M. T. Vaughan1 D. J. Weidner1, Y. B. Wang1, J. H. Chen1, C. C. Koleda1 and I. C. Getting2
Center for High Pressure Research and 1State University of New York at Stony Brook or 2University of Colorado, USA
We have designed and tested a new apparatus for in-situ x-ray diffraction studies under high pressures and temperatures. The apparatus is a two-stage multi-anvil system similar to the now-common 6-8 systems; the first stage is a steel cylinder split into six parts enclosing a cubic cavity (19.5 mm edge length with the [111] axis of the cube vertical) which contains the second stage anvil assembly. The second stage is assembled outside the press and consists of eight 10 mm edge length WC or polycrystalline diamond cubes, separated by spacers. Each cube has one corner truncated into a triangular face; the eight truncations form an octahedral cavity in which the pressure medium is compressed.
Click here for enlarged image (36k):
The cell assembly is an octahedron made of semi-sintered MgO or
boron-epoxy. The incident x-ray beam passes through the gaps between the WC
cubes in the [110] direction of the eight-cube assembly and diffracts parallel to one of
the (100) planes, through gaps between the WC anvils, with a diffraction vector
35.3° from the vertical plane. A special holder is
built for a solid-state detector for energy dispersive measurements at several
2-theta angles.
T-Cup Cell (252k) 
..... T-Cup in SAM-85 press (273k) 
T-Cup with alignment cube (131k) 
The tangential strains in the containment ring were monitored with strain gauges using a solid WC cube in the sample chamber. The gauges were mounted on the inside radius (ID) and on the outside radius (OD) of the ring at positions in the center of the wedges and at the boundaries between two wedges.
The results for loading to 100 tons
are illustrated in the figures below.
ID plot: 15k.
OD plot: 19k.
The ID strains at the boundary between the wedges were a maximum of all measured strains, while the OD strains there were the minimum, indicating a bending moment in the ring at this position.
Considerable hysteresis was observed between loading and unloading. The subsequent pressure measurements do not produce nearly the same amount of hysteresis, indicating that the source of the hysteresis is the friction at the base of the wedges and not in the cell itself. Loading to 210 tons did not produce any unrecovered or non-linear strain, indicating that plastic flow of the confining ring does not occur to these loads.
24 GPa can be routinely achieved in 6-8 high pressure systems such as the USSA-2000 at Stony Brook, using tungsten carbide second stage anvils. Typically, at 800 tons of loading force on a system with 2 mm truncations of the WIC anvils these pressure are obtained. The challenge of the T-Cup system is to achieve comparable pressures with a limit of 200 tons of loading force. The system should be optimized to provide maximum pressurizing efficiency while minimizing anvil failure. We have conducted several test of pressure generating capacity for different system designs.
Illustrated in the figure below are three different anvil designs that we have analyzed.
In A, the anvils have no taper; that is they are the standard configuration of truncated cubic anvils.
B illustrates anvils with positive tapers; that is a low angle taper is ground on the anvil surface just in front of the truncation. We have used tapers of four degrees.
C illustrates negative tapers. Here, the entire face is ground to allow the anvil gap to grow wider as we move away from the truncation.
A cell design with disc heaters is illustrated here.
20 GPa cell (57k)
The boron epoxy pressure medium is cut in half, and the heaters (top and
bottom) are inserted, after which the sample and pressure calibrant
are inserted. The thermocouple is a through-type.
In the figure below, we demonstrate pressure
vs ram load for several cell configurations using Kennametal carbide anvils. The pressure
in these systems was determined using synchrotron-generated x-rays at
the NSLS for diffraction in NaCl. The cell parameters of NaCl were then
used as a pressure marker. Thus, pressure is continuously monitored on
both increase and decrease in load.
Plot of pressure efficiency with Kennametal anvils (25k)
Based on this data, we make the following inferences:
· The pressure efficiency is higher with 1-2mm pyrophyllite gaskets near the truncation (triangular symbols) than with no gasket.
· Positive taper are less efficient than no taper (open symbols have positive taper).
· 7/2 (pressure medium edge length/truncation edge length, in mm) (blue symbols) are slightly more efficient that 7/3 and 8/2.
First series of 9 runs, using Kennametal cubic
anvils with 2 or 3mm truncations on one corner.
The 20x20mm fiberglass backing electrically insulated
the carbide cubes from the first stage, and serves as
lubrication also.
Using these results, we repeated and extended the experiments using Toshiba grade F carbide. We came to these additional conclusions:
· Toshiba grade F tungsten carbide yields somewhat higher pressures than Kennametal carbide. The gain is about 10 kb at the highest pressures.
· Negative taper produces the highest pressure; we gained 15 kb in the test run compared to no taper.
· Keeping the gaskets as close as possible to the pressure media increases efficiency.
· Boron-epoxy octahedra are slightly more efficient
than MgO.
Plot of pressure efficiency with Toshiba anvils (25k)
Finally, we made heating experiment using a disk heater using TiC with diamond for a dilutant and a MgO octahedron. In the plot to the right, we compare the heating efficiency of this cell with the 10GPa cell discussed next (also a MgO cell), and a standard DIA boron-epoxy cell.
8 of the 15 runs made using Toshiba grade F anvils. Shown are all the runs using the best cell and gasket found, with different backing and with a reverse (negative) taper, which resulted in the highest efficiency to date!
In our final test, made at the beginning of August, 1997,
we used untapered ADC anvils in a single run.
ADC refers to Advanced Diamond Compact, which is a natural diamond
product hot-pressed with a SiC binder at modest pressure (~2 GPa).
We achieved a pressure equivalent to our best tapered anvil run,
with a maximum pressure of 22.8 GPa with a load of 110 tons.
The untapered ADC pressure-load curve is tangent to the equivalent untapered
WC curve at pressures below 5 GPa, but with much less curvature above that pressure.
At 110 tons, there is a 6 GPa increase;
we expect tapered ADC anvils to exhibit similar pressure increases.
If we taper the anvils, and extroplate along the best -2° tapered trajectory,
we would expect the pressure to reach 30 GPa at 120 tons!
There is much work to be done!
10 GPa cell (63k) 
While maximum pressure is often a goal of cell design, we have also developed a cell that operates to 10 GPa. Many experiments only need these pressures, furnaces can be made of carbon with no problem associated to diamond formation. We have tested an assembly with a 4 mm truncation on the tungsten carbide with very promising results. The pressure vs ram load for this run is illustrated in the attached figure.
The cell illustrated here has the
cylindrical sample chamber oriented in the standard ([111])
direction in the octahedron. An amorphous carbon furnace with the same
dimensions as our standard DIA cell was used. The thermocouple
penetrated one side of the furnace. A 10 mm MgO
octahedron was used with no gasket.
After loading to 110 tons at room temperature, we heated to 1200°C at constant load.
The pressure, as determined by x-ray diffraction, decreased at first (probably due to stress relaxation), then gradually increased to ~112 kbars. A second heating cycle at constant load resulted in almost the same pressure. We make the following observations for this cell design:
· The sample was visible in the x-ray beam for a range of 0.5 mm, indicating that a layered sample in this configuration would support only one layer for sample and one for the pressure marker.
· The cell did not extrude to fill the entire gap, indicating that this system is not limited by the small size of the cubes.
· The power vs temperature curve
shown before (see below) indicates considerable heat loss compared with
our DIA cells indicating that MgO should be replaced with boron-epoxy.
Plot of heater efficiency (11k)
