I started with a ship full of max-level veterans, and I'm just a complete noob.

Lin Lanying unpacked the packaging paper sent by the semiconductor research group, placed the primary purified silicon rod on the balance, and recorded the weight, length, diameter and batch number in the new process book of furnace No. 3.

The purity of this batch of silicon material is about three nines, and its axial resistivity before loading into the furnace is close to one ohm-cm, which is sufficient to verify the movement of the molten zone and the segregation of impurities, but it is far from meeting the requirements for device materials.

Old Sun pushed the narrowed quartz boat into the alumina tube, with the quartz serving only as a substrate for the silicon rod, and positioned the boat to avoid the area where the highest temperature in the furnace was concentrated.

After the silicon rod falls into the boat trough, he checks it back and forth along the entire stroke of the push rod to confirm that there is sufficient expansion margin between the joint, the stern of the boat and the inner wall of the furnace tube.

The position of the tungsten-rhenium thermocouple was also rechecked, with the temperature measuring end aligned with the hot zone in the center of the furnace, and the outer protective sleeve was fixed with a soft pad to prevent mechanical vibration during the push-in process.

Da Liu closed the furnace tube port, opened the high-purity argon valve, first used a larger flow rate to replace the air in the tube, and then switched to the stable protective gas flow required for regional purification.

Fang Xudong filled in the record table with the loading dimensions, gas flow rate, clamp resistance, and initial state of the lookup table item by item, and then reported to Jiang Ming that the temperature could be increased.

After the main switch is closed, the silicon carbide rod increases the power along a predetermined slope, the lookup table network provides a reference value based on the thermocouple reading, and the PID output remains within a small correction range.

Between 800 and 1200 degrees Celsius, the difference between the material loading curve and the no-load curve is limited, and the heat absorbed by the silicon rod has not yet changed the main heat transfer state of the furnace.

After exceeding 1300 degrees Celsius, the heat capacity of silicon begins to leave traces on the paper tape. With the same increase in input power, the temperature rise rate is slower than when unloaded.

Jiang Ming temporarily retained the original lookup table output, only recording the deviation between the loaded and unloaded data, to avoid modifying the entire reference before the molten zone was formed.

When the furnace temperature approaches 1,420 degrees Celsius, bright spots gradually appear on the surface of the silicon rod. The bright area expands from a single point and then forms a continuous meniscus in the observation window.

"The melting zone has formed; let's hold the width first."

Jiang Ming compared the observation window markings with the thermocouple readings, and instructed Da Liu to make slight adjustments to the feedforward bias. Only after the length of the molten zone stabilized did he signal to Lao Sun to start the propulsion mechanism.

Old Sun adjusted the propulsion rate to about one millimeter per minute. The leading edge of the silicon rod continued to melt, while the trailing edge re-solidified as it left the hot zone. Impurities migrated to the tail end along with the liquid molten zone.

Compared to germanium, silicon has higher thermal conductivity, faster heat transfer along the front and rear edges of the molten zone, and its specific heat capacity increases the amount of heat stored in the furnace, making the temperature gradient steeper in narrow areas.

Fang Xudong stared at the paper tape reporting each deviation, while Da Liu simultaneously observed the reference power of the lookup table, the PID correction amount, and the silicon carbide rod current.

The initial push remained stable, with the furnace temperature mostly controlled within ±0.3 degrees Celsius, and the width of the melting zone maintained within the predetermined range along the observation window lines.

After the push rod enters the middle section of the silicon rod, the ink line on the paper tape begins to expand, first extending above the set value, and then pressing downward after the power correction arrives.

The second cycle then occurred immediately, with the fluctuation range rapidly expanding to ±2.5 degrees, and the duration of the cycle lengthening from the short oscillation during no-load conditions to several minutes.

"Positive 2.3%, currently declining; negative 1.8%, still going down."

Fang Xudong increased the speed of counting, and marked each point along the peak and valley of the paper strip with a red pencil. When Lao Sun heard Jiang Ming's signal, he immediately stopped pushing the lever.

Jiang Ming put the lookup table output together with the PID correction value and found that the PID was trying to pull back the temperature, but the reference power provided by the lookup table was still running at the no-load calibration value.

After the melting zone moves to the middle of the furnace, the thermal radiation of liquid silicon, axial heat conduction, and the heat carried away by the protective airflow all change the load distribution, and the no-load power meter has deviated from the current demand.

The table lookup device only looks at the temperature and cannot identify which section of the silicon rod the molten zone is located in, nor can it determine how much the furnace heat load has been redistributed after the push position has changed.

Jiang Ming switched the controller to semi-manual mode, retained the PID deviation correction, and then rotated the bias potential of the feedforward compensator to gradually increase the reference power of the current operating point.

The first correction only reduces the downswing, and after the temperature rises, it still exceeds the set value. Based on the power difference of the previous cycle, it reduces a portion of the bias, so that the reference value falls between the two peaks and valleys.

After a narrower oscillation, the paper tape returned to the center line, and the fluctuation gradually returned to the range of ±0.3 degrees. The meniscus of the melting zone also returned to its original width.

"Promote recovery, maintain the rate at one millimeter initially, and then perform a final offset check before entering the tail segment."

Old Sun restarted the propulsion mechanism, and Fang Xudong drew a conspicuous mark next to the middle section, copying the three-minute power oscillation and manual correction range into the anomaly record.

Jiang Ming then added a new function to the kraft paper notebook: in addition to temperature, the reference power for the material also needs to include the correction for the melting zone position.

Changes in the push rod position alter the radiation and heat conduction distribution of liquid silicon in the furnace. If future lookup devices are to operate automatically, the push rod position must also be included in the reference power calculation.

The current hardware doesn't have time to incorporate position feedback, so he can only record the offsets for the front, middle, and rear sections of the silicon rod to create a correction table for the next round.

The first melting zone continued to move towards the tail end, and except for the oscillation in the middle section, the temperature fluctuation remained at around plus or minus 0.3 degrees.

Old Sun guarded the push handwheel, reporting the position to Fang Xudong every time it passed a numbered section. Fang Xudong simultaneously recorded the voltage, current, furnace temperature, and feedforward bias.

The last 20 millimeters at the tail end are reserved as an impurity enrichment zone according to the plan. After the melting zone reaches the boundary, Lao Sun pushes the silicon rod out of the highest temperature zone and then cools it down step by step according to the procedure.

After the furnace body is cooled to a permissible operating temperature by a protective gas flow, the quartz boat is slowly withdrawn from the alumina tube, leaving clear traces of the molten zone movement on the surface of the solidified silicon rod.

Lin Lanying first checked the contact area between the rod and the bottom of the boat to confirm that the silicon rod had not stuck or bent significantly at high temperature before agreeing to proceed with axial measurement.

Fang Xudong divided the measuring points along the length of the silicon rod, and the four probes were placed on the front, middle and tail sections in sequence. The resistivity data were filled into the coordinate paper one by one.

Before loading the furnace, the front section was close to one ohm-cm. After the first purification, it increased to about five ohm-cm. The improvement in the middle section was smaller, while the tail end showed obvious impurity enrichment.

These values ​​are still far from the tens of ohms-centimeters required at the device level, requiring multiple purification processes. However, the trend of impurities migrating towards the tail end of the molten zone has been clearly established.

Lin Lanying remeasured the front end and checked the probe pressure and sample temperature again. The second set of numbers still fell within the same range.

She pressed the resistivity curve of the silicon rod next to the record of the first purification of the germanium material. The two materials had different segregation amplitudes, but the axial change direction was consistent.

Standing on the other side of the recorder's table, Jiang Ming looked at the five ohms-centimeter figure at the front, and the tension that had been building up in his heart for many days finally eased.

The furnace tubes, heating elements, thermocouples, clamps, and controllers were almost all replaced. Even after reaching the silicon melting point, this method, which was migrated from the germanium temperature range, was still able to promote impurity separation.

Lin Lanying then unfolded the two rolls of temperature paper side by side. The temperature fluctuation of germanium during the first purification was ±0.08 degrees Celsius, while the temperature of silicon was stable at ±0.3 degrees Celsius except for the abnormality in the middle section.

The two curves show a striking difference in width, but the changes in their envelopes unfold along similar stages of progression, with each stage—heating, stabilization of the melting zone, movement, and exit from the hot zone—corresponding one-to-one.

She picked up a pencil and wrote four words in the conclusion column of the first test of silicon material.

The method is feasible.

Huang Kun arrived at the furnace room that afternoon, first checking the mid-section oscillation record, then reviewing the molten zone position correction item added by Jiang Ming, and finally turning to the axial resistivity curve.

"If the offset is still adjusted manually in the next round, the record must be tied to the advancing position, and a material list must be formed before the third round."

"First, create a coarse table for the first, middle, and last three segments, and then add position nodes for each iteration. Leave the PID parameters unchanged for now to avoid modifying both variables at the same time."

Huang Kun accepted the arrangement, sat down at the record table, opened the hardcover record book, and moved the "Pending Verification" column for the No. 3 furnace's subsequent furnace operation to the "Formal Plan".

The first batch of silicon material undergoes three consecutive rounds of regional purification. After each round, the axial resistivity is remeasured, and the decision is made on whether to continue with the fourth to sixth rounds based on the impurity enrichment effect.

In the approval section, he wrote that the 502 temperature control solution had passed the initial verification in the silicon temperature range, and that subsequent accuracy optimization and material correction would be the responsibility of Jiang Ming Technology, with the record format following the 502 specification.

When Huang Kun wrote the words "Technical Supervisor" in his pen, he checked the mid-section anomaly handling record again before signing his name at the end.

Jiang Ming took the approval page and saw that 502 and the technical supervisor appeared side by side for the first time in the official document of the Institute of Physics' No. 3 furnace. He slowly smoothed the folds along the edge of the paper with his fingers.

Old Sun had already posted the first pass of push position and offset data on the blackboard, while Big Liu disassembled the lookup table panel, preparing to add new nodes to the material curve.

Beside the recorder's table, the silicon rod with its front resistivity increased to five ohm-centimeters was still placed on the cork support, and the impurity-rich area at its tail end had been circled with a red pencil.

Huang Kun pressed the subsequent furnace approval page onto the silicon rod number book, and then opened the furnace loading column for the second purification.