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

At 2 p.m., six small glass-encased tubes were placed on the worktable in Laboratory 502. The tubes had gold-plated feet and the outer shell was printed with the Institute of Physics' number and the date.

Fang Xudong clipped one of them onto the test socket, connected the bias power supply and signal generator, and slowly pushed the frequency knob up from 100 MHz.

The output waveform on the oscilloscope was relatively clean at 300 MHz, but it started to become distorted at 400 MHz. At 500 MHz, the amplitude dropped by half. When the frequency was pushed over 600 MHz, the sine wave on the screen had collapsed into an almost straight line.

"It's over." Fang Xudong turned off the signal generator, took the tube off the test socket, and put it back in the box. "All six are the same. The current amplification factor is zero above 500 MHz, and the cutoff frequency is stuck here. It can't go any higher."

Jiang Ming stood across the workbench, holding the technical data sheet for the tube in his hand, his gaze fixed on the base width column: eight micrometers.

Eight micrometers is the limit that the Institute of Physics can achieve with manual scribing and photolithography; any smaller and the equipment will not be precise enough.

To ensure that the triode has sufficient gain for use as a local oscillator above one gigahertz, the base width must be reduced to less than one micrometer, ideally around 0.5 micrometers.

"Photolithography can't be done." Fang Xudong snatched the data sheet from Jiang Ming's hand, spread it on the table, and poked at the line of numbers with his finger. "The resolution limit of the lithography machine at the Institute of Physics is five micrometers. This is already the best equipment in the country. Even if you wear out the lens, you can't jump to 0.5 micrometers in one step."

Jiang Ming didn't respond to that. He turned around, walked to the front of the blackboard, picked up the chalk, and began to draw.

The first stroke is a rectangle, labeled as a cross-section of a silicon wafer.

The second stroke is a light-colored thin layer drawn on the upper part of the rectangle, labeled as the phosphorus diffusion N-type layer.

The third stroke is an even lighter and thinner line drawn below the thin layer, labeled as the boron-diffused P-type layer.

Fang Xudong looked at the diagram on the blackboard, his brows furrowing and relaxing over a period of about ten seconds.

"You're going to use diffusion as the base?"

"The diffusion coefficient of phosphorus in silicon is about three times that of boron." Jiang Ming wrote the numbers between the two layers. "At the same temperature and in the same amount of time, phosphorus diffuses much faster than boron. If I first diffuse boron in to form a P-type base layer, and then diffuse phosphorus in from the same surface to form an N-type emitter layer, phosphorus will catch up with and surpass boron."

Two curves were drawn on the blackboard with chalk. One curve represents the concentration distribution of boron, which slowly decreases from the surface, while the other curve represents the concentration distribution of phosphorus, which drops sharply from the surface, but its initial concentration is much higher than that of boron.

"The intersection of the two curves is the interface between the base and the emitter." Jiang Ming drew a circle on the intersection with chalk. "And the base width is the distance between the boron diffusion front and the phosphorus diffusion front."

He wrote a number next to the circle: 0.5 micrometers.

"This distance is not etched by photolithography, but is formed naturally by the difference in the speed at which the two types of impurity atoms move through the lattice."

Fang Xudong stared at the two curves on the blackboard for a long time, his lips moving several times before finally uttering a technical question rather than a challenge: "How do we control the diffusion temperature and time? Boron first, then phosphorus, should we cool it in between? Is the concentration gradient at the interface between the two diffusion processes steep enough?"

"Boron diffusion at 1100 degrees Celsius for four hours under nitrogen protection," Jiang Ming said, writing the parameters on the blackboard. "After cooling to room temperature, flip it over to check and confirm that the boron junction depth is about two micrometers. Then phosphorus diffusion at 1050 degrees Celsius for two hours. The initial surface concentration of phosphorus is two orders of magnitude higher than that of boron. After diffusion, the phosphorus junction depth is about 1.5 micrometers. The difference between the phosphorus junction depth and the boron junction depth is the base width."

Old Sun had been standing at the door for a while, carrying a canvas tool bag. When he heard the number 1100, he put the bag on the stool by the door and came in.

"Can your quartz tube diffusion furnace reach this temperature?"

"It's possible." Jiang Ming turned to face Lao Sun, "but the temperature fluctuation must be controlled within plus or minus two degrees, otherwise the diffusion depth deviation will exceed the allowable range."

Old Sun nodded: "I'll recalibrate the thermocouple. The temperature control relay has bad contact; I replaced it with a new set of contacts yesterday. I'll run the empty furnace temperature curve again this afternoon."

Da Liu followed Lao Sun in, carrying an enamel basin containing three brown glass bottles and a stack of filter paper.

"The hydrofluoric acid is here. The concentrations are prepared as you requested: one bottle each of 5%, 10%, and 20%."

Jiang Ming glanced at the three bottles: "First, use 5% of the solution to completely etch away the oxide layer on the surface of the silicon wafer. There must be no residue before diffusion."

Preparations continued from 2 p.m. to 6 p.m. Old Sun ran the temperature curve of the diffusion furnace three times to confirm that the linearity of the heating section was up to standard and the fluctuation of the constant temperature section was within ±1.5 degrees.

Da Liu prepared all the etching solutions and weighed out the boron and phosphorus sources and loaded them into the quartz boat.

At 7 p.m., the first batch of silicon wafers were loaded into quartz tubes.

The furnace temperature climbed at a rate of 150 degrees per hour. High-purity nitrogen gas was passed through both ends of the quartz tube, and the gas discharged from the tube mouth carried a slight heat wave, forming an invisible disturbance above the iron stand.

Four hours later, the boron diffusion was complete, and the furnace temperature began to drop.

Jiang Ming stayed next to the temperature recorder, copying a set of data into his notebook every fifteen minutes, recording the data from 1100 degrees Celsius down to below 300 degrees Celsius before leaving.

At 2 a.m., the silicon wafer was cooled to room temperature and taken out. The surface was a uniform dark blue, which was the color of the borosilicate glass film.

Da Liu used hydrofluoric acid to etch away the glass film, revealing a clean silicon surface.

The next morning, phosphorus diffusion began at 1050 degrees Celsius. Two hours later, the phosphorus source was replaced in the quartz tube.

At 4 PM, the second diffusion was completed, the glass was cooled, the sample was removed, and the surface of the phosphosilicate glass was etched.

Fang Xudong clipped one of the silicon wafers onto the stage of the polarizing microscope, adjusted the focus, and brought his eye close to the eyepiece.

Under polarized light, the cross-section of the silicon wafer clearly shows a three-layer structure: the top layer is an N-type emitter formed by phosphorus diffusion, the middle layer is a very narrow bright line, which is the P-type base layer formed by boron diffusion, and the bottom layer is the original N-type silicon substrate as the collector.

Fang Xudong adjusted the fine-tuning knob three times, aligned the eyepiece scale with the upper and lower boundaries of the bright line, took the readings, and made calculations.

0.48 micrometers.

He straightened up from in front of the goggles, looked at Jiang Ming with a complicated expression, opened his mouth and closed it again, and finally said only one sentence: "Where did you learn all this stuff?"

Jiang Ming did not respond to the question. He reached out and took the silicon wafer from Fang Xudong, examined the uniformity of its surface against the light from the window, and then put it back in the desiccator.

With a base width of 0.48 micrometers, the cutoff frequency can be pushed to over 2 gigahertz, which is sufficient for making a local oscillator.

Old Sun stood beside the diffusion furnace, carefully emptying and storing the residual phosphorus source from the quartz tube. Hearing the number Fang Xudong reported, he turned and asked, "Is it done?"

"It's done."

Old Sun put the quartz boat back on the shelf, wiped his hands on his apron, and didn't say much, but the corners of his mouth turned up slightly.

Fang Xudong has already begun cutting the first batch of double-diffused silicon wafers into single-tube chips, preparing for pin welding and packaging.

At 8 p.m., the three packaged double-diffused silicon transistors were inserted into the socket of the parameter tester. The bias power supply was turned on, and the pointer of the collector ammeter slowly deflected under the extremely high bias voltage, moving to the right from zero.

Fang Xudong pushed the frequency of the signal generator to one gigahertz, and a clear sine wave appeared on the oscilloscope at the output end, with no amplitude attenuation.

He continued pushing it up to 1.5 gigahertz, and the waveform was still there.

At two gigahertz, the waveform begins to distort slightly, but the gain is still greater than one.

Fang Xudong's hand stopped on the knob, and he glanced back at Jiang Ming.

Jiang Ming stared at the still-pulsating green curve on the oscilloscope screen, his pencil unconsciously twirling twice along the edge of the manuscript paper.

With the local oscillator problem solved, the next step is the mixer diode.