particles (21) Clark, D.T.; Thomas,H.R.J . Polym. Sci., Polym. Chem. Ed. 1987,20, 790.
0.5
(22)McIntyre, N.S.;Zetaruk, D. G.Anal. Chem. 1977,49, 1521.
Figure 5. Magnetic hysteresis loop of film 4: (a) saturation magnetization; (b) remnant magnetization; (c) coercivety. Table 11. Coercivities of the Fe(acac)3-ModifiedFilms film no. H,,Oe filmno. H,,Oe 1 104 4 101 2 130 5 32 3 98 6 134
were oriented and parallel to the magnetic film. The shape of the hysteresis loop remained unchanged regardless of the position of the modified films with respect to the magnetic field. This observation suggests that there is no overall preferred orientation of the iron oxide particles in our films. Kunitake and co-workers* have shown that an oriented material composed of an organic matrix and small particles of Fe304(100nm) can be synthesized by the base hydrolysis of FeC12 encapsulated in a laminar matrix. They found that their film could be magnetized more readily ("squaren loop) if the magnetic field was applied parallel to the film. The coercivity (H,) is the force required to demagnetize a magnetically saturated sample. Coercivity data for each film are presenced in Table 11. The coercivities were found to be between 32 and 134 Oe. The values are lower than most literature values for iron oxides that are used in magnetic recording application^.^^ However, the iron oxide particles produced in our films were cubic (-50 nm). Higher coercivities are indicative of acicular particles (0.2 pm X 1.0 The magnetic properties of the ironmodified films suggested that conversion of the paramagnetic dopant to either ferromagnetic yFe2Os or Fe304 (23)Jakubovics, J. P. Magnetism and Magnetic Materials; Bath Press: England, 1987;pp 98-99, (24)Corradi, A. R.IEEE Trans Mag.: 14 1978,5, 655. (25)(a) Ishikawa, T.; Matijevic, E. Langmuir 1988,4, 26. (b) Ozaki, M.; Matijevic, E. J . Colloid Interface Sci. 1985, 107, 199.
734 Chem. Mater., Vol. 4, No. 3, 1992
Bergmeister and Taylor
Table 111: Appearance of the H,FeClr-Modified Films film no. 7 8 9 10 11
12 BTDA/ODAb alone PMDAIODA~done PMDA/ODAc alone
monomer BTDA/ODA BTDA/ODA PMDA/ODA PMDA/ODA DMTC/ODA DMTC/ODA
[Fe], Mol % 0.20 0.33 0.20 0.33 0.20 0.33 ~~
color brown silver silver silver brown silver
flexible yes yes yes yes yes yes
air 495 471 495 481 474 505 593 586 523
N2 502 490 525 553 547 536 628 631 589
polym dec temp, K iron retained" 2.37 88.22 4.30 84.38 2.87 85.52 5.01 78.69 3.14 93.59 5.14 81.30
% Fe found
% C1 found
4.02 6.56 5.03 7.35 4.76 7.92
a Percent retained was determined by mol of Fe present/mol of Fe doped, the moles of Fe present was calculated as follows: all C1- is in the form FeCl,, any remaining Fe is in the form Fe203,moles Fe doped is the moles of Fe per repeat unit (0.25 or 0.5 mol/repeat unit). Poly(amide acid) precursor. Poly(amide ester) precusor.
has occurred. The possibility of forming antiferromagnetic FeOOH or paramagnetic a-Fe2O3 is discarded since there materials do not exhibit remnant magnetization or a coercivity. The magnetic measurments cannot revel the extent of conversion of F e ( a ~ a cto ) ~y-Fe203or Fe304since small amounts of F e ( a ~ a c would )~ not affect the Hc. Small amounts of Fe(acad3would affect the remnant magnetization; however, no known standards of y-Fe2O3 or Fe304 encapsulated in a polymer matrix exist; therefore, the decrease in remant magnetization cannot be determined. One method to distinguish from y-Fe203 from Fe304is X-ray diffraction. X-ray diffractograms of films 2,4, and 6 contain reflection due to y-Fe2O3 or Fe304. It was observed from the X-ray diffractograms that reflections due to FeO, a-Fe203,or any hydroxy oxides were not present. The reflection to distinguish between y-Fe2O3 or Fe304 could not be resolved from the baseline. Films 1,3, and 5 did not contain enough iron to give reflections with a significant signal-to-noise ratio. Since X P S analysis, magnetic measurements, and X-ray diffraction indicated that either y-Fe203or Fe304is formed during the curing process, model reactions were performed to aid in the identification of which iron oxide is actually produced. X-ray powder diffraction was a particularly valuable tool in the monitoring the model reactions since it could easily distinguish between the various oxides and the starting material. It was found that heating F e ( a c a ~ ) ~ to 300 "C in air or nitrogen for one hour produced y-Fe20a. Also, heating Fe304to 300 "C under dry air (the same atmosphere in which the films were cured) produced yFs03quantitatively. Therefore, if Fe(a~ac)~ was converted to Fe304during the cure, the Fe304produced would be readily oxidized to y-Fe203. Furthermore, if FeOOH was produced in the polyimide film as Sed3et al. observed in their Fe3(CO)12-dopedfilms, then it would readily dehydrate to produce y-Fe203a t temperatures above 180 0C.24 For these reasons we believe that only y-Fe203is produced by the thermal decomposition of Fe(acac), in the polyimide matrices. H3FeC16-ModifiedFilms. Physical properties pertaining to the H3FeC& modified films are presented in Table 111. Fully imidized nonmodified films were yellow, flexible, and even creasible. The modified films were typically brown (0.20 mol % Fe) or had a metallic luster (0.33 mol % Fe). All of the modified films were flexible to a certain extent. The films could be creased if folded air-side to air-side; however, they fractured readily if creased glass-side to glass-side. For the f h s reported here, the PDT decreased in value from 138 to 18 "C upon the addition of the metal dopant to the polyimide matrix. The PDT of both the modified and nonmodified films were lower when performed under air as compared to nitrogen.
Furthermore the difference in the PDTs when performed under air versus nitrogen were greater in the presence of the metal dopant. This suggested that the decomposition is catalyzed, initiated, or caused by the metal dopant. We find an increase of 49 to 34 "C in the PDT when comparing the films modified with H3FeC&to films modified with Fe(acac)* This observation has been attributed to the iron oxide surface layers produced on these films (vide infra). The iron-rich surface is believed to act as a protective layer on the film, protecting it or slowing the attack of oxygen on the organic film and/or sealing the polyimide from escaping volatile components which are produced during the thermal decomposition. Elemental iron and chloride analyses are also presented in Table 111. We have postulated that the dopant is H3FeC&;however, there are no reports in the literature where H,FeC& has been isolated in the solid state. Therefore precise thermal decomposition products are not known. It is known that a complex equilibrium exist between FeC13, HFeCl,, and H3FeC&in the presence of HC1 and a solvating solvent.26 Since any excess HCl and solvent will be expelled from the film during the curing process, it was believed that the thermally stable FeC1, species would be formed from the decomposition of H3FeC& in the bulk of the film. If we assume that all the chlorine in the film is in the form of FeC13,then for every 3 mol of Cl- there is 1mol of iron. If the iron which is not in the form of FeC13 is Fez03 (vide infra) and complete imidization has occurred, the percent iron retained by the film during curing can be determined. This is accomplished by subtracting the mass of both FeC13 and Fez03 from the total mass and assuming the remaining mass is only due to polyimide. The films were found to retain 70 to 94% of the original iron. The loss of iron in the fully cured f h was attributed to the sublimationof FeC13from the film during the curing process (FeC13melts and sublines at 300 to 306 "C). Figures 6-8 are typical TEMs obtained from the H,FeC&-modified films. Figure 6 shows a whole section of film 7. In this TEM there are three different regions: a surface layer, a depletion zone (immediately below the surface layer), and a bulk region. The other five films gave similar TEMs. As seen in Figure 6, there are both large particles and holes contained within the bulk region of the film. It is believed that the holes occurred during ultramicrotoming, in part from tearing large particles out of the film but mostly from dissolving the bulk particle of FeC1, into the water used for flotation. It should be ~~
(26)(a) Nicholls, D. In Comp. Inorg. Chem.; Bailar, J. C., Trotman Dickerson, A. F., Eds.; Pergamon Press: Oxford, 1973;pp 1038-1041. (b) Rodriguez, F.; Moreno, M. Trans. Met. Chem. 1985,10, 108. (c) Friedman, H. J. Am. Chem. SOC.1952, 74, 5.
Iron-Modified Polyimide Films lum
Figure 6. Transmission electron micrograph of film 7.
noted that the bulk particles obtained here were significantly larger then those obtained in previous studies when Fe(acac)3or Fe3(CO)1213were used as the dopant. The depletion zone shown in Figure 6 resulted from the dopant either migrating to the surface or migrating back into the bulk of the film and/or subliming of the dopant out of the film during the curing process. Figures 7 and 8 show the two different surface layers observed in the modified f h s . The fmt, Figure 7 (film 12), is typical of the surface layers of metal/metal oxide modified polyimide films that we have found with other metal dopants (eo3and Cu16),while Figure 8 (film 11)shows a well-defined surface layer plus an additional layer termed a diffusion layer. The diffusion layer was found in only three of the films (films 8,9,11). The formation of the diffusion layer does not appear to be the result of any single variable (i.e., polyimide precursor or dopant level), but rather it may be influenced by other nonmonitored variables such as the concentration of HC1 or the viscosity of polyimide precursor. We have found?' as well as others,28that posttreating a fully imidized polyimide f h containing a continuous surface layer of metal of metal oxide will result in a similar architecture (i.e., a surface and diffusion layer). Therefore at this time we can not determine conclusively if this diffusion layer resulted from dopant migrating to the surface or a metal-rich layer being diffused back into the polyimide film during the thermal curing process. Table IV summarizes the dimensions of the surface layers produced in the H3FeC&-modifiedfilms. (27) Porta, G. M.; Rancourt, J. D., Taylor, L. T. Chem. Mater. 1989, 1, 269. (28) Mazure, S.; Reich, S. J. Am. Chem. SOC.1986,90,1365.
Chem. Mater., Vol. 4, No. 3,1992 735 300 - A
Figure 7. Transmission electron micrograph of the air-side surface of film 12. Table IV. Surface Layer Dimensions of Iron-Oxide" Produced in H3FeClsModified Polyimide Films initial [Fe], thickness of surface layers, A polyimide mol % air side glass side BTDAIODA" 0.20 288 70 0.33 255 (401)b 400 PMDA/ODA" 0.20 114 (187) 60 0.33 802 907 PMDA/ODAd 0.20 111 (1514) 0 0.33 1464 1526 a Determined by X-ray photoelectron spectroscopy. Diffusion layer thickness in parentheses. e Poly(amide acid) precursor. Poly(amide ester) precursor.
Table V. Summary of XPS Data from the Analysis of the Air-Side Surfaces of H3FeC1,-ModifiedFilms film at. % at. 94 at. % iron-to-oxide no. iron oxide oxygen chlorine oxygen ratio 7 11.99 20.31 0.87 1.73 8 9.76 16.69 1.39 1.43 9 15.88 24.99 1.26 1.61 10 16.54 25.30 1.98 1.54 11 14.16 21.96 0.97 1.58 12 18.80 30.99 0.97 1.64
The composition of the near surface region of the films was determined using XPS in the same fashion as the Fe(acac)3-modifiedfilms. A summary of the XPS results for the air side and glass side surfaces is shown in Tables V and VI, respectively. The relative atomic concentrations of iron and chlorine were determined using standard Perkin-Elmer operating software. All of the modified films
Bergmeister and Taylor
736 Chem. Mater., Vol. 4, No. 3, 1992 300 A c
Figure 8. Transmission electron micrograph of the air-s-,,. surface of film 11. Table VI. Summary of XPS Data from the Analysis of the Glass-Side Surfaces of H,FeClc-Modified Films at. % iron-to-oxide at. % film at. % oxide oxygen chlorine oxygen ratio iron no. 1.20 1.33 7 2.50 2.80 3.69 1.37 8 9.68 11.60 3.93 1.60 9 2.64 2.13 1.44 13.68 7.21 10 11.88 0.80 1.68 1.89 11 1.39 1.21 12.66 6.07 12 12.48
contained significant amounts of iron on the air side surfaces (12-20 at %); however, only the glass side surfaces of the 0.33 mol % Fe films contained similar amounts or iron. This is not surprising since all these films contained large iron-rich surface layers as determined by TEM (Table IV). The 0.20 mol % Fe glass side surfaces contained less than 3 at. % iron and TEM revealed iron rich surface layers less then 70 A thick. This is rationalized by the fact that these surface layers appeared loose and semicontinuous. The air-side surfaces of the modified f i b contained less then 2 at. % chlorine. This is not surprising since small amounts of FeC13 in the surface layer may have escaped reaction with the oxygen-rich atmosphere. On the glassside surface up to 6.4 at. % chlorine was observed on the surface. Films 7 and 11 contained less than 1.2 at. % chlorine; however, these films also contained very small amounts of iron. A high concentration of chlorine on the glass-side surface was attributed to the lack of oxygen to transform the [FeC1,I3-" (n = 3 or 6) to iron oxide. Furthermore during the curing process the water or methanol or imidization exited the film via the air-side surface thus
leaving the glass-side void of oxygen-containingreactants to transform the [FeC1,I3-" to iron oxide. Atomic concentrations of the iron oxide produced on the surfaces of the fiis were determined by first subtracting out any iron in the form of FeC13 from the total amount of iron. Then the ratio of oxide oxygen to iron was determined and compared to various forms of iron oxide such as Fe2O3, Fe304,FeOOH, and FeO. The ratio of oxide oxygen to iron was found to be between 1.21 and 1.73, which corresponded to either Fe2O3 (FeOlS5)or Fe304 (FeOl,=). The possibility of reducing Fexnto Fen to form Fe304was discounted since there were no reducing agents present in the film's medium. Therefore, it is believed that Fe203 is produced on the surfaces of the films. The iron in the form of FeC13 could not be subtracted from a curve fitof the iron 2p photopeak since this peak was very broad. We also observed that Fe2O3 and Fe304gave similar iron 2p spectra in terms of line width and position. It was observed that the H3FeC&-modifiedf i i did not exhibit ferromagnetic behavior (i.e., remnant magnetization or coercivety, indicative of y-Fe2O3 or Fe304)like the Fe(a~ac)~-modified films; however, the films did show paramagnetic behavior. That is when the magnetic field was applied to the films an induction magnetization was observed; however, when the field was removed the induction magnetization went to zero. This observation suggested that the majority of the iron in the film is in a paramagnetic form such as FeC13. It was estimated that less then 5 % of the iron was converted to iron oxide. Because there was such a small concentration of Fe203,the magnetic measurements could not distinguish the CY from the y phase of Fe203. The magnetic observations were consistent with the results obtained from chemical analysis, indicating that the majority of the iron in the film is in the form of FeC13. In other words, the imidized H3FeC&-modified f i b were found to contain both air- and glass-side surface layers of Fe2O3 while retaining bulk particles of FeC13 at a considerably greater concentration. Metal-Polymer Interactions. In previous studies by Sen et al.l3sZ9 the presence of coordination sites on the polyimide precursor backbone was stated to influence the migration of dopant material to form large bulk particles and iron-rich surface layers. We find a similar trend. However, we also find that the thermal stability of the dopant has a greater influence on the architecture of the composite films. When F e ( a ~ a c was ) ~ used as the dopant, very small particles of y-Fe2O3 (20-80 nm) were produced in the bulk of the film regardless of the polyimide precursor used. This observation showed that removal of binding sites (i.e., amide acid to amide ester) had not effect on the size of the bulk particles or the formation of iron rich surface layers. Because of its low thermal stability F e ( a c a ~was )~ believed to decompose to iron oxide before migration occurred. In fact significant amounts of iron oxide were polyimide observed by X P S from curing a Fe(aca~)~-doped film to only 100 O C . For these reasons we feel that the migration of the F e ( a ~ a c is ) ~not influenced by metal/ polymer interactions. In Sen et al.'s study the metal dopants used were thermally unstable and very reactive (Fe3(CO)12and Cr(CJQ2); therefore, it was though that these compounds decomposed before migration and metal/polymer interactions existed. When the polyimide precursor films were doped with H3FeC&very large surface layers of iron oxide were produced. The surface layers were on both the air and glass (29) Nandi, 1991,3, 201.
M,Conklin,J. A.; Saluati, L. Jr.; Sen, A. Chem Mater.
Chem. Mater. 1992,4, 737-743
sides and in films produced from both the poly(amide ester) and the poly(amide acid). Again we attribute the formation of these layers to the greater thermal stability of H,FeCb. Although we found that FeC13 was formed in the bulk of the film, we feel this can partition into the imidizing film during the thermal process (FeC13mp = 306 OC) and migrate to the air- and glass-side surfaces. In the case of the poly(amide acid) the FeC13 may initiate its migration after imidization has occurred, while in the case of the poly(amide ester) no acid coordination sites exist; therefore, migration most likely occurred throughout the curing process. These postulates are consistent with the surface layers observed via TEM. The poly(amide ester)-derived modified films contained larger surface layers as compared to the poly(amide acid)-derived films. In summary, we find that both the thermal stability of the dopant and the presence of coordination sites on the
737
polyimide precursor affect the formation of large bulk particles and surface layers. It was observed that the thermal stability of the dopant material made a large contribution to the migration process. However, the percent conversion to a magnetic iron oxide of the metal dopant had to be sacrificed in order to obtain the enhanced migration (5% versus 95% conversion for the H3FeCI, and F e ( a ~ a cmodified )~ films respectively). Acknowledgment. We thank Allco Chemical Co. for providing the BTDA and PMDA. We also thank Harshaw Chemical Co. for providing the F e ( a ~ a c ) We ~ . gratefully acknowledge financial assistance from the National Aeronautics and Space Administration at Langley Research Center. We also thank Dr. James Rancourt for helpful discussions and Mr. James Hollenhead for his assistance in obtaining TEMs.
Structural Investigations of New Calcium-Rare Earth (R) Oxyborates with the Composition Ca4RO(B03)3 R. Norrestam* and M. Nygren Departments of Inorganic and Structural Chemistry, Arrhenius Laboratory, Stockholm University, S-10691 Stockholm, Sweden
J.-0. Bovin National Center for HREM, Chemical Center, University of Lund, P.O. Box 124, S-22100 Lund, Sweden Received March 5, 1991. Revised Manuscript Received February 13,1992 A new series of calcium containing rare earth (R) borates has been synthesized by high-temperature synthesis for some of the trivalent R ions, viz., La3+,Nd3+,Sm3+,Gd3+,Er3+,and Y3+. Characterization by X-ray diffraction and electron microscopy techniques has shown them to be isostructural with the compositions CQRO(BO~)~ The structure type of these new oxyborates, determined by single-crystal X-ray diffraction methods applied to the Sm compound, is related to the structure of a recently found calcium fluoroborate and also to common fluorapatite. For Ca4SmO(BOJ3the unit cell parameters are a = 8.114 (2), b = 16.061 (4), c = 3.579 (1) A and p = 101.38 (3)'. The space group is monoclinic noncentrosymmetric Cm with 2 = 2. The structural model deduced was refined versus the amplitudes of the 547 most significant X-ray reflections, with sin (6')lX I0.65 A-1, to an R value of 0.016. The occurrence of Ca, Sm, 0, and B in the synthesized crystals were verified by parallel recording of electron energy-loss spectra (PEELS) in a high-resolution transmission electron microscope (HRTEM). The composition of the crystal used for the X-ray structure investigation was determined by energy-dispersive X-ray analysis (EDX) analysis in a scanning electron microscope to be Ca3,~ml,loO(B03)3. Studies on analogues of the present compounds toward their possible applications as a high-Nd concentration minilaser materials are in progress.
Introduction For smaller trivalent ions, like those in the first row, a series of borates known as the pinakiolite family are frequently formed with divalent ions. The composition of is (M2+)2M3+0W3and the structure type is this formed for a large variety of divalent and trivalent ions (cf* and In attempts to synthesize pinakiolite-related phases containing Ca2+and trivalent rare earth ions (including Y3+), it soon became apparent that some unknown phases occurred. X-ray powder diagrams indicated that the phases obtained were mostly isostructural. Preliminary studies of their enerm-dispersive X-ray spectra coliected in a scanning ek-ctron microscope suggested a composition of Ca4RO(B03)3,in(1)Norrestam, R.;Bovin, J.-0. 2.Kristallogr. 1987, 181, 135.
0897-4756/92/2804-0737$03.00/0
dicating that a new oxyborate type had been found. To characterize these new phases, crystalline specimens were prepared, and the present structural study of the Sm compound was out. The number of well-characterized solid alkaline earthrare earth borates is so far very and includes the SP-containing phase Sr3R-JB03)4 (see, e.g., Abdullaev and Mamedov2) and the Ba2+-containing with similar composition, Ba3R2(B03)4, but with a different structure. Recently another family of oxyborates, with the composition A&lM'(BO3), where A = Sr or Ba, has been f0und.49~ The metal ions and M, comprise several metal ions, (2) Abdullaev, G. K.; Mamedov, K. S.Kristallografiya 1982,27,795. (3)Yan. J. F.:Hone. H. Y.-P. Mater. Res. Bull. 1987.22. 1347. (4)Thompson, P. 6:;Keszler, D.A. Chem. Mater. 1989,'1, 292.
( 5 ) Schaffers, K. I.; Alekel, 111, T.; Thompson, P. D.; Cox, J. R.; Keszler, D. A. Chem. Mater. 1989,1, 7068.
0 1992 American Chemical Society