1962 IRE TRANSACTION8 ON COMPONENT PARTS 53
Solid Aluminum Electrolytic Capacitors with Etched Aluminum Foil*
Summary-Some suggestions are made for further developments in the field of electrolytic capacitors using etched aluminum foils and solid electrolytes.
Various laboratory-tested construction procedures are discussed in detail. The behavior of capacity, dissipation factor, and leakage is illustrated as the function of temperature and frequency during storage with and without current. It is shown that the production of aluminum electrolytic capacitors with commercially available etched foils and solid electrolytes will be entirely feasible in the near future.
In conclusion, the optimal data found so far in a series of experi- ments are reported, pointing the way toward results that may reasonably be expected from further investigations.
VER SINCE the introduction of solid tantalum electrolytic capacitors, attempts have been made to replace the relatively expensive tantalum with
other materials. Several patent applications indicate that aluminum, in the form of wire or sintered slugs, has been investigated in this connection. Attempts with aluminum foil are scarcely known, but a British patent, No. 845 698 (patent holder: U. S. Secretary of the Army), indicates that some preliminary work has been done in this line. In the meantime etched aluminum foils of high purity and surface gain have become commercially available, and forming methods resulting in highly stable aluminum oxide layers are now known, so that the successful con- struction of solid aluminum electrolytic capacitors seemed feasible. A series of experiments were therefore carried on in the laboratories of the Aluminium-Walzwerke Singen, to determine the usefulness of etched aluminum foils in solid aluminum capacitors. Preliminary results indicate that such capacitors can definitely be made.
II. ALUMINUM FOILS
All foil samples used in our investigation were taken from commercially available types containing at least 99.99 per cent pure aluminum. In order to cover the entire range from 3 to 600 volts, we largely confined ourselves to the Universal foil type 2013 made by Aluminium-Walzwerke Singen. Fig. 1 shows the capacity yield of this foil type, in microfarads per square inch, in the conventional, wet electrolytic (dashed lines) as well
* Received November 16, 1961; revised manuscripts received November 27, 1961, and December 7, 1961. Presented at the Thir- tieth Annual Meetmg of the Conference on Electrical Insulation, Pocono Manor, Pa., October 23-25, 1961.
t Aluminium-Walzwerke Singen G.m.b.H., Singen-Hohentwiel, West Germany (a subsidiary of Aiuminium-Industrie A. G., Zurich, Switzerland).
- SolId capacitors
- - -Wet capacitors
4 Capacaty with foil type 2013 0.7
0.2 a 2
0.1 E 0.07 'g 0.04 2 0.02 z 5 0.01 .o 0.007 B 0.004 z
I I III I I I II I I I I
2 34 6810 20 30 50 s0100 200 400600 1000 Foil-forming potentlo (volts DC.)
Fig. l-Capacity and anodic area of solid and wet aluminum capacitors.
TABLE I CAPACITIES OF SOLID AND WETLOW-VOLTAGE CAPACITORS
Voltage of the capacitor
Capacities in the capacitor
Si;ogp Foil-forming %VJrki;g Surge Wet Solid potential voltage capacitor capacitor
type (volts dc) (volts dc) Wsq in)
2013 6 2021 6 I 2
80 140 190 230
as in solid capacitors (solid lines). To test a foil of higher gain, a number of experiments were made at low voltage, using type 2021 foil made by the same producers. Com- parative values, listed in Table I, show that the use of etched foils is parhicularly favorable in low-voltage solid capacitors, due perhaps to the higher series capacitance of their cathode coating.
III. FOIL FORMATION
Foil formation has an even greater influence on the quality of solid than of wet capacitors. The most stable oxide layers known up to now result from alternating the formations in aqueous solutions of boric acid with immersions in boiling deionized water. In this process the water appears to react with the basic metal wherever a fault in the oxide layer exists, according to 2Al + 4H,O *Al,Os. H,O + 3H,, or perhaps 2Al + 6H,O * Al,O, .3H,O + 3H,, causing the resulting relative large Al,O, .H,O or Al,O, .3H,O molecules to tear and lift the oxide layer around faulty or contaminated sites and to remove forcibly any impurities. Each subsequent forming cycle then closes any resulting gaps in the oxide
54 IRE TRANSACTIONS ON COMPONENT PARTS June
layer. Since these cycles are repeated four or five times, each bringing about an improvement of t#he foils, it seems evident that falJ1.s ancl impurit,ies are successively removed.
The importsnet: of using extremely pure water cannot be overemphasize 1 because t,he slightest contaminant in the water might 1: e introduced underneath the protective forming layer. In addition, t,hc foil must be thoroughly rinsed between eac:h forming and boiling operation because the presence of eve11 0.1 per cent boric acid in the boiling bath would greatly impair the effectiveness of the whole method. Most of our experiments were made at a foil- forming potential If 110 volts dc for electrolytic capacitlors of 35-40 volts dc.
Our foil was formed in the following steps:
ten minutes of formation in queous n~mmonium pentaborate solution at room temperature, with a resistance of 3.50 ohm-cm; thorough rirsing of foil in very pure water at room temperature; three minutes of boiling in very pure water (re- sistivity, ml:asured within the deionizer at room temperature prior to exposure to the atmosphere > lo7 ohms 1; five minutes of formation as in step 1) ; three-fold repetition of steps 2) to 4).
In tests at < 110 volts dc the boiling time was reduced to one minute. Abolre 110 volts we used an electrolyte con- taining 14 per cer.t boric acid at 85 C.
11. COiWTKUCTIOR OF CAP.4CITORS
To prepare capacitors we used foil strips of l-cm width and 0.003- to 0.0035-inch (Sci- to 90-p) thickness. A length of 1 cm was indic:ated by means of V-cuts on both sides of the strip. Aftc:r fo.rming, this foil area of 1 cm2 was immersed in liqllid manganese nitrate, in some cases under vacuum. he presence of a highly stable oxide layer is essential because in the subsequent pyrolytic decomposition oj manganese nitrate into manganese dioxide very hil;h temperatures are employed which could otherwise damage the oxide film. Also, faulty oxide-layer spots arc much more difficult to correct in solid-electrolyte t:nan :in licluid-electrolyte capacitors. For this reason the m:,nganese nitrate coating must be applied only on foil surf:.ccs that carry a firm, closed, and un- damaged oxide fil:n.
The manganese nitrate was decomposed into manganese dioxide at varioulj temperatures. Some of the capacitors were formed once more in ammonium pentaborate at 3.50 ohm-cm and roo:m temperature, after pyrolysis. A silver counterelec trode was applied to the manganese dioxide coating by pa,inting or immersion, and the lead wire attached to it.
Fig. 2 shows the, capacitors after the three basic manipu- lations: a) after foil formation, b) after immersion in manganese nitratl: and pyrolysis into manganese dioxide,
,.., . ._ j a) &i-foil etched Iaf Cbated wifh
I cf Coated witk !
, and formed MnCi2 silvw 1
Fig. 2-Manufacturing stages of solid aluminum capacitors.
Fig. 3-High-cnpaci t,y solid aluminum capacitor.
and c) after application of the silver coating and attach- ment of the negative lea,d wire. The untreated foil acts as t,he positive lead.
Following these steps, the capacitors werle connected through a resistor to a dc voltage and slowly formed until a peak potential was obtained and there maintained until the leakage current reached a minimum. After 24 hours of storage without current at room temperature, the e: ectrical parameters were measured. Capacity and disi;ipation factor were determined in a bridge circuit without de polarization at 1 volt ac and 50 cps. The leakage current was measured at room te:mperature after one minute with rated voltage applied. Aging tests with a,nd without voltage were then made.
To avoid damaging the oxide layer, it is advisable to use not wound but stacked capacitors for high-capacity units. Fig. 3 shows front and side views of such a stacked capacitor in various stages. At the left there is 9 single element [cf. Fig. 2(c)] with both negative a,nd :?ositive
1962 Post: Solid Aluminum Electrolytic Capacitors
leads attached. In a 3-4 volt dc capacitor, for example, this element has a capacity of 35 pf. In the center there is a combination of fourteen individual elements con- nected by means of conductive silver into a solid block with a volume of 1 cm3 and a capacity yield of about 500 pf at 34 volts dc. The stacked capacitor shown in the right-hand side is lacquer-coated for protection against moisture and damage.
Unless otherwise specified, all foils were formed at Time (minutes)
110 volts de for a capacitor voltage of 3540 volts dc. Fig. 4-pyrolysis duration of solid aluminum capacitors vs tem- perature.
A. In,terdepen,dence of Temperature and Time in Pyrolysis
After forming and after a single immersion in manganese capacity Dissipation factor
nitrate at room temperature, the foils were subjected to pyrolysis in a drying oven at various temperatures and times until the weight became constant. il safety factor of 25 per cent was added to the Gme elapsed and the dat#a plotted (Fig. 4). To determine the optimal pyrolysis
temperature for capacitors, we constructed a series of capacitors of 35-40 volts and subjected them t,o one single pyrolysis at various temperatures for time periods as Leakage ot 25vDC Test formation after shown in Fig. 4. A few samples mere then formed without x silver coating in an aqueous electrolyte to determine, by noting the time required for formation, any destruction of the forming layer. Fig. 5 shows the various electrical
yk ,,,, z p+ ,,Ys, , ~
data as well as forming times. The pot,ential reached 9 0 200 400 600 0 200 400 600
never exceeded 2.5 volts de. For all pyrolysis temperatures appreciab1.y below 4OOC, the readings become less
Fig. 5-Data of solid electrolytic capacitors at various pyrolysis temperatures.
favorable, -while above 400C there w& no noticeable change except for the dissipation factor. At, 55OC the majority of the capacitors broke down. All further experi- ments were therefore made with a pyrolysis temperature of 400C.
Up to this point only slightly heated manganese nitrate was used; in the following experiments, however, 12 IO the foil was immersed in a manganese nitrate melt at dissociation temperature. This resulted in much better dissipation factors, but other readings, notably the Leakage at 25vDC leakage current, suffered. Unfortunately this was dis- g 3. covered only after some further tests were already under way, so that it entered into the next test series.
Fig. 6 shows data and forming time of a test series with a pyrolysis temperature of 4OOC. The capacitors again reached a maximum potential of only 25 volts dc, and pyrolysis times ranged from 2 to 1000 minutes. Since 10 minutes appeared to be the most favorable time, all further tests were made for 10 minutes at 400C.
Time (minutes )
Fig. 6-D&a of solid aluminum capacitors with varying pyrolysis duration at 400C. Foils immersed into Mn(NOs)s melt at dis- sociation temperature.
B. Capacitor Characteristics for Repeated Mn(nO,), Coating and Pyrolysis
In the tests following, the effects of one- to five-fold manganese nitrate coating alternating with pyrolyses were studied. Between and after the individual pyrolysis no foil formation in aqueous solutions was carried out; only the finished capacitor was formed.
4 Dissipation factor
Test formation after
Fig. 7 shows values for the capacitors and the test formation. Again, only about 25 volts dc was reached regardless of the number of pyrolyses. Test formation times show that repeated pyrolyses cause some fairly heavy damage to the oxide layer, which, however, was not found to have any effect on the capacitor data.
In this experiment capacitors made from foil formed in
56 IRE TRANSACTIONS ON COAlPONENT PARTS June
an aqueous an tmonium pentaborate solution after pyrolysis did re::,ch the required 35-40 volt potential. The following exI)eriments therefore deal with formations after each pyroly;Gs.
C. Capacitor Characteristics for Formation After and Between Pyrolysis Cycles .
Capacitor data and test formation readings are shown in Fig. 8, for which capacitors were used that had been formed in aqueous ammonium pentaborate solution between each cc jating-and-pyrolysis cycle (solid line). In some cases all additional formation took place after the final pyrolysis (dashed lines). Capacitors with a single pyrolysis ::eached only 25 volts, those with five pyrolyses, 35 volts, while all others reached the desired 40 volts dc. Starling with this experiment, the maximum forming voltage that could be reached without destroying manganese dioxid.e coating was recorded.
It will be seen that the most efficient capacitors were obtained after tvro tat three coating-and-pyrolysis cycles; where the last stl:p should always be a pyrolysis so as to avoid excessive dissipation factors.
D. Immersion I:npregnation in Alanganese Nitrate in Vacua
Since our best results were obtained by immersion in slightly heated n~anganese nitrate, we now immersed the foils under vaculun. YEvacuation was continued until air bubbles ceased to appear in the nitrate. Nearly all other steps were exactly as previously outlined, except that the first mangacese nitrate impregnation and pyrolysis were doubled. In Fig. I) the solid lines again show values for electroIytic capa,cit.ors not formed in ammonium pentaborate after, the final pyrolysis, whereas the dashed lines indicate da;a where formation constituted the last process before silT,er coating. Here all capacitors reached 40 volts dc. The moc,t favorable leakage results were obta,ined after three pyrol,rses and one or two formations. Earlier results, however, tend to indicate that similar readings, but with better tlissipation factors, can be obt,ained after two pyrolyses separated by a forming cycle. .
E. Frequency Behavior of Solid Aluminum Ca,pacitors with Etched Foil
Fig. 10 shows t,he frequency dependence of capacity and dissipation f:tctor of some capacitors made with three pyrolyses and Tao formations. The values shown are comparable with those of solid tantalum capacitors.
F. Capacitor Characteristics During Storage without Voltage
In Fig. 3 1 the behiavior of capacitors during idling for 500 hours at roonl t.emperature and during 14 temperature cycles, each of 20 hours at 65C, resp. 85C, and 4 hours at room temperature are plotted. Measurements were made at room temperature, and fluctuations were found to remain essentiall,y within allowable tolerances.
Leakage at 25 v DC 0 : 30 0 > 20
Li 4 IO a a 0 L *-
IX 2x 3x 4x 5X
Test formation a Ye last pyrolysis
-II IX 2x 3* 4:: 5x
Fig. 7-Data of solid aluminum capacitors d.ependix.g on numbers of coatings wil;h Mn(NOz)s and pyrolyses.
- Formed between each two pyrolyses but not after Ias. pyrolysis - - - Formed after last pyrolysis also
Leakage at 35 vDC Test formation aft ?r lost
IX 2x 3x 4x 5x IX 2x 3x 4x 5x MnOe Mm&
Fig. S-Data of solid aluminum capacitors formed iafter pyrolysis.
-Formed between Znd,3rd,and 4th but not after last pyrolysis - - - Formed after last pyrolysis also
Dissipation foctol, 2.5
0 2x 3x 4x
Leakoge at 35 v DC
x 30 c 0 T 20
IL .$ IO
a // 40 -
2x 3x 4x
x c 9 t
Test format ion of, er pyrolysis
01-U-+ 2x 3x 4x
Fig. g-Data of solid aluminum capacitors coated with in vacuum.
196% Post: Solid Alurrkuun Electrolytic Capacitors 57
G. Capacitor Characteristics During A.ging with Voltage
Fig. 12 shows aging characteristics of capacitors under voltage for 1000 hours at room temperature, 65C, and 85C. Initial measurements were made at room tempera- ture, all subsequent ones at the test temperatures. This work is being continued and will be reported in a later publication. Here it is particularly the leakage current which shows a marked increase at the first measurement, other values remaining essentially constant.
--- Room temperature - 85T
IO' IO' 103 104 105
Fig. IO-Capacity and dissipation factor of solid aluminum capaci- tors at different frequencies and temperatures.
,, 1.0 .?z g 0.8 8 0.6
Fig. 11-Data of solid aluminum capacitors during idling under different temperatures.
Room 65T 85 T
0 500 1000 0
Time ( hours )
Fig. I2-Data of solid aluminum capacitors during aging under different temperatures.
Whereas the tests here reported concerned capacitors of 3540 dc rating and a foil-formation potential of 110 volts dc, several solid electrolytic capacitors were made in some very recent experiments with a foil-formation potential of only 50 volts. They attained 45 volts dc, their capacity per 1 cm2 foil area was 2.5 pf (corresponding to a yield of 16 pf/sq in), their dissipation factor about 8 per cent, and their leakage current for 25 vo1t.s dc < 0.02 pa, for 35 volts de < 0.2 pa per pf.volt de.
Other capacitors, whose foils were formed at very high voltages (540 volts) and which were not further formed either as foil or as capacitor, after a single pyrolysis reached 150 volts dc before breakdown.
Several capacitors were made for 34 volts de, with foil formed at 6 volts. The yield of these capacitors, each about 1 X 1 X 0.75 cm, amounted to 35 pf/cm3. These results are definitely of interest in comparison with tantalum capacitors. Such a capacitor was found to operate perfectly in a transistorized radio.
Summarizing these preliminary studies, the most favorable conditions found for solid aluminum electrolytic capacitors with etched foils were:
Foil formation must be carried out under the best possible conditions. After foil formation, all damage to the oxide layer must be avoided. Only fully formed foil areas can be allowed to come into contact with manganese nitrate. The manganese nitrate is to be brought into the most intimate contact with the formed foil by means of heat, vacuum, dilution, etc. Excessive tempera- tures, which would cause dissociation of the man- ganese nitrate, should be avoided before pyrolysis. The most favorable pyrolysis temperature and duration were found to be 400C and ten minutes, respectively. A two- or three-fold manganese nitrate coating and pyrolysis gave the best values. Slow formation in an aqueous electrolyte between each pyrolysis and the subsequent manganese nitrate coating was advantageous up to the voltages that could be obtained without difficulty. It is advisable to proceed with the capacitor forma- tion slowly, with continuous voltage increase through resistors.
The author wishes to express his appreciation for valuable help extended by his co-workers at Aluminium- Walzwerke Singen, especially Miss H. Wiebe. He also thanks J. Stein for his translation, and F. Rupp, Jr., of Conalco for having delivered the address.